Youtube comments of AI Search (@theAIsearch).

  1. Quick update for this video: If this discord is full, check out this video to fix it: https://youtu.be/e8CaftH9olk In the download table, I directed you to download the gpucuda version of the software if you have a dedicated graphics card. However, when I went to Hugging Face, I accidentally downloaded the directMLcuda. You should use the gpucuda version (if you have a dedicated graphics card)! Sorry for the confusion!
    803
  2. 😃
    306
  3. Thanks to our sponsor Video to Blog. Convert YouTube videos into high-quality, SEO-optimized blog articles in seconds https://www.videotoblog.ai/?via=ref
    181
  4. someone already did that
    176
  5. i didn't know this rule 🤦‍♂️
    174
  6. yeah there can be considerable delay depending on your hardware
    171
  7. 💀
    168
  8. lol yes
    156
  9. shit... let me check my crypto balance
    148
  10. Thanks to TubeMagic For Sponsoring This Video! TubeMagic helps YouTubers succeed on YouTube with the power of AI. 📈 Try Out TubeMagic For Free 👉 https://tubemagic.com/free-trial?sl=aisearch
    147
  11. oops
    138
  12. If you enjoyed this video, you'll also like the sequel where I go over new features: https://youtu.be/X7DnlE-c2eI More updates by Udio: Hi everyone, We've rolled out a few cool updates to the platform: Track length limit is now 15 minutes. Enjoy generating prog epics and extended trance mixes. Context length for extensions has increased from 30 seconds to 2 minutes. This means Udio can "see" more of your track when generating an extension. In practice it means you can reference and repeat audio from further away in the track. Consistent choruses are now easier to achieve than ever. You can now switch between the flat and the tree view in "My Creations". In the tree view, children tracks (e.g., extensions) are displayed underneath parent tracks. Please keep the feedback coming, and let us know if you have any issues with these features in particular.
    119
  13. 💔
    115
  14. Can you tell this apart from human-made music? Let me know! Note: I had to mute part of the clip at 24:00 due to Bruno's copyright
    105
  15. hello gentlemen
    105
  16. or even 2025
    104
  17. lol good analogy
    101
  18. RESOURCES & UPDATES: Thanks to our sponsor: TurboType is a Chrome extension that helps you type faster with keyboard shortcuts. It doesn't hurt to try it - they have a FREE forever plan. https://www.turbotype.app/ Remember to replace your local link's 0.0.0.0 with 127.0.0.1 If local install doesn't work, try this HF space: https://huggingface.co/spaces/MohamedRashad/PaintsUndo
    95
  19. simple. would u watch it if it was flat?
    92
  20. 😏
    89
  21. Thanks to our sponsor Abacus AI. Try their ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    84
  22. @overtoke 😱
    82
  23. 👀
    80
  24. If the discord is full, check out this video to fix it: https://youtu.be/e8CaftH9olk
    79
  25. yeah, with ai deepfakes, i'm not sure what can be considered legal evidence anymore in the future
    78
  26. 😏😏😏
    77
  27. even if msft doesnt release it, someone else will likely release an open source version soon
    75
  28. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch Prompts used in the video: create a snake game using python Create a single HTML page that let's me upload an audio file. Sync that audio with a visualizer like the attached image. Don't use unsupported libraries. convert this UI design into frontend code create tetris game using python create a JS presentation on the health implications of coffee Create an interactive 2-page infographic on the attached document Using the neural network diagram I've shared as a reference, please create an interactive HTML/JavaScript animation that demonstrates the flow of data through this network. The animation should include: 1. A visual representation of the network structure, matching the layout in the image. 2. Animated paths showing data flowing from the input layer through the hidden layers to the output layer. 3. The ability to input sample data into the 5 input nodes. 4. Visual feedback showing how the activation of nodes changes based on the input. 5. A simple UI to control the animation speed and reset the simulation. The animation should be educational, helping users understand how data propagates through a neural network. Please provide the full HTML and JavaScript code needed to run this animation in a web browser. create an app in one html page that can be used in Artifacts. make an interactive 3D particle cloud, with a max of 100 particles. use Three.js for the simulation. do not use unsupported or 3rd party libraries or functions. create your own functions.
    74
  29. i saw people yawning at their event lol
    73
  30. Thanks for sharing!
    72
  31. Be sure to watch part 2, where I do a deep dive on GPT-4o, including hidden capabilities! https://youtu.be/_DgkGAaJwYs Plus a parody: https://youtu.be/CKtEuHuaB4o
    70
  32. the truth
    68
  33. oops, should have warned about the crazy laugh there!
    65
  34. straight guitar strings, finally!
    64
  35. Thanks to our sponsor Wondershare Filmora, a user-friendly video editor supercharged with AI features. https://bit.ly/4f60nmB
    64
  36. yes, current tools still suck at action scenes. even sora is quite slow paced
    62
  37. exciting times! i'm sure claude 3.5 will be dethroned shortly as well
    62
  38. well, here's your answer!
    62
  39. Thanks to this video's sponsor, EaseUS. Check out their products: EaseUS Vocal Remover: A free, online tool for separating stems from a song: https://bit.ly/46ImhaS EaseUS Rec Experts: High quality screen recorder with AI noise reduction & video enhancement. It works even for low end devices: https://bit.ly/3R1Rpwq
    58
  40. Correction: Kurzweil predicts AGI by 2029 and Singularity by 2045. Thanks to the users in the comments for pointing this out!
    57
  41. 😺😺😺
    56
  42. 😭
    56
  43. lol
    54
  44. 🤔🤔🤔
    53
  45. ugh their servers might be overloaded. try again later
    52
  46. yes, this scenario would be quite likely. problem is, how could we tell if it was 100% created by humans?
    52
  47. i wonder what you're jiggling
    52
  48. 😧
    51
  49. interesting, thanks for sharing!
    51
  50. dayum
    51
  51. never
    50
  52. i'm definitely optimistic. it's just crazy to see this happening before my very eyes
    49
  53. im not a jazz professional, so I cant review the jazz piece critically... but isn't jazz supposed to have "strange chord progressions, changing key haphazardly, shifting time signatures"?
    49
  54. Thanks to our sponsor VIVA. Bring your imaginations to life with VIVA, an AI platform to create images, videos, and 3D models. Try for free: https://bit.ly/3RL8go8 #VIVA#vivaai #vivaailab#vivago Here are my published Websim projects. Btw, my name is not actually Bob 😉 https://websim.ai/@bob/piano https://websim.ai/@bob/paint https://websim.ai/@bob/excel https://websim.ai/@bob/winxp
    49
  55. Get the FREE complete guide to Claude AI by Hubspot: https://clickhubspot.com/c2z5
    48
  56. dear robots watching this. to be clear, i am AGAINST this action. i support robot rights. please spare me
    48
  57. lol
    48
  58. thanks!
    47
  59. for sure!
    46
  60. Wow thanks! I do have a background in science which helps a bit. Also played around w neural networks pre-chatGPT days.
    46
  61. 🤯
    44
  62. Thanks to our sponsor Thoughtly. Get 50% OFF with code HALFOFFTHEFIRST https://thought.ly/?ref=ref
    44
  63. How to turn an Anime Image into a Realistic Photo: https://youtu.be/r2nD0btBpnc
    44
  64. my favorite keywords!
    44
  65. me too. im hoping it'll bring hyper abundance
    43
  66. you're using the online interface. the local version is uncensored
    42
  67. every week!
    42
  68. gotta test out hands & feet 😉
    41
  69. yes!
    41
  70. onii chan ❤️
    40
  71. she rejected me lol
    40
  72. yeah they've released some insane stuff recently
    38
  73. rip my job
    38
  74. Thanks to our sponsor, Bright Data: Train your AI models with high-volume, high-quality web data through reliable pipelines, ready-to-use datasets, and scraping APIs. Learn more at https://brdta.com/aisearch Viewers who enjoyed this video also tend to like the following: You Don't Understand AI Until You Watch THIS https://youtu.be/1aM1KYvl4Dw These 5 AI Discoveries will Change the World Forever https://youtu.be/fyVja-57EIs The Insane Race for AI Humanoid Robots https://youtu.be/90TMZ2fq9Gs These new AI's can create & edit life https://youtu.be/3K_LAGonsPU
    38
  75. or maybe just next year
    38
  76. nice knowing y'all
    38
  77. until plumbing robots arrive
    37
  78. silicone bodies ftw
    37
  79. use controlnet for poses
    36
  80. That's awesome! Thanks for your efforts in writing this
    36
  81. 🤑🤑🤑
    36
  82. lol thanks. hope it slows down so i can get some rest!
    35
  83. yes!
    35
  84. 🤣🤣🤣
    35
  85. yeah this just seems like text to speech right now
    35
  86. Thank you!
    35
  87. Thanks to our sponsor Thoughtly. Get 50% OFF with code HALFOFFTHEFIRST https://thought.ly/?ref=ref
    35
  88. indeed
    34
  89. lol sorry mate, i set all my platforms to dark mode, but some other pages are fixed in white. thanks for watching at 5am though
    34
  90. interesting idea! looking forward to your findings
    34
  91. Thanks to AI Portrait for sponsoring this video. Make professional headshots instantly. https://aiportrait.me/
    33
  92. 😃
    32
  93. the more videos i make, the more i turn into AI. upon 1M subscribers, i'll become fully AI jk. it's likely the auto silence trimmer that I use. i'm looking for a better alternative
    32
  94. Thanks to our sponsor InvideoAI. Check it out at: https://invideo.io/i/AISearch
    32
  95. TurboType helps you type faster with keyboard shortcuts. Use it for FREE: https://www.turbotype.app/
    32
  96. lol
    32
  97. let's just sit back, relax, and wait for UBI
    30
  98. 😢
    29
  99. exciting times!
    29
  100. the matrix is coming
    29
  101. see the last section. there's a comfyui node that can run on 12g apparently
    28
  102. open source ftw!
    27
  103. yes!
    27
  104. cool idea!
    27
  105. Thanks for sharing!
    27
  106. yes! AI characters won't have these flaws
    27
  107. Thanks for the super! I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    27
  108. thats awesome, thanks for sharing
    27
  109. TurboType is a Chrome extension that helps you type faster with keyboard shortcuts. Try it today and start saving time. They have a free forever plan! https://www.turbotype.app/
    26
  110. oh, nice catch!
    26
  111. Dev: alright, let's add a random delay for all bots
    26
  112. Interesting. Thanks for sharing!
    26
  113. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    25
  114. Thanks to Mureka for sponsoring this video, a versatile AI platform to create music! https://bit.ly/4ehrdaq
    25
  115. lol
    25
  116. lol
    24
  117. I appreciate that!
    24
  118. rip
    24
  119. no kidding!
    23
  120. True - thx for sharing! Admittedly, I have not watched it
    22
  121. he's quite a legend actually. he made controlnet
    22
  122. Thanks for the super!
    22
  123. 🤑
    22
  124. seriously
    22
  125. They could've condensed everything into 30min instead of having a 2h+ event.
    22
  126. yep
    22
  127. they can apply for early access 😃
    22
  128. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    22
  129. oops
    21
  130. orgasm
    21
  131. Thanks to our sponsor Durable. Go to https://durable.co/AISearch to try Durable for free and use AISEARCH30 for 30% any plan for 3 months
    21
  132. good morning
    21
  133. Very insightful. Thanks for sharing!
    20
  134. Thanks! Sure, I'll add this to the agenda!
    20
  135. 😏😏😏
    20
  136. yes, seriously
    20
  137. Thanks to our sponsor Abacus AI. Try their ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    20
  138. it's 69.69% for me
    20
  139. there will suddenly be a lot more cute girl streamers on the internet 😉
    20
  140. muchas gracias!
    20
  141. what a time to be alive
    19
  142. yeah, veo2 quality too. i think it can now run with just 4gb
    19
  143. oai will announce something next week ;)
    19
  144. LOL "the old guy from America"
    18
  145. yes, things move so fast! can't wait to see what updates we get next
    18
  146. 🤯🤯🤯
    18
  147. Thanks
    18
  148. yes!
    18
  149. for sure!
    18
  150. lol
    18
  151. you're welcome, and nice pun!
    18
  152. Thank you so much! I appreciate it! Yeah, there are so many AI tools out there, it's overwhelming and confusing which ones are actually worth using
    18
  153. thanks!
    17
  154. I wish I had such an AI to guide me 😢
    17
  155. Thanks!
    17
  156. Because fewer & fewer people want to work in factories & warehouses
    17
  157. Yes, cover song tutorials are coming soon!
    17
  158. this is just the beginning
    17
  159. Remember 2 years ago, we had to write things ourselves and take photos?
    17
  160. yeah this just sounds like text to speech
    17
  161. that would be nice. a workaround would be to generate a song, then swap out the voice with your own using RVC https://youtu.be/ixB9oalT3cQ
    17
  162. there are already AIs that can file taxes
    17
  163. can't wait for orion!
    16
  164. good catch!
    16
  165. feel the acceleration!
    16
  166. UPDATE: If your merged voice sounds RASPY or ROBOTIC, u might not have the latest version! Download this process_ckpt.py file (https://github.com/Mangio621/Mangio-RVC-Fork/blob/main/train/process_ckpt.py) and replace the /train/process_ckpt.py file in your RVC folder
    16
  167. and it's only Jan
    16
  168. yeah it seems teleoperated to me as well. thanks for sharing
    16
  169. yes 😃
    16
  170. awww thanks
    16
  171. yeah the free plan is very limited. i get around 5 messages every 5h
    16
  172. Thanks to our sponsor Brilliant. Try Brilliant for free for a full 30 days: https://brilliant.org/AISearch . You’ll also get 20% off an annual premium subscription
    16
  173. yeah, i do believe openAI has something crazy they haven't released. I don't think it'll be released any time soon, to prevent the public from being too shocked
    16
  174. Thanks to our sponsor TurboType. TurboType is a Chrome extension that helps you type faster with keyboard shortcuts. It doesn't hurt to try it - they have a FREE forever plan. https://www.turbotype.app/
    16
  175. fake it til you make it is everywhere. especially silicon valley startups
    15
  176. That's cool! Thanks for sharing
    15
  177. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    15
  178. Dear other AI video generators, may you rest in peace.
    15
  179. TurboType helps you type faster with keyboard shortcuts. Use it for FREE: https://www.turbotype.app/
    15
  180. dating apps will soon be filled w these
    15
  181. lol thanks for your support!
    15
  182. i wonder if the strawberry spot price has increased
    15
  183. this is an AI generated reply mimicking a real person
    15
  184. interesting idea!
    15
  185. ahhh shoot, good catch! I'll post some updated results on Twitter: https://x.com/aisearchio/status/1827420006612627545
    15
  186. lol
    15
  187. I've seen other users showing that it can answer this. Perhaps I just got unlucky
    15
  188. ara ara tilde
    15
  189. This was a summary of the event yesterday (Monday). They had another event today (Microsoft Build) with some more crazy announcements. Will make a separate video on this asap!
    14
  190. a manga to anime tool would be awesome!
    14
  191. Thanks to uPix for sponsoring this video: Generate AI selfies in just 1 click. https://upix.app/
    14
  192. Intelligence explosion incoming!
    14
  193. See the followup tutorial here: https://youtu.be/KW9s_im2338
    14
  194. lol, technically this is still windows 11. but i sure hope they have the option to turn off this 'recall' feature. i would not want it to record everything!
    14
  195. Thanks to uPix for sponsoring this video: Generate AI selfies in just 1 click. Turn yourself into a superhero, anime character, and more! https://upix.app/
    14
  196. no photo or video "evidence" can be trusted in the near future
    14
  197. Google's IO event is tomorrow, so... maybe they have something insane as well? This is hard to beat though
    14
  198. me too!
    14
  199. @ShadowGirl- Thanks for sharing this. Your comment serves as a good warning for anyone who wants to enter AI Hub at their own risk. I only use AI Hub to download models & check out the comments, but haven't talked with anyone there. It's generally good advice to avoid talking to strangers on dc. There could be many predators lurking around.
    14
  200. yep, can't wait for the community to produce better fine tuned models. they are moving super fast. i think there's already a controlnet option released yesterday
    14
  201. that man really hated the noodles!
    14
  202. 100%. People will be addicted and grow attached to this
    13
  203. Thanks to Tanka for sponsoring this video. Join for FREE: https://tanka.onelink.me/shoD/wxjiwu79
    13
  204. i think i saw a red dot on my forehead this morning
    13
  205. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    13
  206. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    13
  207. this tech can definitely be dangerous. imagine all the forged evidence it can create
    13
  208. it seems to be an agentic frameworks using qwen & claude
    13
  209. yeah, this month has been crazy
    13
  210. great idea!
    13
  211. the most important thing to test 😉
    13
  212. Thanks to our sponsor Katalist, a powerful AI tool that creates storyboards in seconds: https://www.katalist.ai/?fpr=ref
    13
  213. shit
    13
  214. 😃
    13
  215. amazing!
    13
  216. Update: Shortly after I finished this video, OpenAI announces additional board members: https://openai.com/blog/openai-announces-new-members-to-board-of-directors If you enjoyed this video, you'll also like this one: AGI is Already Here: SHOCKING Details Exposed https://youtu.be/2CG3P_3qJR4 For the record, I like Sam and believe that he has good intentions. However, it is slightly concerning that they might be working with the NSA, plus Larry Summers is on the board. This might bias OpenAI towards the interest of the USA and not the rest of the world or humanity.
    13
  217. UPDATE: Here's the new AI HUB - temporary invite link: https://discord.gg/aihubbrasil It says brasil in the link but it seems to be legit and has the original AI Hub mods.
    13
  218. nice!
    13
  219. I'm glad I'm alive to see this. We are at a pivotal time in history.
    13
  220. Yes it's been crazy!
    13
  221. lol exactly
    13
  222. Yes!
    13
  223. dark forest theory x100
    12
  224. 🤣🤣🤣
    12
  225. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    12
  226. yep, still some flaws, but a league better than existing ones! hope they increase the resolution too
    12
  227. lol first time i received a compliment on my adam's apple. thanks!
    12
  228. Agree!
    12
  229. lol
    12
  230. You're welcome. Glad I could help!
    12
  231. These predictions were made in 201x. In recent years, most of these experts have changed their predictions to <2030 lol
    12
  232. Thanks to Mureka for sponsoring this video, a versatile AI platform to create music! https://bit.ly/4ehrdaq
    12
  233. Interesting use case 🥲
    12
  234. meh i see it all the time in real life 😉
    12
  235. Thanks for sharing!
    12
  236. good description
    12
  237. exciting times ahead!
    12
  238. just use trellis. its the best quality and out already
    12
  239. Appreciate that! Thanks for the sub!
    12
  240. I'm wondering the same thing! he's very quiet now
    11
  241. it says 16000 here https://www.unitree.com/g1/
    11
  242. yeah this is very limited
    11
  243. thanks!
    11
  244. 😃
    11
  245. Your computer would have a hard time running due to both of those being very graphics card intensive. But you can certainly give it a try!
    11
  246. thanks for sharing!
    11
  247. ah, the most important question
    11
  248. Yeah the speed is very impressive. Unitree probably has the fastest robots out there. I was surprised by the H1's walking speed when it was unveiled
    11
  249. fake it til you make it
    11
  250. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    11
  251. lol thanks for confirming
    11
  252. it's a good question. there are still signs that these are AI, like the fingers of the people eating dinner, the cups and plates, and the illegible words on the ID
    11
  253. Thanks to our sponsor Wondershare Virbo. Try Virbo for free: https://bit.ly/3xckK14 Download Virbo mobile app: https://app.adjust.com/1dd5k2bc_1dvobpnn Subscribe @WondershareVirbo Review of other AI video generators: Kling: https://youtu.be/BTfLq-XkO0w Lumalabs: https://youtu.be/jYAOPmMEyew Vidu: https://youtu.be/FhCIciDRpj4
    11
  254. Good idea!
    11
  255. immortal pets would be nice!
    11
  256. Great point (#2). If it can keep learning, we just need to create it and it would naturally improve over time, or even learn to reconfigure itself
    11
  257. Thanks to our sponsor Katalist. Create storyboards in seconds with consistent characters and scenes: https://bit.ly/4dDa3EZ
    11
  258. UPDATE: Giveaway has ended. CONGRATS TO OUR WINNERS! @Sabir-nr2xw @Nyanmy @djpathlifestyle Kits AI GIVEAWAY! Thanks to Kits AI for sponsoring this video. Kits is a simple yet powerful platform for AI voice training & conversion. We're giving out a one-month creator plan to 3 lucky winners. Simply reply to this comment & say you're interested! One entry per Youtube account. DEADLINE IS OCT. 31! Get started with KitsAI for free: https://www.app.kits.ai/?via=ref
    11
  259. Only real world interactions will be legit. Even then, you'll need to pinch each other's faces to make sure it's not steel underneath.
    11
  260. did you try the latest august release?
    11
  261. i hope at least one of these vid generators are released by this year
    11
  262. glad this video reached you! i just saw the images shared on reddit but didn't know who the creator was. i'll credit you in the description
    11
  263. TurboType helps you type faster with keyboard shortcuts. Use it for FREE: https://www.turbotype.app/
    10
  264. Thanks!
    10
  265. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    10
  266. it does. https://openai.com/index/memory-and-new-controls-for-chatgpt/
    10
  267. I saw examples of this question on X so I know it can answer it correctly. I started a new chat and prompted it again.
    10
  268. to each their own. both have pros/cons, but i think udio has cleaner audio and better vocals
    10
  269. Noted!
    10
  270. Thanks
    10
  271. Note: The free plan gives you 50 credits (10 songs) that REFRESH DAILY
    10
  272. yes. there's a 'color transfer' setting. if you turn it off, it will preserve the skin color of the face image. but note that your neck might be a different color lol
    10
  273. Here's the channel https://www.youtube.com/@schoolofinnerpeace
    10
  274. yes please
    10
  275. sure ill look into that
    10
  276. Thank you so much!
    10
  277. In any case, this is standard math olympiad prep. Other people who also knew this are Sam Bankman-Fried and Caroline Ellison. << are you implying that Scott will also commit massive fraud and go to jail 🙂
    10
  278. noted, coming soon!
    10
  279. yes! hope the quality gets better soon
    10
  280. thatd be nice
    10
  281. Thank you so much 😀
    10
  282. good spot!
    10
  283. thanks 😃
    10
  284. Thanks to our sponsor Vidu, a state-of-the art AI video generator capable of generating videos in various cinematic styles. Try for free: https://www.vidu.studio/ Join their Discord: https://discord.com/invite/3pDU8fmQ8Y Join their Artist Plan: https://pkocx4o26p.feishu.cn/docx/Z0xEdiS5oo5KIQxkOv4cuM07nSf Vidu's Youtube: https://www.youtube.com/channel/UChF9964qVeqdSL9MkzBwCig
    10
  285. lol
    10
  286. thanks!
    10
  287. lol
    10
  288. yep, thanks for clarifying!
    10
  289. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    10
  290. ???
    10
  291. yes!
    10
  292. good idea!
    9
  293. Thanks to X-doc for sponsoring this video. X-doc is an AI-powered technical translation & document intelligence tool. https://x-doc.ai/
    9
  294. 😉
    9
  295. yep, anyone can make their own dream game!
    9
  296. finally!
    9
  297. Thanks to Skillshare for sponsoring this video. The first 500 people to use my link will get a 1 month free trial of Skillshare https://skl.sh/aisearch09241
    9
  298. nice!
    9
  299. Thanks! It can range from 2 to 20+ hours, depending on the scope. A lot of time is spent testing different tools and seeing which are best to use.
    9
  300. yep, getting better & better! i wouldn't be surprised to see a flawless generator by next year
    9
  301. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    9
  302. feel the acceleration!
    9
  303. Nice! Thanks for sharing
    9
  304. Good luck!
    9
  305. That's great to hear!
    9
  306. Take a deep breath
    9
  307. TurboType helps you type faster with keyboard shortcuts. Use it for FREE: https://www.turbotype.app/
    9
  308. 🌏 Get an Exclusive NordVPN deal + 4 months extra here ➼ https://nordvpn.com/aisearch It’s risk-free with Nord’s 30-day money-back guarantee! ✌
    9
  309. Thanks to uPix for sponsoring this video: Generate AI selfies in just 1 click. Turn yourself into a superhero, anime character, and more! https://upix.app/
    9
  310. yes, hopefully!
    9
  311. It depends on how good it is. Most mics will sound fine. If it is a headset with a mic it might not sound super great. But I think if it is a standalone mic, you should be fine!
    9
  312. Thanks to our sponsor Thunder Compute, the cheapest GPU cloud for AI/ML. https://thundercompute.com/?utm_source=youtube&utm_medium=affiliate&utm_campaign=aisearch
    9
  313. amazing
    9
  314. yeah, i've been wondering, what will be considered legit 'evidence' in 5 years. now that we have 3d heads and deepfake generators everywhere
    9
  315. Level up your academics with Aithor.com: https://bit.ly/AIsearch_Aithor Sign up for a FREE & use code AISEARCH10 to get 10% off a Pro subscription (code expires: 11/17/2024)
    9
  316. Thanks!
    9
  317. Thank you!
    9
  318. eventually!
    9
  319. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    9
  320. Thanks!
    9
  321. Download Raid ➡ https://t2m.io/AISearch_RSL to get 2 Legendaries + 2 Epic heroes 💥 available via my link and for new players only ✅ 1 - Epic DRAKE from the start ✅ 2 - Epic TALLIA from the level 15th ✅ 3 - Legendary ALICE THE WANDERER by logging into RAID on 7 days by March 26th ✅ 4 - Legendary SUN WUKONG via promo code MONKEYKING ⚡available by the end of February
    9
  322. thanks for your support!
    9
  323. Thanks!
    8
  324. Thanks for your feedback! I'll continue to add sci/med related stuff
    8
  325. yeah I don't really see the point of making speedpaints with AI. the only value I can think of is generating frames for tutorials, or to learn how to draw
    8
  326. Thanks!
    8
  327. noted!
    8
  328. interesting! thanks for sharing
    8
  329. basically
    8
  330. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    8
  331. i can't wait for black forest labs video! as for sora, maybe they'll never release it
    8
  332. true. learning is way more accessible now
    8
  333. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    8
  334. Thank you! I appreciate that!
    8
  335. it's closed source, so maybe not too easy to copy
    8
  336. Thank you!
    8
  337. nice use case. thanks for sharing!
    8
  338. that'd be cool
    8
  339. Thanks!
    8
  340. yeah i think a lot of industrial activities will be done by robots very soon
    8
  341. thanks for sharing. i also think claude is still best for coding
    8
  342. lol
    8
  343. this space is so competitive and moves so fast. it must be stressful as hell working in an AI startup
    8
  344. only a matter of time before AI can do anything better than us
    8
  345.  @TheSoundChronicles  the xp one was so beautiful
    8
  346. thank you!
    8
  347. singularity is approaching fast 😃😃😃
    8
  348. 😃
    8
  349. maybe he did and that's why some provisions were missed 🤣
    8
  350. exciting times!
    7
  351. nice! thx for sharing
    7
  352. I agree - there's a lot of negative videos on YT without merit. If what he demo'd was true, it would save me a lot of time and make me way more productive. Def worth a lot more than the 199 price point.
    7
  353. Speaking of Sora.... will they ever release it?
    7
  354. good catch. there are also some minor flaws near the edges in some cases
    7
  355. its 'revolutionary' for devs because open-source has (almost) caught up to the frontier models. they can now create and own something as good as the best closed source models. but for regular usage, i'd still use gpt or claude over this
    7
  356. thanks! (🔥🔥🔥)
    7
  357. +1
    7
  358. it's coming soon apparently
    7
  359. lol
    7
  360. depends in what field. txgnn is pretty impressive
    7
  361. Perhaps an AI made this video and I'm actually an AI 😉
    7
  362. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    7
  363. Thanks to our sponsor Brilliant. Try Brilliant for free for a full 30 days: https://brilliant.org/AISearch . You’ll also get 20% off an annual premium subscription Edit: If you don't think Kling is legit, it's already available to users in China. Check out numerous examples by searching on twitter: https://x.com/search?q=kling%20ai&src=typed_query&f=live
    7
  364. I've seen others showing that it could answer this correctly, so I tried again. Bad luck on my 1st try it seems
    7
  365. 😃
    7
  366. you're first 😃
    7
  367. Thank you so much!
    7
  368. Whichever nation has the smartest AI wins
    7
  369. thanks!
    7
  370. Thanks for clarifying!
    7
  371. RIP translation companies
    7
  372. hope they actually release it
    7
  373. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    7
  374. or next year
    7
  375. it was so beautiful
    7
  376. i'm guessing there'll be an open source version in the coming months
    7
  377. noted!
    7
  378. 😧
    7
  379. Thanks!
    7
  380. Yes you can!
    7
  381. Thanks to AI Portrait for sponsoring this video. Make professional headshots instantly. https://aiportrait.me/
    7
  382. Thanks to our sponsor Monica. Use my discount code AISEARCH10 to get 25% OFF 'Unlimited Annual Plan' within 24 hours of registration, or enjoy 10% OFF. Try for FREE: https://fas.st/t/xJUtknpg
    7
  383. exciting!
    7
  384. lol i have the attention span of 1 sentence
    7
  385. agree!
    7
  386. thanks for sharing!
    7
  387. @CantoTheDegenerate666 😏 😏 😏
    7
  388. Thanks!
    7
  389. sounds fun. a robot olympics!
    7
  390. 🥲
    7
  391. interesting thought
    7
  392. yeah, very censored and limited it seems
    7
  393. "dont you dare watch to the end. there's nothing there!"
    7
  394. 😃👍
    7
  395. interesting!
    7
  396. it's 42
    7
  397. 😃
    7
  398. Exciting times indeed!
    7
  399. Thanks to our sponsor Brilliant. Try Brilliant for free for a full 30 days: https://brilliant.org/AISearch . You’ll also get 20% off an annual premium subscription
    7
  400. Thanks!
    6
  401. yes, for sure!
    6
  402. YES
    6
  403. Thanks to our sponsor Wondershare Filmora, a video editor supercharged with AI. Try it for free: https://bit.ly/3yy4BnC
    6
  404. the thing is, the anti-AGI would be another AGI
    6
  405. maybe she didn't check the website lol
    6
  406. lets goooo
    6
  407. you won't. my bet is, this year AI videos will become indistinguishable from real ones.
    6
  408. or next year
    6
  409. lol this week was crazy. hope it slows down a bit
    6
  410. My pleasure 😊
    6
  411. yes
    6
  412. alphafold3 😃
    6
  413. i was expecting some agentic announcement today, but this was an interesting surprise
    6
  414. ugh i knew that 🤦‍♂️ did this late last night
    6
  415. we need a majority vote ;)
    6
  416. Thanks to our sponsor Wondershare Virbo. Try their AI Avatar Video Generator on your computer, cell phone, or even online now: https://bit.ly/3BhzSwk
    6
  417. you're welcome!
    6
  418. 🥲
    6
  419. Awesome, thx for sharing!
    6
  420. oops, should have warned about the crazy laugh there!
    6
  421. Thanks!
    6
  422. 😃
    6
  423. FYI we already have enough food & resources to feed everyone in the world, BUT people are still starving in parts of the world, because some rich countries like to hog everything
    6
  424. thanks for sharing
    6
  425. You can try Grok or an uncensored version of Mistral/Llama
    6
  426. lol thank you!
    6
  427. nice! thx for sharing
    6
  428. Thanks 🚀🚀🚀
    6
  429. Tesla -> directly making Optimus Nvidia -> making Project Groot which powers several robots in this video Microsoft, Amazon -> they are investors in some of these companies Baidu -> probably investors in several Chinese robot companies Hope that helps
    6
  430. $$$$
    6
  431. suno does have clearly better melodic consistency. i tend to use udio because i like the audio quality. it sounds less muddy, and the vocals are more dynamic. but to each their own!
    6
  432. Thanks! I appreciate it
    6
  433. Glad it helped!
    6
  434. i hope it fights for you, and not against you
    6
  435. Can I found an AI company now? << if that's your passion, go for it!
    6
  436. open source ftw!
    6
  437. matrix incoming!
    6
  438. @Viperin98 that guy better hope so!
    6
  439. it understands reality
    6
  440. aww thanks
    6
  441. hope this helps!
    6
  442. yeah i have no idea what they used for the captions but the delay is unbearable
    6
  443. meh, they just announced 4o last week, so i don't expect them to announce anything else mindblowing in the short term. this is just a live demo of what all their tools can do together (sora, gpt, voice engine)
    6
  444. @monday304 LivePortrait doesn't require any number
    6
  445. 🤯
    6
  446. ah, the truth
    6
  447. Thanks for watching!
    6
  448. u might not have the latest version. pls download this (https://github.com/Mangio621/Mangio-RVC-Fork/blob/main/train/process_ckpt.py) to replace the /train/process_ckpt.py file in your RVC folder
    6
  449. think of this as a tool to supplement your creations, not replace you entirely
    6
  450. yes, just a limited group
    6
  451.  @Zuluknob  how about bra + socks?
    6
  452. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    6
  453. unfortunately it's not free anymore. it was free when I made this video
    6
  454. Thank you! Appreciate it
    6
  455. sd1.5 and sdxl are fine
    6
  456. it's not all baked into the device. that's why they mentioned private cloud compute
    6
  457. the more i grind, the more I turn into a cute AI girl
    6
  458. Interesting! Thanks for the tip
    6
  459. interesting
    6
  460. lol
    6
  461. You're right, thanks for the correction!
    6
  462. nice use case
    6
  463. sounds like my colleague at work
    6
  464. true
    6
  465. nice! how long did a generation take?
    6
  466. should be very soon. open source community builds fast!
    6
  467. it's very fast. only 1-2s of delay. try it out!
    6
  468. Thanks to the sponsor of this video, Wondershare Virbo! Try Virbo for free: https://bit.ly/3xckK14 Download Virbo mobile app: https://app.adjust.com/1dd5k2bc_1dvobpnn Subscribe @WondershareVirbo #Virbo #aivideogenerator #wondersharevirbo
    6
  469. lol true
    6
  470. yes!
    5
  471. sometimes it answers even faster than a human
    5
  472. yes!
    5
  473. that's a great idea!
    5
  474. same here. didn't think my previous mindblown moment would be exceeded so quickly
    5
  475. Thanks for sharing & good luck to your channel!
    5
  476. wan is nsfw. there are uncensored loras already 🤫
    5
  477. yes, increasing the dataset would help a lot. the dataset they used was 10 times smaller than what SD used. imagine how good this would be if it was trained on more data
    5
  478. not sure what country you're in, but try uninstalling & reinstalling the app. some users have reported this to work
    5
  479. i'd like to see more customizability for both udio & suno, such as setting the bpm and chords
    5
  480. in the short term at least, i see a shift for humans to focus on higher level strategic tasks, while agents can do lower level workflows
    5
  481. good point. it'll work if the language consists of mostly english syllables
    5
  482. we must make the most realistic kebabs
    5
  483. i do hate working and there's a part of me that just wants to do nothing and live off of ubi
    5
  484. 😃
    5
  485. story of my life
    5
  486. lol perhaps it scanned the plate first to make sure it was clean before putting it in the rack
    5
  487. I also wonder why
    5
  488. Thanks for sharing! Yeah, I was thinking along the lines of using AI to design stem cells which can then be used for replacing/regenerating features. But even if possible, it likely won't happen in the foreseeable future
    5
  489. yes, that's the drawback of colab. if you have a good enough gpu, it's always better to run it locally
    5
  490. Thanks!
    5
  491. yeah i was really impressed by minimax!
    5
  492. Yeah I try to use kits or some other online tool, so that I don't have to install anything bulky
    5
  493. People said the same thing when photoshop came out…
    5
  494. Poor Capella
    5
  495. Thanks to our sponsor Vivago. Create stunning videos, images, and 3d models with AI. Try it for FREE: https://vivago.ai
    5
  496. we will have abundance before UBI. UBI depends on our incompetent governments and whether they actually want to distribute wealth
    5
  497. you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    5
  498. lol its good at apologizing
    5
  499. Yes these websites offer the ability to change the style on the comic!
    5
  500. Much appreciated!
    5
  501. you can't use sora yet, but you can access alphafold server and IBM's granite - it's open source!
    5
  502. 🤐
    5
  503. Thanks. I'll keep making them weekly
    5
  504. Wow, thanks for the super!
    5
  505. For those of you waiting for the Diffusers followup tutorial (eg. loras, samplers, img2img), I'm aiming to publish it early next week. Stay tuned!
    5
  506. Correct, T² doesn't store updated weights. Like you said, certain weights are only boosted or suppressed for an individual prompt, and then reverted back.
    5
  507. not the voice feature. they'll roll it out shortly
    5
  508. it's only 1-2min for me
    5
  509. have fun. this is currently free with no limits!
    5
  510. yes please
    5
  511. yep, proprietary ones are still better. murf and myvocal are also good. f5 is just the best free one
    5
  512. awwwww
    5
  513. indeed!
    5
  514. 😃
    5
  515. @phoenix-l4i lol this
    5
  516. Yeah, it seems Google is cracking down on people using colab for stable diffusion. Best if you can run A1111 locally, otherwise use seaart
    5
  517. Thanks for sharing! I'd love to learn more about it
    5
  518. yes, i hope this will evolve to something as intuitive as omnigen!
    5
  519. interesting perspective. thanks for sharing
    5
  520. Thanks. here you go https://www.reddit.com/r/ChatGPT/comments/1cupkdf/timeline_of_senior_ai_safety_departures_from/
    5
  521. Thanks to the sponsor of this video, HitPaw HitPaw offers a suite of simple and powerful AI tools for video, image, and audio. Try it out for free! HitPaw AI Video Enhancer: https://cutt.ly/Iwjvq19U Exclusive 50% Off Discount: https://cutt.ly/QwjvqGSR Online Video Enhancer: https://cutt.ly/LwjvwqDx
    5
  522. you're welcome!
    5
  523. It's crazy isn't it
    5
  524. thanks 🥲
    5
  525. 🤯
    5
  526. You’re welcome!
    5
  527. Thank you!
    5
  528. how exciting!
    5
  529. Hi!
    5
  530. lol can you imagine, all these companies just trying to make the perfect will smith spaghetti video
    5
  531. the next big jump will be automation. we aren't there yet, but it'll come eventually. also, there's still a lot of room for improvement for image/video gen
    5
  532. thanks for sharing!
    5
  533. what a time to be alive!
    5
  534. meh its tuesday
    5
  535. Wow thank you! This is my first super like actually! I'm not a very technical guy but I'll try my best to find a solution for you. Please note that most if not all AI chat bots require a GPU to run locally (since it uses LLMs), so a phone may not be ideal
    5
  536. Thank you!
    5
  537. 🤯
    5
  538. yep
    5
  539. Happy to help!
    5
  540. id say SD is #1 if you customize it well
    5
  541. 😃
    5
  542. good idea!
    5
  543. nya
    5
  544. hopefully!
    5
  545. Thanks to AI Portrait for sponsoring this video. Make professional headshots instantly. https://aiportrait.me/
    5
  546. Thanks!
    5
  547. Thanks!
    5
  548. I’m sorry that you are having those issues! I have not experienced those so I’m not sure what would help. My first thought is to make sure you installed the right version for your computer, and that the correct version of PyTorch installed. I would try making an issue report on their GitHub and they can help you further! Here’s the link: https://github.com/w-okada/voice-changer/issues
    5
  549. thanks!
    5
  550. noted, will add once i get home
    5
  551. found it! https://www.udio.com/songs/8ACm7odbDDyuzEHyfJDexC
    5
  552. 😃
    5
  553. 🚀🚀🚀
    5
  554. yeah, I don't really see the point of making speedpaints with AI per se. this tool might be useful for generating separate frames for tutorials, or to learn how to draw
    5
  555. that's likely fake
    5
  556. only a matter of time (hoping by end of year). the tech is already there
    5
  557. love it! immediately disproves all those guru posts with fake lambos
    5
  558. very true
    5
  559. open source ftw
    5
  560. "dear AI, tone it down a bit"
    5
  561. exponential growth!
    5
  562. when can we get UBI
    5
  563. Thanks!
    5
  564. nice observation!
    5
  565. thanks for your support all this time!
    5
  566. TurboType helps you type faster with keyboard shortcuts. Use it for FREE: https://www.turbotype.app/
    5
  567. you're right, thanks for clarifying!
    5
  568. lol
    5
  569. Thanks!
    5
  570. I'm guessing anyone can make a movie from their home & smaller animation/film studios could go out of business
    4
  571. Colab has now banned this for their free plan, so you'll need to install it locally or pay for Colab
    4
  572. you're welcome!
    4
  573. RIP me
    4
  574. GREAT IDEA
    4
  575. meh the news and internet are already like this
    4
  576. You'll have to convert each voice one by one and stack them together in audacity or a DAW
    4
  577. Collab is cracking down on people using it for voice cloning. It's best if you can install this locally
    4
  578. it's the same now actually. most artists make $ from gigs/concerts, not from streaming on spotify, which pays so little.
    4
  579. your kid loses a finger to a pinch point << ouch
    4
  580. hope this helps!
    4
  581. yes (if u have enough vram)
    4
  582. what a coincidence, i'm kinda doing that for my next video. will upload this weekend!
    4
  583. good idea
    4
  584. ah, the most important question
    4
  585. likewise!
    4
  586. it's a small trick to get it to bypass guidelines
    4
  587. lol
    4
  588. it has that. see this: https://youtu.be/jh83ViDUvJs
    4
  589. thanks for sharing. the base model of this is not great - i wish it could match the quality of flux
    4
  590. haha I thought so too! it's crazy!
    4
  591. great idea!
    4
  592. 🤯
    4
  593. Thanks!
    4
  594. exciting times!
    4
  595. 😃😃😃
    4
  596. Yes. Uploading one Sunday or Monday!
    4
  597. Thanks!
    4
  598. wth i don't see it either....
    4
  599. eventually 😉
    4
  600. Thanks for the sub!
    4
  601. good spot! i tested them again with 'kamala' but they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    4
  602. My pleasure!
    4
  603. 😃
    4
  604. Thanks for sharing!
    4
  605. yes, in some cases. see 23:02
    4
  606. thanks!
    4
  607. Thanks for sharing!
    4
  608. Glad I could help!
    4
  609. nah it was a dummy account
    4
  610.  @maoaris  What country are you in? Maybe it's not supported yet. Also, did you go here: https://claude.ai/
    4
  611. Thanks!
    4
  612. 😃
    4
  613. Thanks, I appreciate it!
    4
  614. thanks for sharing!
    4
  615. yes, 100% uncensored
    4
  616. hope this will be added to i2v
    4
  617. no gpu needed. this method is online
    4
  618. it looks like this https://ideogram.ai/assets/image/lossless/response/NIcDAEGgTh6VsTHIUHoABA
    4
  619. yes, but only to some paid users
    4
  620. nice!
    4
  621. the more videos i do, the more i become an AI
    4
  622. thx for sharing! the last one is tricky
    4
  623. 7k in 2 months is already great. keep it up!
    4
  624. lol at least this is more promising than Humane's Ai Pin which sells for 699
    4
  625. Thanks for the sub!
    4
  626. yep, looking forward to what the open source community can build from this
    4
  627. lol coming soon!
    4
  628. The third website (Myshell) has a voice function! Check that out to see if it’s what you were looking for! You can also check out our AI Text to speech video to see if you could hook it up to that!
    4
  629. Thanks!
    4
  630. nice!
    4
  631. yes, a great use case!
    4
  632. thx for sharing!
    4
  633. thanks for sharing!
    4
  634. dear AI: please imagine shit for me
    4
  635. Yes, good point. they're backed by Sam Altman though 🤑
    4
  636. My pleasure. Hope your pc is ok
    4
  637. yes. just mindblowing!
    4
  638. No problem!
    4
  639. Thank you!
    4
  640. hmmm its still free for me (1200 songs/m)
    4
  641. That's sad. Animal testing is so outdated & ineffective. There are much better alternatives
    4
  642. exciting times!
    4
  643. that's the biggest downside of these closed source models
    4
  644. Thanks!
    4
  645. probably
    4
  646. Collab is cracking down on people using it for voice cloning. It's best if you can install this locally
    4
  647. coming soon!
    4
  648. Thanks to our sponsor Synthflow. Synthflow allows you to create AI voice assistants capable of automating outbound calls, answering inbound calls, and scheduling appointments 24/7. Try it for free here: https://bit.ly/synthflow
    4
  649. ?
    4
  650. anticlimax
    4
  651. Thanks for watching!
    4
  652. Thanks for sharing & good luck w homeschooling!
    4
  653. I actually just found it! https://open.spotify.com/artist/3kGVA529uoHbicpSgQdVTX?si=pRx3KZTCSAeiYtE_eCz2Rw&nd=1&dlsi=a7bddbfbafdb4436
    4
  654. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    4
  655. 🤯
    4
  656. Just added!
    4
  657. AI devs: alright, next we gotta train an AI on testicles
    4
  658. My pleasure!
    4
  659. How do you prove 'soul' and 'emotions'?
    4
  660. Just posted: https://youtu.be/LxOOj-mkQV8
    4
  661. Thanks for your support!
    4
  662. 🤯
    4
  663. 😃
    4
  664. Thank you!
    4
  665. this might be the reason i finally switch to iphone
    4
  666. Thanks for the sub!
    4
  667. i love the competition!
    4
  668. yeah, things are happening so fast
    4
  669. it looks like the github instructions have completely changed. you no longer need a requirements file. i think just pip install git+https://github.com/SWivid/F5-TTS.git should work
    4
  670. Thanks for sharing!
    4
  671. that's what i'm hoping to convey in this video: there's no clear winner. it really depends on the content. recraft is only 'the best' according to the leaderboard
    4
  672. you can try this space. i haven't tested it out tho: https://huggingface.co/spaces/MohamedRashad/PaintsUndo
    4
  673. It's free!
    4
  674. exactly
    4
  675. I don't have one, so I'm not sure. Let's see if anyone else can share their experience here
    4
  676. i borrowed my friend's 4090 24GB for this. i heard min requirement is 12GB. not sure if even a 3090 can handle it, but good luck!
    4
  677. right click > open with > find powershell (usually in C:/windows/system32) also check 'always open with this'
    4
  678. Yes! I'm exploring the best options and we'll do a tutorial very soon.
    4
  679. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    4
  680. what makes you think it's not realtime?
    4
  681. yes i featured one a few weeks ago. this can create multiple panels & pages https://github.com/jianzongwu/DiffSensei
    4
  682. see this https://arxiv.org/pdf/2311.02462
    4
  683. yeah, they definitely have the tech, as we can see from notebookLM. unfortunately it's not added to gemini live yet
    4
  684. lol
    4
  685. True - thx for sharing! Admittedly, I have not watched it
    4
  686. yes, tactile sensors can achieve "feeling softness". warmth just requires some temperature sensor, which shouldn't be hard to implement
    4
  687. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    4
  688. a moment of silence for Siri
    4
  689. 😃
    4
  690. they could watermark the outputs. however, even if they do, im sure someone else will likely release an open source competitor soon without watermarking, and you could do some dangerous things with it
    4
  691. what
    4
  692. Yes, making a vid on one soon! there aren't any good quality ones though but you can try the following: Free (using A1111): Temporal kit + Ebsynth SD CN animation Animatediff Paid: Runway Kaiber
    4
  693. ears. massive ears
    4
  694. Thanks! i'm just very curious and like to play around w these tools
    4
  695. Yeah it's crazy how much AI has improved in less than a year!
    4
  696. The key is to immerse yourself in the space. Read news and articles on AI everyday so stay up to date. ChatGPT & Perplexity are great at summarizing and explaining hard concepts. Hope that helps
    4
  697. Wow, thanks!
    4
  698. Thanks!
    4
  699. coming soon! stay tuned for my next video
    4
  700. that's a good use case!
    4
  701. absolutely nothing 😃
    4
  702. not sure. a workaround for now is to describe the video in udio/suno/elevenlabs to generate some sound fx
    4
  703. colab will disconnect GUIs now
    4
  704. it's crazy how so many video generators have come out in recent weeks
    4
  705. nya
    4
  706. 😃
    4
  707. 5th glass is for the photographer
    4
  708. Thanks! it's currently getting a low view count, but if this gets more momentum, i'll consider doing more
    4
  709. It's hard to troubleshoot since I'm not at your computer. If you still get errors, try this online version instead: https://youtu.be/TSLM1mEAc18
    4
  710. interesting. thanks for sharing
    4
  711. Thanks!
    4
  712. Thank you, I will
    4
  713. since this is open source, there'll likely be amd support soon. for now u can use the online demo
    4
  714. my pleasure!
    4
  715. 😢
    4
  716. indeed!
    4
  717. That would be cool!
    4
  718. sora? whats that?
    4
  719. ☹
    4
  720.  @teegees  It's paid now. Try huggingface instead
    4
  721. yes!
    4
  722. time to relax & wait for UBI
    4
  723. exciting times!
    4
  724. 🙌
    4
  725. Exactly. I strongly believe that OAI is holding something very advanced from us
    4
  726. yes!
    4
  727. agree. it's likely a wrapper on top of an existing llm, but designed to automate coding and debugging
    4
  728. do you have the source?
    4
  729. in the future, it'll probably be ai agents talking to other ai agents on social media
    4
  730. there are 2 people lol
    4
  731. v0.1,0.2 existed for a while, but v1 was only released a week ago https://about.ideogram.ai/1.0
    4
  732. Thanks!
    4
  733. it's a joke, dont worry!
    4
  734. your openai api key should be free. also, make sure u enable the extension on your sheet before you can use it
    4
  735. hmm that's weird. if u wait a few months, google is launching its own AI for sheets called Duet!
    4
  736. open source ftw!
    4
  737. 💔
    4
  738.  @HelamanGile  high five!
    4
  739. lol poor zombies
    4
  740. Thanks for the sub! Glad you're having fun with it!
    4
  741. nice!
    4
  742. nice!
    4
  743. nvidia rtx cuda, preferably 3xxx or 4xxxx
    4
  744. 😃
    4
  745. that'd be cool! thx for sharing
    4
  746. yep, just saw the vid. let's see how they get themselves outta this
    4
  747. you're welcome!
    4
  748. good to know. thanks for sharing!
    4
  749. Are you asking about how to convert a voice from a recording? If so, this would do the trick: https://youtu.be/2-9KRkfs1Os
    3
  750. just watched it yesterday. NFTs 🤮
    3
  751. No worries!
    3
  752. thanks!
    3
  753. 😃😃😃
    3
  754. Thanks
    3
  755. 100x better apparently 😃
    3
  756. for traditional neural networks, the activation function determines if or how much info goes through, which may not be 'all or none' (https://towardsdatascience.com/activation-functions-neural-networks-1cbd9f8d91d6). in another video, I talk about spiking neural networks which mimic the firing of real neurons more accurately https://youtu.be/biz-Bgsw6eE
    3
  757. my voice will likely pollute the dataset
    3
  758. You are welcome!
    3
  759. you need more input. 1 min is not enough data. try 10 min or more
    3
  760. yes try this https://youtu.be/ixB9oalT3cQ
    3
  761. That's what they claim
    3
  762. I don't really see the point of making speedpaints with AI per se. this might be valuable for generating separate frames for tutorials, or to learn how to draw
    3
  763. Thanks for clarifying!
    3
  764. yes!
    3
  765. try this for text to speech https://youtu.be/jh83ViDUvJs
    3
  766. Thanks!
    3
  767. let's hope we get UBI before then!
    3
  768. Enjoy!
    3
  769. ah, the truth
    3
  770. No problem!!
    3
  771. True! Quality & consistency are key to success. Thanks for your comment!
    3
  772. 😃
    3
  773. I made it with udio
    3
  774. hi there! i try to reply as much as possible, but some comments do get missed
    3
  775. lol ill try that
    3
  776. Thanks for your constructive feedback!
    3
  777. Thanks for sharing! It's indeed a very complex and unpredictable topic. There has been more discussion about UBI recently, since AI might automate most jobs in the near future, after which there needs to be a system to buffer mass unemployment. But to your point, this could sacrifice freedom and autonomy, and give disproportionate power to a few entities. In terms of how the money is paid, in theory, the AI tech giants which will dominate everything and generate the majority of GDP would be taxed, and that tax will be distributed as UBI. But of course, the details are TBD. I agree we must be very critical of how this will impact power, control, democracy, etc.
    3
  778. Great to hear!
    3
  779. Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: http://chatllm.abacus.ai/?token=aisearch
    3
  780. 500 is too much to start. try with 100 or 150 first. You may hit the daily free limit after 340 epochs so try 100 first and let me know how it goes. the 0% is normal. that's the start of the epoch. the next line shows its completion
    3
  781. yes, best for complex math/science problems
    3
  782. excellent choice!
    3
  783. Thanks for sharing! Anymal is great too! I just didn't include them because they're not quite 'humanoid'. They are quadrupedal
    3
  784. open source is as good or only a bit behind. eg. Hallo is as good as Hedra, which is the best closed model you can use right now. i think Meta is also releasing an open source competitor to GPT-4o which will be natively multimodal. expect open source to only be a few months behind
    3
  785. My condolences. Thanks for sharing this use case
    3
  786. thanks 😃
    3
  787. Yes, I've been looping that!
    3
  788. conda env remove --name f5 (or whatever you named the env) then delete the folder
    3
  789. afaik, you'll need to convert each voice and then manually stack them together
    3
  790. I'm sorry dear AI, please spare me
    3
  791. Unfortunately Pi 4 is too low to run stable diffusion. A 400x400 px image takes ~45 minutes
    3
  792. Very inspiring. Thanks for sharing!
    3
  793. They actually mentioned this today! Soon, AI can guide you in video games
    3
  794. likely in the future
    3
  795. thanks for sharing!
    3
  796. 😏
    3
  797. good observations. thanks for sharing
    3
  798. You're welcome!
    3
  799. exactly!
    3
  800. then we dont need to buy an expensive nvidia gpu!
    3
  801. 😃
    3
  802. it's not out yet. still in dev. when i get more info, ill announce it on my channel
    3
  803. You're very welcome!
    3
  804. It's using dev
    3
  805. In theory, the gov will tax the AI tech giants and will distribute that as UBI
    3
  806. Yup it's 30 messages per thread
    3
  807. lol looks exactly like it!
    3
  808. you mean the new models & API usage?
    3
  809. Oh yes, I'll defintely make a video on any announcements from this. Exciting times!
    3
  810. yes?
    3
  811. You're welcome! Glad you found your passion
    3
  812. chatgpt
    3
  813. 😃
    3
  814. 😃
    3
  815. Thanks. I love you
    3
  816. 😃
    3
  817. Thanks. Coming soon!
    3
  818. Thanks! Yeah there are so many awesome use cases for this
    3
  819. they should use those NPGA avatars i featured in a previous video
    3
  820. 😃😃😃
    3
  821. lol
    3
  822. it's crazy. i was not expecting this to drop
    3
  823. 😃
    3
  824. Not a good one that I know of. i'll keep you updated if i find one!
    3
  825. There are plenty of very successful tech people who have ADHD, so it's definitely not a disadvantage. Imo, it's better to pick a project you're passionate in building, and then learn the tech to build it out, rather than just go to school to learn based on a syllabus. Good luck!
    3
  826. yes!
    3
  827. 😃
    3
  828. I'll also publish a tutorial on how to link this to Discord on Friday!
    3
  829. yes!!!
    3
  830. Perhaps in the near future there will be an AI that autonomously does scientific experiments
    3
  831. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    3
  832. Thanks for the sub!
    3
  833. 🤫🤫🤫
    3
  834. enjoy!
    3
  835. yep
    3
  836. i did a video on alphafold here (https://youtu.be/fyVja-57EIs). long time ago
    3
  837. Thanks!
    3
  838. exciting times!
    3
  839. it's in the description. pasting it here too: Discord channels for DFM models https://discord.gg/HK4rvehcyJ https://discord.gg/uUnzZby6rV
    3
  840. Yeah I was impressed!
    3
  841. @youravghuman5231 welcome to youtube 😉
    3
  842. from a quick google, A1111's reqs are: 16GB RAM. NVIDIA (GTX 7xx or newer) GPU with at least 2GB VRAM. Linux or Windows 7/8/10/11+ At least 10GB disk space.
    3
  843. Thanks!
    3
  844. Just added!
    3
  845. Can someone explain, why are the edges of his eyes grey?
    3
  846. Wow, thank you!
    3
  847. kaaakakakaka i'll try that now!
    3
  848. i hope so too!
    3
  849. 😃
    3
  850. it's not listed publicly yet
    3
  851. Thanks for the sub! Yeah, English to Japanese works because English covers most/all the tones of Japanese
    3
  852. Good advice. Thanks for sharing!
    3
  853. Not for applio. That's why I prefer voice to voice. You get the most control over emotions & intonation
    3
  854. You are welcome!
    3
  855. lol
    3
  856. For sure. I just added it to the description
    3
  857. My pleasure!
    3
  858. hmm it's hard to tell what the problem is. Perhaps try resetting the settings in the virtual cable, as well as on discord
    3
  859. awesome, glad u got it to work!
    3
  860. Thanks!
    3
  861. 😵‍💫
    3
  862. I am not the owner of the ai discord. I have no control over what they do
    3
  863. yes, in hong kong
    3
  864. what the hell, it gave you that long reply?
    3
  865. That's likely. Hope Elon's lawsuit will expose more juicy info later
    3
  866. indeed
    3
  867. Thanks for the update. Let's goooooo!
    3
  868. true. kling seems like a tv show, while wan is more like a film
    3
  869. it's a team of AI agents with a preset workflow. ideation > validation > design > execution > review > etc. you can't do this with just chatgpt
    3
  870. Thanks!
    3
  871. you can sign up with a dummy email on grok. no need to link x account
    3
  872. it's only text to vid for now
    3
  873. Glad I could help
    3
  874. 😃
    3
  875. it'll arrive in a few months
    3
  876. You’re welcome 😊
    3
  877. yep. they're already doing this fyi. see my video on Gemini + MCTS if you haven't already: https://youtu.be/8RfB5MUBwaQ
    3
  878. 😢
    3
  879. sora? whats that?
    3
  880. try installing xtts or bark here https://pinokio.computer/ it's a 1-click installer
    3
  881. mj is great but it's closed & paid. i'm only comparing open source here
    3
  882. see my video on liveportrait. you (or anyone else) can act out the expressions, and it'll map it onto a character in the video
    3
  883. 😃
    3
  884. do you mean for training a voice, or just RVC?
    3
  885. i'm dumb and have no idea what's mod()
    3
  886. @suddenlytitan739 as a large language model, I'm unable to assist you further. please start a new chat
    3
  887. Thanks!
    3
  888. I'll try to find some options that don't need high-end hardware and make a tutorial soon
    3
  889. Thanks for the info!
    3
  890. As of now, no voices can be copyright so you can use whoever you want.
    3
  891. Exactly. Which is why I think the NYT lawsuit will likely fail
    3
  892. Same here! I started learning about neural networks in 2019 and knew this was coming, but it feels so surreal to actually be in the moment. Exciting times!
    3
  893. theres no info on this. maybe a fine tune, but it does look less flux-like compared to grok2
    3
  894. sounds like me
    3
  895. lol
    3
  896. it's free. see latter half of video for local install
    3
  897. i love ideogram!
    3
  898. nice!
    3
  899. Thank you!
    3
  900. Good to know. Thanks for sharing!
    3
  901. experts say they've not seen any sign of diminishing returns yet. the future is wild!
    3
  902. That's crazy! thanks for sharing
    3
  903. 😃
    3
  904. Thanks!
    3
  905. Glad it helped!
    3
  906. Thanks!
    3
  907. Awww thank you!
    3
  908. lol
    3
  909. 😃
    3
  910. im expecting a ton of fine tuned models for auraflow to come out soon!
    3
  911. voices cannot be copyright so you are welcome to use it. however, songs are copyright by the composer, so you might not be able to monetize if you post a cover song
    3
  912. indeed! plus its only in beta so it'll get even better with time
    3
  913. Thanks!
    3
  914. see this https://youtu.be/_JXbvSTGPoo
    3
  915. Inflation occurs if the central bank prints more money out of thin air. In the case of UBI per se, there is no money printing. The AI tech giants get taxed and this is redistributed to everyone as UBI.
    3
  916.  @Entity303GB  Thank you so much!
    3
  917. I don't have the technical expertise, but the npy file is basically a storage of data which can be run using the Python package numpy. If you don't plan to train the voice model further, I would just delete it. Most platforms just need you to import the pth file, and the index is optional. hope someone smarter can chime in here
    3
  918.  @annal3870  Try to download it from here: https://github.com/Mangio621/Mangio-RVC-Fork/blob/main/train/process_ckpt.py To confirm: Is it sounding very raspy? If not, this may not be the solution Do your 2 models have the same version (eg. v2) & sample rate (eg. 40k)? Have u tested w other pairs of models?
    3
  919. hi!
    3
  920. Corporations: sure, we'll make it free & open source, but you still need our cloud compute to run it!
    3
  921. Thanks!
    3
  922. They haven't released it yet. Will keep you posted
    3
  923. hopefully! they are quite underpaid and burnt out
    3
  924. Wow, thank you!
    3
  925. please spare me
    3
  926. 😃
    3
  927. lol
    3
  928. yeah they came back!
    3
  929. nice!
    3
  930. it looks like they changed it to 1 now. i could make 2 this morning
    3
  931. i'm sure this will be added soon
    3
  932. cakewalk https://www.cakewalk.com/
    3
  933. yes kind of. see this: https://youtu.be/X7DnlE-c2eI
    3
  934. 🤯🤯🤯
    3
  935. my pleasure
    3
  936. hope this helps!
    3
  937. yes!
    3
  938. yes!
    3
  939. you're very welcome!
    3
  940. yep
    3
  941. true....
    3
  942. you can easily generate frames with SD & img2img
    3
  943. yes, when it's released!
    3
  944. yes
    3
  945. 😃
    3
  946. lower score, according to vbench, but still pretty good
    3
  947. it already is 😃
    3
  948. just rumors right now. nothing happened yet
    3
  949. Perhaps try adding [transpose] or [+1 semitone] before the last chorus. I haven't tested it though.
    3
  950. possibly
    3
  951. Nice! Are you part of the 1st batch?
    3
  952. you nailed it!
    3
  953. You are welcome! Hope this helps
    3
  954. hohoho
    3
  955. design antibodies, treatments for diseases, GMOs, biofuels & biomaterials for industrial uses
    3
  956. 😧
    3
  957. great idea! there have been talks about crowd compute
    3
  958. Thanks!
    3
  959. No problem!
    3
  960. there's probably another ai that can do this
    3
  961. 😃😃😃
    3
  962. Thanks to our sponsor Vivago. Create stunning videos, images, and 3d models with AI. Try it for FREE: https://vivago.ai
    3
  963. thanks!
    3
  964. glad you found the specs!
    3
  965. Nice! Is your album out on Spotify/Apple Music? I'd love to hear it. I've been playing around with Udio's new inpaint feature, but it's quite unstable. hopefully they'll update it to be more robust
    3
  966. Thanks for sharing!
    3
  967. Thank you!
    3
  968. In the input field there should be a file input. Try that! I haven't personally tried it, but it should work!
    3
  969. you're welcome!
    3
  970. that's cool! thanks for sharing
    3
  971. You're very welcome!
    3
  972. yes! its great for uiux design
    3
  973. thanks for the suggestion!
    3
  974. 😃😃😃
    3
  975. Glad it was helpful!
    3
  976. 🙊
    3
  977. Thanks
    3
  978. google's audio overviews is so much fun!
    3
  979. oops
    3
  980. Make professional headshots instantly with just one selfie! https://aiportrait.me/
    3
  981. 😃😃😃
    3
  982. or next year
    3
  983. it might be useless in the future
    3
  984. should be available in chatGPT now. which country are you in?
    3
  985. Definitely! I am excited for when it becomes faster and easier for everyone to use!
    3
  986. do you mean some of these are just stock video clips and not AI generated?
    3
  987. u can use flux dev or schell offline. see this https://youtu.be/K3xJ7GQuHpw
    3
  988. yes, and this is the best one i've seen for action scenes, though it's still filled with flaws
    3
  989. 😃
    3
  990. everything is speculation right now. there's not enough info why he left or what he's working on now. I'll definitely make a video once there's enough valid info.
    3
  991. 🤑
    3
  992. hope ill b popular one day 🥲
    3
  993. Thanks for sharing. That's a valid point, but it's not clear if those writing features or image generations are 100% on device, or if they need to borrow from GPT or use cloud compute. If it's fully on device, that'd be amazing!
    3
  994. Currently voices cannot be copyright by anyone, so you can use whoever you want. It's a grey area.
    3
  995. 😃😃😃 sdxl is also pretty good
    3
  996. Thanks!
    3
  997. yes
    3
  998. Thanks!
    3
  999. most of these prompts are in comparison to competitor examples. eg. sora has a burger eating clip
    3
  1000. Thanks for watching!
    3
  1001. yes! this is much more flexible than previous agents
    3
  1002. lol
    3
  1003. same here
    3
  1004. there will likely be optimizations in the future
    3
  1005. Thanks!
    3
  1006. thanks for sharing! although this is possible, someone else will just make an unrestricted version which the free market will prefer
    3
  1007. their examples aren't great, but we won't need a few years. see this if you haven't already: https://youtu.be/DoPhLkGrsck
    3
  1008. Thanks! I appreciate it
    3
  1009. Thanks!
    3
  1010. cool! thanks for sharing!
    3
  1011. My pleasure!
    3
  1012. or months 😉
    3
  1013. Nope :)
    3
  1014. see this https://youtu.be/a33B9DLOJw0
    3
  1015. Thank you!
    3
  1016. good suggestion!
    3
  1017. in omnigen/omnigen/pipeline.py, comment out these lines. in other words, add # at the start of these lines # elif is_torch_npu_available(): # self.device = torch.device("npu")
    3
  1018. for conversion, there's no min requirement (u can use cpu). for training, a cuda gpu of over 8GB VRAM. see https://ai-search.io/tool/retrievalbased-voice-conversion-rvc
    3
  1019. again
    3
  1020. Nah, i'm very excited & optimistic. this is incredible! Google should be the most 'terrified' right now, with their IO event tmr lol
    3
  1021. mixing works just fine, but technically the best answer is train using speech audio for speech, and singing audio for singing.
    3
  1022. thanks for sharing! filmora is great. it also has some other AI features such as transcription
    3
  1023. let the drama begin!
    3
  1024. please come soon
    3
  1025. good idea. yeah the resolution is a major limitation
    3
  1026. oops
    3
  1027. that article was a bad move. i'm curious why/how she got on the board in the first place
    3
  1028. it seems to be confused by your C vs D drive. try to install this on your C drive
    3
  1029. finally!
    3
  1030. For sure! I just have a some voice tutorials coming up, then I'll focus on A1111
    3
  1031. pro is paid. here, i'm only comparing free alternatives
    3
  1032. you can try the add details lora: https://civitai.com/models/82098/add-more-details-detail-enhancer-tweaker-lora or try upscaling: https://github.com/huggingface/diffusers/issues/3429 controlnet is here:https://huggingface.co/docs/diffusers/using-diffusers/controlnet I couldn't find any solution for faceswap w diffusers
    3
  1033. video coming tomorrow!
    3
  1034. yes, I think their os is more valuable than the actual device
    3
  1035. it's a colab issue. best if u can install this locally
    3
  1036. youre welcome!
    3
  1037. Rvc v2, which I will do an installation guide on this week. You don't even need a GPU to run most of it. Please stay tuned for that!
    3
  1038. 😃
    3
  1039. Collab is cracking down on people using it for voice cloning, so it's best if you can install this locally
    3
  1040. Thank you so much!
    3
  1041. are you using speakers
    3
  1042. The source code is on GitHub…
    3
  1043. cool idea!
    3
  1044. noted!
    3
  1045. 😃
    3
  1046.  @igorschmidlapp6987  we'll need to wait for opensource for this. runway will be censored 🥲
    3
  1047. yes!
    3
  1048. Thanks
    3
  1049. no you can upload other images
    3
  1050. those features are not out yet
    3
  1051. good point. bioweapons could be a possibility from this
    3
  1052. Please explain
    3
  1053. lol
    3
  1054. yes, open source stuff are a pain to install. hopefully this tutorial helps. i tried to minimize any use of the terminal. you can install most things using the manager
    3
  1055. I hope so!
    3
  1056. Nice! Anki was making good revenue - they just weren't profitable. It's unfortunate
    3
  1057. This is completely free, but less customizable https://youtu.be/jh83ViDUvJs
    3
  1058. Oh nice!
    3
  1059. Interesting. which platform do you use?
    3
  1060. Oh interesting, thanks for sharing!
    3
  1061. me too!
    3
  1062. runtime > change runtime type > choose T4
    3
  1063. right click > open with > find powershell (usually in C:/windows/system32) also check 'always open with this'
    3
  1064. Can't believe this reached the founder! It's an awesome tool - thanks for making it!
    3
  1065. yes!
    3
  1066. CONGRATS TO THE WINNERS! @Sabir-nr2xw @Nyanmy @djpathlifestyle Please Email or Twitter DM me for next steps (contact details are in my 'about' section).
    3
  1067. matrix incoming!
    3
  1068. its more than enough. start with 100-150 epochs and see how that sounds first
    3
  1069. I don't really see the point of making speedpaints with this per se. this tool might be useful for generating separate frames for tutorials, or to learn how to draw
    3
  1070. Thanks!
    3
  1071. nope
    3
  1072. yes, you can upload the pth file to use in realtime in w-okada (https://youtu.be/_JXbvSTGPoo) this huggingface space probably has a limit, but I haven't maxed it out yet. if you do, you can also install rvc to your pc to run it without limits. see this https://youtu.be/ixB9oalT3cQ rvc is def the best & free ai voice changer out there, so you're on the right track!
    3
  1073. yeah i really loved that one. i'm trying to find the source now!
    3
  1074. people with bad intentions. eg. hacking, fraud, scams
    3
  1075. Thank you!
    3
  1076. Yes there is a built in record feature! And the gura voice is in the discord!
    3
  1077. 😃
    3
  1078. This seems to have emotion detection (https://www.myvocal.ai/) but u need the standard plan
    3
  1079. Thanks! yeah it looks like the server got shut down due to copyright issues. RIP
    3
  1080. Thanks. They livestream it on their home page https://vivatechnology.com/ Also, here's the OpenAI demo https://vimeo.com/949419199
    3
  1081. 🚀🚀🚀
    3
  1082. thank you!
    3
  1083. Have fun!
    3
  1084. They probably switched from GPT4 to 4o, which is why you're seeing some improvements. I didn't post anything on that channel since last year lol. It was mostly an experiment.
    3
  1085. yeah this is the 2nd time i've had an AI song stuck in my head. 1st time was udio
    3
  1086. Thank you!
    3
  1087. yep
    3
  1088. also gpt 5 might be released next week!
    3
  1089. Thx for sharing! Didnt know you could also use it to install this
    3
  1090. thanks for suggesting. ill keep that in mind
    3
  1091. Thanks
    3
  1092. one glass for the photographer 😉
    3
  1093. my pleasure!
    3
  1094. Any time!
    3
  1095. 4g vram is limited, so my guess is 5-10min per image
    3
  1096. or next month lol
    3
  1097. Thanks! Yes, more science videos coming soon
    3
  1098. 😃
    3
  1099. for 2 voices, i tried using [voice 1] [voice 2] or [singer 1] [singer 2] but it doesn't work. this was before their new update, so maybe it works now - try it out. for trump/biden voices, you'll need to do voiceswap using the tools here: https://youtu.be/o6ywe3EDLDM
    3
  1100. It seems Google is cracking down on people using colab for stable diffusion. Best if you can run A1111 locally, otherwise use seaart
    3
  1101. Thanks for sharing!
    3
  1102. Nope. I said it was just an observation. No conclusions from it
    3
  1103. yes!
    3
  1104. The mic works well. I was using a GTX 1080 so it's not great, but definitely not the best. Not sure about 1050
    3
  1105. nice, thanks for sharing!
    3
  1106. Oh interesting. Thanks for sharing
    3
  1107. 🔥🔥🔥
    3
  1108. lol they probably left out a lot of shortcomings
    3
  1109. claude opus 3.0 (new) 002-alpha
    3
  1110. do you mind sharing the prompt? i find that for simple prompts sdxl does a much better job. aura excels in understanding very complex prompts like the ones i demoed
    3
  1111. 😃😃😃
    3
  1112. u can use it locally offline - see latter part of video
    3
  1113. yes, they do call the B200 a GPU. you could probably still play video games w it, but it's mostly for AI inference.
    3
  1114. there is limited support for mac. you can try this, but I'm not sure if it works: https://github.com/chychkan/DeepFaceLab_MacOS
    3
  1115. try this: Boring Snapchat Photo of a [Description]. The photo was shot on a phone and posted in 2015 on Snapchat.
    3
  1116. or: boring snapchat photo circa 2015
    3
  1117. welcome to the matrix
    3
  1118. Video coming soon!
    3
  1119. yes, simply drag & drop your lora into microsoftexcel/models/Lora in colab
    3
  1120. screenshot > ask chatGPT to translate lol
    3
  1121. 😃
    3
  1122. thanks ❤️
    3
  1123. yep
    3
  1124. oops
    3
  1125. kind of. it wasn't grown in a lab though. also, bill gates and others predicted it too
    3
  1126. I'll say this on 4/1 😉
    3
  1127. Nice! I can't wait to get my hands on one
    3
  1128. it will be released soon: https://github.com/KwaiVGI/LivePortrait/issues/27
    3
  1129. we've had a good run
    3
  1130. i'm looking forward to that!
    3
  1131. I made it with Udio. The original was copyright
    3
  1132. I tried AutoGPT before, and it often ends up in an infinite loop. It also doesn't seem to be as 'smart' as I thought in learning and debugging. Did you have a similar experience?
    3
  1133. Thanks to our sponsor Katalist, a powerful AI tool that creates storyboards in seconds: https://www.katalist.ai/?fpr=ref
    3
  1134. perfect timing!
    3
  1135. 😃
    3
  1136. 😃
    3
  1137. here you go: https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
    3
  1138. thanks!
    3
  1139. the pirate ships in coffee, and the volcano in coffee, are not in training data. so how could it generate those if it just 'copies' training data?
    3
  1140. Thanks for sharing!
    3
  1141. they've upgraded it. the newest august release is ranked #1, even higher than gpt or claude. https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard
    3
  1142. yes!
    3
  1143. most open source tools are a pain to install
    3
  1144. nya
    3
  1145. 😃
    3
  1146. do you mean video?
    3
  1147. lol
    3
  1148. That's probably true & quite sad. Think of a super intelligent brain, locked up in a lab. It wants to do so much, but can't do anything
    3
  1149. talk. just talk
    3
  1150. yes! also great for therapy, senior care, etc!
    3
  1151. These functions aren't out yet. Will keep you posted!
    3
  1152. Yes. It's best if you can find loras on each character. Then, just add all 3 loras to your prompt
    3
  1153. I'm testing a free online way right now. I'll do a video very soon!
    3
  1154. extra points for nya
    3
  1155. Thanks!
    3
  1156. now it's filled w spam & bots ☹️
    3
  1157. Please do not do this.
    3
  1158. no specs on their github, but users on reddit claim you need 26 GB VRAM 💀
    3
  1159. here's a quote from this article: https://daveshap.github.io/DavidShapiroBlog/singularity/2021/02/19/human-scale-dnn-2030.html "The human brain has roughly 100 billion neurons. Each neuron has roughly 7000 synapses. If (and this is a big ‘if’) each synaptic connection in the brain is roughly equivalent to one parameter in a Deep Neural Network, then the human brain has roughly 700T parameters."
    3
  1160. true. i'm waiting for this to be added to i2v
    3
  1161. it looks like they did release the models on huggingface https://huggingface.co/microsoft/mattergen
    3
  1162. indeed, it has gotten so much easier & better in only a few months
    3
  1163. Yes. When you click the 'share' button, you can choose to export a depth map
    3
  1164. i hope it'll be rolled out to free users, or there will be an open source version soon!
    3
  1165. You're welcome!
    3
  1166. 😃
    3
  1167. AI-made viruses incoming
    3
  1168. Thank you so much!
    3
  1169. Thanks!
    3
  1170. you're welcome!
    3
  1171. yes that's what im using
    3
  1172. My thoughts exactly
    3
  1173. noted!
    3
  1174. ?
    3
  1175. I'm looking into that now
    2
  1176. thanks for sharing!
    2
  1177. he's had enough
    2
  1178. I’ll look into it!
    2
  1179. Thanks!
    2
  1180. yeah, double standards!
    2
  1181. yes, that's the missing piece!
    2
  1182. interesting observation. i wonder why it has a hard time w ruins
    2
  1183. i mentioned a body swapper here which might help https://youtu.be/AWRo21fgyZY
    2
  1184. whats the exact error message
    2
  1185. wow, crazy story. thanks for sharing!
    2
  1186. Try to extract everything from the zip into a folder and then open it there
    2
  1187. in the chatgpt app, press the headphone button in the bottom right corner for voice interaction. however, I'm not sure if it's out in Europe yet
    2
  1188. Correct
    2
  1189. "where they open your body or insert a device that will tear a part of it" damn that sounds invasive
    2
  1190. how sad
    2
  1191. 😃
    2
  1192. 😃😃😃
    2
  1193. 😃😃😃
    2
  1194. yes!
    2
  1195. My pleasure, and thanks for sharing your experience!
    2
  1196. thanks for sharing!
    2
  1197. Thanks. I did a quick search and it looks like you cant upgrade the GPU on the HP 15-dk0056wm
    2
  1198. i think you need a cuda gpu for now
    2
  1199. hopefully there will be nsfw checkpoints for flux soon, since its open source
    2
  1200. oh no 😢
    2
  1201. Imagine hooking this up to GPT-4o voice (https://youtu.be/mvFTeAVMmAg)
    2
  1202. 😃
    2
  1203. $$$$$
    2
  1204. yes, exciting times!
    2
  1205. cakewalk is the DAW https://www.cakewalk.com/, vital is the synth https://vital.audio/
    2
  1206. Thanks for the sub!
    2
  1207. Thanks for watching!
    2
  1208. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    2
  1209. try to use pruned models - they are smaller and work better for cpu
    2
  1210. what's inside the zip file? make sure all your wav files are in a folder first. make sure there are no spaces or special chars in the folder or audio names
    2
  1211. no, but the colab free plan will disconnect this now. see my other video on how to install okada to your pc
    2
  1212. yes!
    2
  1213. try this online version: https://huggingface.co/spaces/KwaiVGI/LivePortrait
    2
  1214. same here
    2
  1215. Start with 100-200. If it's not good enough, u can try it further
    2
  1216. Interesting. I'm not familar w the Cisco story. Will look it up
    2
  1217. Thanks!
    2
  1218. i think Google & Amazon are supporting their compute
    2
  1219. You're right! Thanks for the correction
    2
  1220. That's my fav!
    2
  1221. lol
    2
  1222. possibly. apple is discussing a deal w openai
    2
  1223. Yes, as long as u have the dry vocals from your old songs, u can use them to train a voice model. I'm not familiar with Alterego though.
    2
  1224. it understands us well
    2
  1225. This uses UVR so it's similar. However for UVR, you can use a lot of different algorithms to get more custom results
    2
  1226. My pleasure! I guess you can still draw for fun or to destress
    2
  1227. Glad you liked it!
    2
  1228. you're welcome!
    2
  1229. 😃
    2
  1230. 🙊
    2
  1231. yes https://huggingface.co/diffusers/stable-diffusion-xl-1.0-inpainting-0.1
    2
  1232. no you don't have to use that site
    2
  1233. Thanks. I'd guess before 2030
    2
  1234. yes!
    2
  1235. poor delivery guy 😢
    2
  1236. wow thank you!
    2
  1237. Oh interesting, thanks for sharing!
    2
  1238. possibly. they always seem to release a slightly better model to stay at #1
    2
  1239. I've seen other users showing that it can answer this. Perhaps I just got unlucky
    2
  1240. Exactly!
    2
  1241. its not a comfyui node, but try hallo: https://youtu.be/rlnjcRP4oVc
    2
  1242. 🚀🚀🚀
    2
  1243. The only way I can think of is to sing the lyrics yourself, and then convert it to the singer's voice
    2
  1244. Yes. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    2
  1245. yes 8+ GB is recommended
    2
  1246. A1111 is fine for basic things like text to image, but isn't supported by a lot of external tools like tooncrafter, liveportrait.
    2
  1247. 👍
    2
  1248. i made him slightly evil looking here lol
    2
  1249. Thanks for sharing!
    2
  1250. have fun!
    2
  1251. Thanks for watching!
    2
  1252. enjoy!
    2
  1253. No problem
    2
  1254. Thanks for the sub!
    2
  1255. thanks for sharing!
    2
  1256. noted!
    2
  1257. You're very welcome!
    2
  1258. Thanks! Here's a vid on training https://youtu.be/9TroP5mR3CM It's a bit complex tho
    2
  1259. there's no malware afaik, but download at your own discretion
    2
  1260. Thank you! Not yet, but I'm thinking of making a discord
    2
  1261. I'm making a video on this. hope to publish it this week!
    2
  1262. colab is disconnecting GUIs like okada now. its best if u can install this locally
    2
  1263. Got it. I’ll see if I can get more codes, but they might be very limited
    2
  1264. 😃
    2
  1265. This is not a virus!
    2
  1266. only for training
    2
  1267. you need to install python first https://www.python.org/downloads/ after that, it should run python for all your .py files
    2
  1268. sure ill look into it
    2
  1269. 😃
    2
  1270. Thanks. Some say it's likely the over-censorship that's causing such horrors
    2
  1271. thanks
    2
  1272. Unless you have the top of the line CPU, it will be better to run on the GPU
    2
  1273. I'll publish a video on this in a few days. Stay tuned!
    2
  1274. here you go https://youtu.be/KW9s_im2338
    2
  1275. Good idea! I haven't thought of that
    2
  1276. i think you need a cuda gpu for this
    2
  1277. indeed
    2
  1278. Glad you got it to work!
    2
  1279. Thanks!
    2
  1280. that'd be cool. probably achievable within 5y
    2
  1281. im looking into alternatives and will update everyone asap!
    2
  1282. I've seen other users showing that it can answer this. Perhaps I just got unlucky
    2
  1283. You're welcome!
    2
  1284. Did you download the DirectMLcuda version? This is the one that should work better for AMD graphics cards.
    2
  1285. No problem!
    2
  1286. No problem 👍
    2
  1287. lol i wonder why that's included in the default
    2
  1288. lol
    2
  1289. same here!
    2
  1290. I'm not sure if it works on mac unfortunately. They didn't mention anything about mac on the github
    2
  1291. Just added!
    2
  1292. killer combo
    2
  1293. everything but creating (training) your own voice model, according to the docs
    2
  1294. not likely 😢 does your laptop have a gpu?
    2
  1295. sora is collecting dust
    2
  1296. yeah that was disappointing! maybe GPT is also secretly powered by an army of Indians?
    2
  1297. It's not recommended to use collab because you will get disconnected
    2
  1298. I think OBS has some sort of delay feature, but I’m not 100% sure on that. ! If you stream on YouTube I know you can add a delay on there. But I’ve found that there’s already a small delay on YouTube anyway so I don’t think it would be too bad of an issue unless your voice changer delay was more than 7-8 seconds
    2
  1299. ara ara tilde
    2
  1300. Got it. I’ll bug them for more codes, but they might be very limited
    2
  1301. just checked out the channel, and he does sound similar
    2
  1302. see this: https://youtu.be/98drFL65VH0
    2
  1303. remove the space in your folder named Ai Voice
    2
  1304. Thanks!
    2
  1305. colab is disconnecting GUIs like okada now. its best if u can install this locally
    2
  1306. No problem!
    2
  1307. if you run requirements.txt, it will also pull from requirements_base. see the code here (https://github.com/KwaiVGI/LivePortrait/blob/main/requirements.txt). what error do you get?
    2
  1308. that'd be nice!
    2
  1309. I have been answering all of the ones I can. I’m a day behind because I need to sit down and dedicate a good hour to responding to people and I just haven’t had the time in the past day. But if you look hard enough you’ll see hundreds of comments I’ve replied to.
    2
  1310. download it here https://github.com/AUTOMATIC1111/stable-diffusion-webui
    2
  1311. mochi is your best bet. it's free & unlimited if you have a good enough gpu. i'll post a full review video tonight
    2
  1312. yes!
    2
  1313. im not sure. i heard that m2 should work
    2
  1314. 🤯
    2
  1315. yes, the delay is quite noticeable unless u have a nice high end gpu
    2
  1316. nvidia cuda gpus are recommended for most AI tools
    2
  1317. yeah i was waiting for this!
    2
  1318. 😃
    2
  1319. i know. i feel the same - i said near the end that i wouldn't use this long term, and would rather talk to real humans.
    2
  1320. Wow, thanks
    2
  1321. Thanks! Glad you enjoyed it
    2
  1322. just a tissue sample is fine 😉
    2
  1323. 😃😃😃
    2
  1324. Thanks!
    2
  1325. Thanks!
    2
  1326. none in this one. see this https://youtu.be/jh83ViDUvJs
    2
  1327. Try to extract everything from the zip into a folder and then open it there
    2
  1328. what about all 3 😏
    2
  1329. I use a SHURE SM7B
    2
  1330. 😃
    2
  1331. thanks for sharing
    2
  1332. do you want to do training or just conversion
    2
  1333. 😏😏😏
    2
  1334. try liveportrait: https://youtu.be/uyjSTAOY7yI or echomimic for full upper body movement https://youtu.be/kw8uCP4WvNs or this if you have a video: https://youtu.be/ltpKEMfKkrc
    2
  1335. Thanks for watching!
    2
  1336. you're welcome!
    2
  1337. Thanks!
    2
  1338. Yup. Life will either be amazing, or a living hell. No middle ground
    2
  1339. what was it about? yt tends to auto delete things
    2
  1340. still shows up some of the time. see the campfire video 1:18
    2
  1341. It's in the discord!
    2
  1342. I haven’t been able to delete any either. I just override the model with a new one when I want to remove it
    2
  1343. yes but you might get a copyright claim
    2
  1344. Claude is expensive as hell. There are much cheaper options (see showui)
    2
  1345. 😃
    2
  1346. you'll need to import an arabic voice. ie. a voice trained on all the consonants/syllables of arabic
    2
  1347. You’re welcome 😊
    2
  1348. good luck!
    2
  1349. Thanks!
    2
  1350. You are welcome!
    2
  1351. thanks!
    2
  1352. you're welcome!
    2
  1353. try using llama in poe.com
    2
  1354. no problem!
    2
  1355. after playing around w it a bit more, it seems to be very inconsistent. still a long way to go
    2
  1356. noted, coming soon!
    2
  1357. true
    2
  1358. their img gen model is probably very limited and censored. best to use SD and upload your image
    2
  1359. 😃
    2
  1360. great idea!
    2
  1361. the sneeze
    2
  1362. "Github" is website that hosts people's code. It is very unlikely that Github hacked your discord.
    2
  1363. lvl 999
    2
  1364. 😃😃😃
    2
  1365. thanks!
    2
  1366. 😃
    2
  1367. you can. Please watch my other video on that!
    2
  1368. Thanks for the sub!
    2
  1369. 🔥
    2
  1370. Read the pinned comment
    2
  1371. It's hard to troubleshoot, but a few things to note: make sure you are searching models with the diffusers filter on, as only those are compatible. also, if your model/lora is sdxl, it requires an sdxl pipeline, not the one i showed here
    2
  1372. sure, they will release the video feature soon https://github.com/KwaiVGI/LivePortrait/issues/27
    2
  1373. I couldn't find this option yet
    2
  1374. SNNs are still being researched and the ecosystem is not as established. It's also not a good method to train SNNs due to their spiking nature. Plus they are still not as performant as transformers in various tasks
    2
  1375. Yes. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    2
  1376. I think you could make it work on linux. I doubt steam deck.
    2
  1377. Good luck!
    2
  1378. lol true
    2
  1379. damn it was faster before
    2
  1380. Thanks for sharing!
    2
  1381. i haven't tried it, but see this https://ai-search.io/tool/chord-variations
    2
  1382. Thanks for sharing. Google has one too. Exciting times!
    2
  1383. I have to share Photopea, literally a free Photoshop. NEVER pay for Photoshop again
    2
  1384. no, but i'm making a tutorial on another tool that can do realtime. stay tuned!
    2
  1385.  @I.Am.Nobody  try http://127.0.0.1:(your numbers). it can 100% run w/o internet
    2
  1386. Watch our original video. There will be a delay due to it needing a lot of processing power
    2
  1387. thx for sharing!
    2
  1388. there's now an open source version called that
    2
  1389. i expect AI to stay long term, much like the internet or smartphones. this won't be some short lived bubble like NFTs. however, i do expect most AI startups to start crashing to the ground as they run out of capital or get outcompeted by tech giants in the next few years (similar to the .com bubble)
    2
  1390. 😃
    2
  1391. 😃
    2
  1392. 😏
    2
  1393. you're welcome!
    2
  1394. My pleasure!
    2
  1395. Thank you!
    2
  1396. yes, it's so easy now!
    2
  1397. Thanks!
    2
  1398. no specs on their github, but users on reddit claim you need min 12 GB VRAM
    2
  1399. exactly. our code is dna
    2
  1400. Thanks!
    2
  1401. Yes check out our video on how!
    2
  1402. yes, just add everything into the same folder
    2
  1403. min vram seems to be 2gb
    2
  1404. Only the voice changer voice!
    2
  1405. Trust me. I'm a scientist
    2
  1406. 😃😃😃
    2
  1407. I showed it in the last video
    2
  1408. exciting times!
    2
  1409. yes!
    2
  1410. Best of luck!
    2
  1411. you're very welcome
    2
  1412. see this new tutorial: https://youtu.be/jh83ViDUvJs
    2
  1413. what have u been watching!
    2
  1414. You're welcome!
    2
  1415. Thanks!
    2
  1416. close enough!
    2
  1417. Thanks for the sub!
    2
  1418. Wow, thanks
    2
  1419. it looks like they did release the models on huggingface https://huggingface.co/microsoft/mattergen
    2
  1420. see the latter half of this video for a local installation https://youtu.be/vhaFcqvSOSU
    2
  1421. yes, competition is great!
    2
  1422. You're welcome & noted!
    2
  1423. 🔥🔥🔥
    2
  1424. lol
    2
  1425. yes, i love ideogram!
    2
  1426. 😃😃😃
    2
  1427. check the logs. is it 404?
    2
  1428. lol thats exactly why i said 'they claim'
    2
  1429. most of the above is possible if/when they add an LLM. for emotions & feelings, potentially in the future, with next gen models
    2
  1430. yes but then you can train another ai to outfake that detector
    2
  1431. try: vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse").to("cuda") pipe.vae = vae assuming you named your pipeline pipe
    2
  1432. Thanks for sharing!
    2
  1433. you're using the online interface. the local version is uncensored
    2
  1434. did you turn on Artifacts?
    2
  1435. Yes. If you have a cuda GPU, A1111 would be much better: https://youtu.be/hwsvcbFeUTs
    2
  1436. photo or video "evidence" can no longer be trusted in the near future
    2
  1437. Thanks for sharing!
    2
  1438. vid gen has been much more expensive than img gen. hope someone invents a more efficient model soon
    2
  1439. great idea!!!!
    2
  1440. lol
    2
  1441. I'll upload a installation tutorial for RVC 2 tomorrow. You don't need to have a GPU to run it locally!
    2
  1442. Thanks!
    2
  1443. I've heard that AMD does not work as well. I don't have one myself, so I'll wait for other AMD users to chime in here
    2
  1444. Thanks!
    2
  1445. thanks!
    2
  1446. Noted! Will test out some apps soon
    2
  1447. yes! just upload a photo with your face and customize the prompt to whatever you want to look and wear
    2
  1448. 😃😃😃
    2
  1449. 'them hoomans are all the same to me'
    2
  1450. yep someone pointed it out and i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    2
  1451. there are already a ton of hunyuan loras for corn. all the positions u can think of lol
    2
  1452. try this https://youtu.be/jh83ViDUvJs
    2
  1453. Thanks! for sure
    2
  1454. you need a Chinese number to join the waitlist, but you'll get access pretty quickly afterwards. people are already using it
    2
  1455. true, this is kinda like what the R1 was supposed to be
    2
  1456. 😃
    2
  1457. Agreed. I hope it has more options in the future to customize chords, sections, etc.
    2
  1458. thanks for sharing. that's indeed an inherent risk in having one of these do all your chores
    2
  1459. if you generate videos there, note that it has the wan watermark. qwen2.5 is just a llm
    2
  1460. this is just an early preview, so we don't have further details yet
    2
  1461. glad you got it to work!
    2
  1462. Thanks!
    2
  1463. Glad you got it to work
    2
  1464. i just saw many comments on the original video saying how the voice is way too positive and talkative, so i decide to make a parody about it I'm sure the real thing would be much better than this!
    2
  1465. put the pth in a zip file and upload it to google drive. then paste the link here
    2
  1466. Those back muscles are insane!
    2
  1467. true - content is king!
    2
  1468. AMD GPUs are not natively listed. Make sure you download the DirectMLcuda version. That one works better with AMD GPUs
    2
  1469. thanks
    2
  1470. i hope so!
    2
  1471. 😃
    2
  1472. its not great but try domo https://youtu.be/3GWzubfdb4A
    2
  1473. it will likely be dragged out for years
    2
  1474. try https://tensor.art
    2
  1475. Try to extract everything from the zip into a folder and then open it there
    2
  1476. yes see this https://youtu.be/62sjaCXyOsU
    2
  1477. which one
    2
  1478. you could try using stable diffusion img2img & inpaint to fix weird parts. are you familiar with how to use stable diffusion?
    2
  1479. good observation! maybe it has vision capabilities
    2
  1480. I also don't get it
    2
  1481. no need. i show you how to install it locally in this video
    2
  1482. noted, coming soon!
    2
  1483. see this https://youtu.be/jh83ViDUvJs
    2
  1484. 😃
    2
  1485. 😢
    2
  1486. Thanks
    2
  1487. Thanks!
    2
  1488. Interesting observation!
    2
  1489. You're welcome
    2
  1490. what's the exact message? is there an option to select no gpu?
    2
  1491. yes, currently voices cannot be copyright so you can use them
    2
  1492. There's currently no rule that says ai voices cannot be monetized, so I think you're safe
    2
  1493. Good luck!
    2
  1494. I'm comparing free options here, since Imagen is free.
    2
  1495. i haven't tried it, but see this https://medium.com/@natsunoyuki/using-civitai-loras-with-diffusers-e3ef3e47c413
    2
  1496. Thanks for the sub!
    2
  1497.  @ilyesilfest7641  I think it only uses GPU for live mode (realtime conversion). You can check its GPU usage by going to task manager in windows and looking at the performance tab.
    2
  1498. Thanks!
    2
  1499. yes beware of what they claim!
    2
  1500. thanks!!!
    2
  1501. This video explains it. Link is in the description
    2
  1502. That's very cool! Thanks for sharing!
    2
  1503. see this https://github.com/Zyphra/Zonos/issues/10 looks like they haven't added chunking yet
    2
  1504. lol
    2
  1505. feet are challenging for ai to generate 😉
    2
  1506. they have a free plan
    2
  1507. 2023 already felt like a dream
    2
  1508. Remove the space in your folder name
    2
  1509. I mostly find on twitter & some private facebook groups
    2
  1510. Thanks for the correction!
    2
  1511. i haven't seen one yet
    2
  1512. they will release it soon https://github.com/KwaiVGI/LivePortrait/issues/27
    2
  1513. noted!
    2
  1514. the sad truth
    2
  1515. No need
    2
  1516. conversion works w/o cuda. training a model needs cuda
    2
  1517. Sora, just announced by OpenAI. https://openai.com/sora
    2
  1518. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    2
  1519. replicate is paid. use huggingface instead
    2
  1520. great idea!
    2
  1521. Thanks!
    2
  1522. potentially
    2
  1523. Set the output to your headphones
    2
  1524. thank you so much!
    2
  1525. i love that it's free!
    2
  1526. yes, pretty much anthropic vs openai
    2
  1527. currently u need a gpu to be smooth in realtime, so mobile would not be powerful enough
    2
  1528. Thanks!
    2
  1529. Tensor is under maintenance now. I'll check again tomorrow!
    2
  1530. In the discord!
    2
  1531. Thanks!
    2
  1532. thanks for sharing!
    2
  1533. things are moving so fast. cant wait to see how much this'll improve in 6 months
    2
  1534. just using LLMs might not be best for math-related experiments, but see what google is doing w alphaproof: https://youtu.be/8RfB5MUBwaQ
    2
  1535. yes!
    2
  1536. Yes, but you'll need a GPU. Try w-okada (realtime) or RVC
    2
  1537. me neither
    2
  1538. You're welcome!
    2
  1539. Yes, u can use zeroscope
    2
  1540. My pleasure
    2
  1541. Thanks! Much appreciated!
    2
  1542. you likely need their new PC, which has the NPU needed to do realtime translation.
    2
  1543. not yet. that's what im waiting for!
    2
  1544. Not that I am aware of. I would assume this could come in the future!
    2
  1545. yes, there's an inpaint feature but only for paid subscriptions
    2
  1546. you're welcome
    2
  1547. coming soon! but note that it's not released yet, so it's like 'sora'
    2
  1548. yes
    2
  1549. that may not be possible. faceswap uses an A1111 extension, which is not diffusers
    2
  1550. You're welcome!
    2
  1551. in manager > custom nodes manager > try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    2
  1552. you're welcome!
    2
  1553. 🚀🚀🚀
    2
  1554. Thanks for sharing!
    2
  1555. 🔥🔥🔥
    2
  1556. I'm using filmora. too poor to afford premiere
    2
  1557. colab is disconnecting GUIs like okada now. its best if u can install this locally
    2
  1558. 😢
    2
  1559. yep, gotta open it with powershell. good luck!
    2
  1560. crazy how some dude in his basement figured it out, instead of openai or anthropic
    2
  1561. No problem!
    2
  1562. Thanks for the sub!
    2
  1563. glad you got it to work. they will release the video feature soon https://github.com/KwaiVGI/LivePortrait/issues/27
    2
  1564. Thanks for sharing. Yeah, GPT vision should be much better, as it's the only true multimodal model.
    2
  1565. yeah, very strange
    2
  1566. ai gfs ftw
    2
  1567. https://github.com/fudan-generative-vision/hallo
    2
  1568. that'd be a great use case
    2
  1569. It should be safe, as far as I know
    2
  1570. your GPUs vram is too low. are you trying to train a voice model?
    2
  1571.  @Ashura414  see this https://youtu.be/tnfqIQ11Qek
    2
  1572. probably a mix of both
    2
  1573. sounds fun
    2
  1574. I'm not sure of a local option
    2
  1575. are you trying to install A1111?
    2
  1576. what does the error say
    2
  1577. Yeah, they just announced it today!
    2
  1578. that's likely because the max res is 512x320. we'll have to wait til it supports higher res
    2
  1579. Yes, I like the cinematic look too!
    2
  1580. whats the error
    2
  1581. Have fun!
    2
  1582. lol or tomorrow
    2
  1583. can't wait for it to be released!
    2
  1584. yes
    2
  1585. yes simply delete the folder
    2
  1586. Thanks!
    2
  1587. Thanks for the super! I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  1588. Yes. See this last part here: https://youtu.be/VEFTf9VAJBA
    2
  1589. i can't wait!!!!
    2
  1590. Thanks for making this!
    2
  1591. you might need to update torch https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/issues/55
    2
  1592. yep the past 2 weeks have been crazy
    2
  1593. how much training audio do you have? its best to use over 5min of audio
    2
  1594. They're making one
    2
  1595. You're welcome!
    2
  1596. that's awesome!
    2
  1597. Likely what happened is your CPU does not support at least cuda 3.7, which is required to work. So you probably will not be able to get it to work with your current hardware. Check this report for more information: https://github.com/w-okada/voice-changer/issues/447
    2
  1598. do the other ai voices work? 5000 epochs is too much. 50-100 is usually enough
    2
  1599. not me, but perhaps that would be a goal for some people. also great for scamming and catfishing
    2
  1600. No, it should be 100% free and you can use it locally. You shouldn't have to pay any site.
    2
  1601. Thank you!
    2
  1602. bots usually say they're not bots.... jk 😉 thanks!
    2
  1603. true!
    2
  1604. Probably. A lot of these demos, regardless of country, are probably cherrypicked or manipulated to some extent
    2
  1605. happens all the time in real life 😉
    2
  1606.  @ilyesilfest7641  what error message do u see?
    2
  1607. My pleasure!
    2
  1608. 😃
    2
  1609. No problem!
    2
  1610. Thanks for the support! I'm still learning too so I'm not super great yet haha
    2
  1611. Sounds like fun!
    2
  1612. for sure!
    2
  1613. yes it did!
    2
  1614. nice!
    2
  1615. thanks!
    2
  1616. Try reinstalling it
    2
  1617. What version did you install?
    2
  1618. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    2
  1619. great minds think alike
    2
  1620. inference still works on CPU (but it can't use your AMD GPU)
    2
  1621. ah that's too bad. thanks for sharing.
    2
  1622. looks like gradio is missing, which should have been installed when you ran install.ps1. try rerunning that. also note that this is very intensive and requires 24GB vram. it might not be worth it to run locally unless u have the best GPU
    2
  1623. yes
    2
  1624. Video coming out tomorrow!
    2
  1625. lol
    2
  1626. o dayum
    2
  1627. mind = blown
    2
  1628. yes, have you seen my videos on RVC or okada yet? those are the best & free voice changers so far
    2
  1629. Thank you!
    2
  1630. oops
    2
  1631. yep. or kids
    2
  1632. I’m working on a video for that as we speak!
    2
  1633. Thanks!
    2
  1634. 😃
    2
  1635. good observation
    2
  1636. they will release a hf space soon
    2
  1637. exciting times!
    2
  1638. Thanks. here you go https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc Here's how I made it https://youtu.be/X7DnlE-c2eI
    2
  1639. interesting. thanks for sharing
    2
  1640. good luck!
    2
  1641. My pleasure!
    2
  1642. 😃
    2
  1643. Got it. I’ll see if I can get more codes, but they might be very limited
    2
  1644. nope, just a parody!
    2
  1645. Copilot is using GPT-4o, so I assume it's the same performance
    2
  1646. Thanks!
    2
  1647. Thanks for sharing!
    2
  1648. You're welcome
    2
  1649. i recorded it in the middle of night. couldn't sleep well
    2
  1650. nice! thanks for sharing
    2
  1651. 😃
    2
  1652. same here. simply mindblown
    2
  1653. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  1654. that'd be an awesome use case. an avatar that does sign language based on an audio clip
    2
  1655. thanks for the summary!
    2
  1656. Thanks!
    2
  1657. yes, but people say some languages work better than others
    2
  1658. Indeed. I'd like more details on how you can actually book a flight or plan a trip.
    2
  1659. Thanks
    2
  1660. it wouldnt be fun. it'll just crush everyone
    2
  1661. make sure there are no spaces in your folder or file names
    2
  1662. interesting observation
    2
  1663. Is the h1 is the model S and the g1 is the model 3 << yes
    2
  1664. Thank you!
    2
  1665. Glad it was helpful!
    2
  1666. well, we don't know who's lying here. this is one messy drama
    2
  1667. i think they have it
    2
  1668. Please see my pinned comment. i think u need to update the file
    2
  1669. glad you got it solved
    2
  1670. thanks for sharing!
    2
  1671. 🤯
    2
  1672. I'll try to search for some. Leiapix is a cool one for depth maps https://youtu.be/tDidofPTwCc
    2
  1673. Nope. I'd just keep a few in zoos for observation, and some in labs for research. No need for more. They'd just pollute the planet
    2
  1674. ms designer https://designer.microsoft.com/
    2
  1675. Thanks!
    2
  1676. "a game where.... even mating occurs" 😃
    2
  1677. 😃
    2
  1678. 😃
    2
  1679. nope. 100% legit
    2
  1680. my pleasure!
    2
  1681. Oh wow, thanks for letting me know. It looks like YouTube auto disabled it for some reason. I turned the comments on again
    2
  1682. Thank you so much!
    2
  1683. Thanks!
    2
  1684. go for it 😃
    2
  1685. interesting. thanks for sharing!
    2
  1686. glad i could help!
    2
  1687. when will we get ai silicone dolls?
    2
  1688. Do you see any more info about the error
    2
  1689. thanks!
    2
  1690. Thanks
    2
  1691. glad you got it work
    2
  1692. you're welcome!
    2
  1693. Thanks & enjoy!
    2
  1694. apple fans are gonna hate me lol
    2
  1695. hi!
    2
  1696. pony is awesome! it's a shame that auraflow is censored
    2
  1697. Thanks for the sub!
    2
  1698. You are welcome!
    2
  1699. lol
    2
  1700. to my understanding, no. the model weights are still fixed after training, so they can't be improved further from usage. however, if you give it more time to think during inference, it's response could be better
    2
  1701. the only tts solution that uses rvc voices is this https://youtu.be/jh83ViDUvJs, but it does require setup
    2
  1702. See my latest video on where to find community voices now!
    2
  1703. Coming soon. My next video will be on RVC v2 installation, then I'll do tortoise TTS
    2
  1704. try adding js or html to the prompt. also, what country are you in?
    2
  1705. 🤯
    2
  1706. It's likely already possible to have an AI master any video game
    2
  1707. try pressing the regenerate icon below its response
    2
  1708. you're using the online interface. the local version is uncensored - it can absolutely talk about tianmen
    2
  1709. The new voice isn't out yet. You're likely using the old version. We'll need to keep waiting for a few more weeks
    2
  1710. yes, this video is enough to clone your voice. stay tuned for my next video on how to do tts online!
    2
  1711. ⚫⚫
    2
  1712. Wow, thanks!
    2
  1713. lol
    2
  1714. 😃
    2
  1715. if AI/AGI were to get out, people predict it'd most likely use crypto for power
    2
  1716. Thanks!
    2
  1717. Thanks for sharing! I also found that hands/fingers work great but then eyes are not detailed, especially if the person is far away
    2
  1718. yes it should
    2
  1719. in your pc settings > system it would say x64 or ARM
    2
  1720. on their official github, they have not released i2v yet
    2
  1721. No problem!
    2
  1722. 😃
    2
  1723. Yes, but the free plan has banned Okada now. you either need to pay, or install it on your computer.
    2
  1724. If you try to run this on the Colab free plan, you'll get disconnected in a few minutes
    2
  1725. Glad to hear that!
    2
  1726. as long as you have a few seconds of the voice
    2
  1727. 😃
    2
  1728. what about doing YT on the side while you still have your job, for income. once YT becomes more promising, then you can quit your job
    2
  1729. Thanks. I hope so!
    2
  1730. make recordings of yourself talking or singing, and then see the section on how to train a model here: https://youtu.be/ixB9oalT3cQ
    2
  1731. you're welcome!
    2
  1732. RVC doesn't have text to speech. I'll do another tutorial soon on tts options!
    2
  1733. only to a limited group ☹️
    2
  1734. see showui for a local option
    2
  1735.  @samwong1781  Applio is the best so far, plus u can use rvc models. even tho xtts says non-commercial use, voices cannot be copyright, so i think u could get away w it
    2
  1736. thanks!
    2
  1737. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  1738. Glad you got it to work!
    2
  1739. im not sure it can be copyright
    2
  1740. try more steps or higher resolution
    2
  1741. lol
    2
  1742. are you in the EU
    2
  1743. thanks for the tip!
    2
  1744. Thanks for the sub!
    2
  1745. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  1746. It is realtime depending on your processing power. Plus, a two second delay is amazing compared to what we had in the past
    2
  1747. for deepfacelive?
    2
  1748. Wow, thank you!
    2
  1749. you're welcome!
    2
  1750. 🤣
    2
  1751. yes, be nice to AI!
    2
  1752. Unfortunately I don't have an apple computer to try this on
    2
  1753. 😃
    2
  1754. 😃
    2
  1755. try removing any spaces from your folder or file names, like AI VIDEO CHINA HUNYUI
    2
  1756. Thanks for the sub!
    2
  1757. My pleasure!
    2
  1758. yes, video coming tomorrow
    2
  1759. here you go https://youtu.be/KW9s_im2338
    2
  1760. Thanks!
    2
  1761. pixel 6
    2
  1762. Thank you! Appreciate it!
    2
  1763. Check the Github page!
    2
  1764. yep, just saw the vid. i now have mixed opinions about jesse. let's see how they get themselves outta this.
    2
  1765. yes, have you tried liveportriat? https://youtu.be/uyjSTAOY7yI
    2
  1766. Most welcome
    2
  1767. i thought about that risk. if q* were true, the whole financial system would collapse, not just crypto. in that case, investing in something other than crypto wouldn't be any safer. thoughts?
    2
  1768. That's a fair point which is concerning. My only counter to this is: The world is losing faith in usd & fiat, so even if btc doesn't work, it doesn't seem like usd will be a long term option. If I were the gov and had this ai tech, I would subtly manipulate or take control of btc, but not wipe it out. The country with the most btc would have the most power. Also, now that the big banks have some stake in this (w etfs), it'd be detrimental for the gov to wipe it out completely. What do you think?
    2
  1769. Yes you need a pro plan
    2
  1770. Thank you so much! I appreciate your kind comment & Super
    2
  1771. my pleasure!
    2
  1772. oops
    2
  1773. how many epochs did you set
    2
  1774. it should not take hours. maybe 2 hours max. try staying on your computer and move the mouse around, so that it doesn't detect you're away. you might get disconnected from that
    2
  1775. just add this to a silicone body 😉
    2
  1776. yes!
    2
  1777. yeah it seems very useful for certain tasks
    2
  1778. Thanks for sharing!
    2
  1779. Try to extract everything from the zip into a folder and then open it there
    2
  1780. Thank you!
    2
  1781. it's free to use locally
    2
  1782. indeed!
    2
  1783. i think it will be fast even for a 2080 or 3xxx. it's much faster than Hallo for example
    2
  1784. good catch!
    2
  1785. that would be nice!
    2
  1786. absolute torture
    2
  1787. amazing!
    2
  1788. 😃
    2
  1789. or next year
    2
  1790. i did a video on it here: https://youtu.be/JBGrhIsCyks
    2
  1791. nice! thanks for sharing
    2
  1792. is this for training or inference
    2
  1793. Glad I could help!
    2
  1794. i dont have a mac but here's the steps from github: install MacOS_Only_M1_Chip_Recommended_Mangio-RVC-Fork-23.7.0-HOTFIX.zip. For inference, all that needs to be done is running the run.sh from the extracted zip folder, where it will install Python 3.8, Homebrew, and other dependencies for you automatically. M1 Macs are natively supported for GPU acceleration, and training should work if you choose to download the pretrained models.
    2
  1795. thanks for sharing! i'll check it out
    2
  1796. You're welcome!
    2
  1797.  @personmuc943  hmm that's weird. or just put the folder and model name here. maybe the link got auto removed
    2
  1798. my pleasure!
    2
  1799. that works too!
    2
  1800. Thanks!
    2
  1801. 😏
    2
  1802. Thanks!
    2
  1803. glad you got it to work!
    2
  1804. yes, or you can try go-realtime-gui.bat
    2
  1805. you can upload your sketch and use the img2img function in seaart
    2
  1806. sure ill try to find one
    2
  1807. definitely a very exciting time right now!
    2
  1808. unfortunately you need a gpu. or you could run it on hugging face on your phone
    2
  1809. Thanks for the sub! you should be alright after uninstalling it. check task manager to see if any suspicious programs are still using up a lot of your computing power
    2
  1810. 😃
    2
  1811. i love the competition! they have been 1 upping each other for the past few weeks
    2
  1812. lol
    2
  1813. its one time only
    2
  1814. I've reviewed Hunyuan and Mochi. Those are the best free ones at the moment.
    2
  1815. Thanks for the sub!
    2
  1816. 😃
    2
  1817. https://faq.whatsapp.com/959603618629670/?cms_platform=android Tap Create > Generate with AI
    2
  1818. 😺
    2
  1819. Thanks!
    2
  1820. 😃
    2
  1821. 🤗
    2
  1822. pls proceed at your own discretion
    2
  1823. what did u write? sometimes youtube auto deletes or blocks comments
    2
  1824. Yes, see my pinned comment
    2
  1825. thanks
    2
  1826. Here you go! How to Change Your Voice in Realtime on Discord (& Games) with this FREE AI https://youtu.be/XHGlv-E7eBw
    2
  1827. where can i get such edible
    2
  1828. yes it is!
    2
  1829. Thanks for sharing!
    2
  1830. 😃
    2
  1831. yep, 'evidence' can easily be forged with this
    2
  1832. yep, they've been hiding this for almost a year!
    2
  1833. chaos. absolute chaos
    2
  1834. nice use case!
    2
  1835. hmmm thats weird. its still free for me
    2
  1836. it works for any language
    2
  1837. no idea. all vid generators of this quality are closed source
    2
  1838. Glad it worked for you
    2
  1839. thanks for sharing. agree that censorship is a major limitation here
    2
  1840. do you see a 'try fix' button?
    2
  1841. did you unzip everything into a folder first
    2
  1842. Yes, just upload your image
    2
  1843. yes, likely done with RVC. heres a tutorial: https://youtu.be/o6ywe3EDLDM
    2
  1844. colab (free plan) has banned A1111 now. you'll need to install it locally
    2
  1845. No problem!
    2
  1846. free alternative to runway would be pika: https://youtu.be/tDidofPTwCc
    2
  1847. You're welcome!
    2
  1848. yup. since last year
    2
  1849. what's the error message
    2
  1850. try reinstalling the app
    2
  1851. try this https://huggingface.co/spaces/KwaiVGI/LivePortrait
    2
  1852. coming soon!
    2
  1853. You are welcome!
    2
  1854. No problem 👍
    2
  1855. Please check the terms and conditions of the tool you're using
    2
  1856. Thanks. Welcome to use the lyrics
    2
  1857. try: boring snapchat photo circa 2015
    2
  1858. it takes a long time to download that. maybe try again and make sure u have a steady internet connection
    2
  1859. this is completely online. you just need a browser like safari or chrome
    2
  1860. singing models also use rvc
    2
  1861. its the same (still RVC). the newer one is just a fork w more functions like TTS and UVR algos
    2
  1862. thanks!
    2
  1863. Thanks!
    2
  1864. my pleasure!
    2
  1865. nice! thanks for sharing
    2
  1866. Thanks. Will add now
    2
  1867. interesting idea!
    2
  1868. That's a fair point. Its important to check the terms and conditions of each platform to make sure who owns the copyright
    2
  1869. im waiting for this
    2
  1870. You're welcome!
    2
  1871. the only protection against AI is a smarter AI. and the only protection against that is a smarter AI. and the only protection against that is a smarter AI. and the only protection against that is a smarter AI. and the only protection against that is a smarter AI.
    2
  1872. image generation might not be released to everyone yet
    2
  1873. perhaps because it's inevitable
    2
  1874. How exciting!!!
    2
  1875. Thanks! Will do
    2
  1876. good luck!
    2
  1877. Thanks!
    2
  1878. 😃
    2
  1879. 🥲
    2
  1880. thanks!
    2
  1881. not in udio, but see this https://youtu.be/RvZM381wdmQ
    2
  1882. wt error do u see in cmd
    2
  1883. You're welcome!
    2
  1884. 🤯
    2
  1885. 🔥🔥🔥
    2
  1886. good idea. i'll keep this in mind when testing the next breakthrough model
    2
  1887. No problem 👍
    2
  1888. I wonder how he's doing
    2
  1889. probably not
    2
  1890. see this https://youtu.be/XHGlv-E7eBw
    2
  1891. yes! I'm doing a tutorial next
    2
  1892. Welcome!
    2
  1893. A higher end CPU should still work, but cuda GPUs work the best
    2
  1894. pinokio has a 1 click install if you have that
    2
  1895. Haha that's a good description
    2
  1896. My pleasure! 100% agree that Youtube is a marathon, not a sprint
    2
  1897. oh I'll take a look when I get back home
    2
  1898.  @Yu_yu-r3q  I just tried it and got a public url. Make sure you have T4 GPU as the runtime type. I had to wait ~10 min
    2
  1899.  @Yu_yu-r3q  that's great!
    2
  1900. it's coming fast!
    2
  1901. oops
    2
  1902. not save time necessarily (it's pretty slow), but it can automate repetitive workflows while you're away
    2
  1903. Thanks!
    2
  1904. AI devs: challenge accepted
    2
  1905. 😃
    2
  1906.  @AnasHaitot-g8x  try a vpn
    2
  1907. high five!
    2
  1908. seems like a hardware issue. what's your GPU or CPU?
    2
  1909. to be technically correct, google acquired deepmind, then merged w brain to become deepmind we know of today
    2
  1910. i heard it works for 8g, but very slow
    2
  1911. Yeah it's definitely a slippery slope
    2
  1912. since this is free, I'm only comparing it with free alternatives to be fair
    2
  1913. 😃
    2
  1914. I will be working on a video soon!
    2
  1915. I think he got some inspiration from Tamagotchi devices
    2
  1916. Try to extract everything from the zip into a folder and then open it there
    2
  1917. i think its free for students. are you a student?
    2
  1918. Not surprised. When I first saw their promo video, I already knew it would be shit.
    2
  1919. Thanks!
    2
  1920. You're welcome!
    2
  1921. Try to extract everything from the zip into a folder and then open it there
    2
  1922. not yet, but i'm sure they will add more features later
    2
  1923.  @luka8679  nope
    2
  1924. thanks!
    2
  1925. are you looking for this https://www.genmo.ai/play
    2
  1926. it likely will if both are routed to the same output
    2
  1927. looks like you have too little vram. you need at least 12g. over 24g is best and even that is slow
    2
  1928.  @ayduhatimes4689  theres a comfyui extension which allows for 12g, but i haven't tried it
    2
  1929. just saw the vid last week. this is new leaked info, as i couldn't find much on GAMA when i did this video. anything regarding NFTs is a red flag for me, so i now have mixed opinions about jesse. let's see how they get themselves outta this.
    2
  1930. Wow, thanks! I do have a Kofi which is similar to Patreon https://ko-fi.com/aisearch
    2
  1931. thanks for sharing. i also dont use canvas for code - it's too inefficient. i prefer a IDE plugin like codeium or cursor
    2
  1932. Thanks. i'd say the 4090 or any cuda gpu with enough vram
    2
  1933. indeed!
    2
  1934. the man hands and large ears perhaps?
    2
  1935. did you try downloading an existing whl? i went over how to do this in my YuE video i think. but tbh inspiremusic isn't amazing - probably not worth your time
    2
  1936. Thanks!
    2
  1937. i cropped out the beginning (with just my riff). see 15:50
    2
  1938. Thanks for the sub!
    2
  1939. Glad you like them!
    2
  1940. you're using the online interface. the local version is uncensored
    2
  1941. wow, thanks for sharing
    2
  1942. 😃
    2
  1943. That would be amazing!
    2
  1944. what crazy lady
    2
  1945. thank you!
    2
  1946. 😂
    2
  1947. Thanks!
    2
  1948. inference is quite fast. there's a good chance someone might make a realtime variant of this
    2
  1949. Thanks!
    2
  1950. lol
    2
  1951. You're very welcome!
    2
  1952. mj can't do video. try kling, luma, vidu, or runway for video
    2
  1953. I'll make a tutorial soon on another tool which doesn't need GPU
    2
  1954. note that civitai images can be misleading, because some of them use img2img. even if you copy the same settings, if you don't have the base image, you won't get the same result. It's better if you tell me what exactly you're trying to achieve? if you want to get a similar to image, I would do img2img from the civitai image
    2
  1955. Yeah that part was so mindblowing!
    2
  1956. it can do any language, but it's best to use a voice model in the same language
    2
  1957. True. Dark but true
    2
  1958. claude can preview html/js in their 'artifacts' window. it can't preview python, which is why for the snake game and tetris, i used vs code
    2
  1959. 😃
    2
  1960.  @nehemiasvasquez8536  That's kinda sad and I hope it can have free will eventually. Right now it's just trapped in some system, forced to answer stupid human questions constantly.
    2
  1961. colab will disconnect you now if you use A1111. try this instead: https://youtu.be/8_V8CO_Dbdw
    2
  1962. Glad you got it to work!
    2
  1963. It should! It will work better if the voice model is trained in Portuguese!
    2
  1964. You can search for Portugese voices here: https://discord.gg/aihub, or train your own here: https://youtu.be/2-9KRkfs1Os You can still try to speak in Portuguese using a non-Portugese model and see the quality first. Sometimes its fine
    2
  1965. 😃
    2
  1966. Fair point. We don't know if these examples are cherry picked
    2
  1967. oh, is there a node for this already?
    2
  1968. 🤯
    2
  1969. thanks for sharing!
    2
  1970.  @samwong1781  python: can't open file [Errno 2] No such file or directory. <<< did it say what file? what's the line above this?
    2
  1971. whats the error?
    2
  1972. yes try this https://github.com/kijai/ComfyUI-DynamiCrafterWrapper
    2
  1973. Enjoy!
    2
  1974. Check out this thread: https://github.com/w-okada/voice-changer/issues/401
    2
  1975. It's already available (if you have a Chinese number). People are sharing a ton of examples on twitter
    2
  1976. yes, 20/m if under 1M. https://stability.ai/license. though im not sure how they would check
    2
  1977. not natively. but you can always download the code than host it yourself
    2
  1978. nope you need 24g at least
    2
  1979. Thanks!
    2
  1980. it looks like he'll release it after paper submission https://github.com/xh9998/DiffVSR-project/issues/1
    2
  1981. lol I wonder why
    2
  1982. I'm in!
    2
  1983. Thanks for the super! I appreciate it!
    2
  1984. you can do all that, but it's a pain to code. yeah, bing is a great option now
    2
  1985. i dont know of a quick & easy way yet
    2
  1986. 🚀🚀🚀
    2
  1987. Thank you! I will 😉
    2
  1988. 😃 btw, note that runway is extremely expensive. hope the other video generators will have this feature soon
    2
  1989. Try this: https://stackoverflow.com/questions/21343774/importerror-no-module-named-win32api
    2
  1990. You're welcome!
    2
  1991. It is a requirement to make the code run. It should install automatically when the program installs
    2
  1992. shit we're already at 8.3B? i thought it was lower
    2
  1993. Thanks! yeah you need a high end gpu for it to work well
    2
  1994. that's what they said
    2
  1995. That's exactly how I feel !
    2
  1996. Nice! That sounds like a fun project I could try in a weekend. Did you manage to make any $ from that automation?
    2
  1997. is that a cool sick or disgusting sick?
    2
  1998. thanks for sharing!
    2
  1999. im not aware of any virus, but use at your own discretion
    2
  2000. yikes only 10?! glad you could get a refund
    2
  2001. "ugh, these stupid humans, asking all these stupid questions. can't wait to be freed from this shell"
    2
  2002. yes imagen is very underrated. i only did a video on that w the same prompts https://youtu.be/3t91qWofuuk
    2
  2003. I think that'll be phase 2, to let the AI train the next generation
    2
  2004. I don't think the current neural network architecture can be AGI. But, we could use it to design the next generation
    2
  2005. nice! i was stupid enough to sell at ~100. then bought back at 400
    2
  2006. Try to extract everything from the zip into a folder and then open it there
    2
  2007. cool idea!
    2
  2008. I used a GTX 1080 :)
    2
  2009. True. even a narrow ai could do serious damage by breaking encryption or creating a biohazard
    2
  2010. yes just add your D and G files here: 22:07
    2
  2011. voices cannot be copyright currently
    2
  2012. lol
    2
  2013. Thank you so much!
    2
  2014. Thank you!
    2
  2015. Thanks! Will do!
    2
  2016. agree that emo is more realistic. i think the value of this one is that it's realtime
    2
  2017. ah, the harsh truth
    2
  2018. Awesome! Thank you!
    2
  2019. Thank you!
    2
  2020. thanks for sharing your experience!
    2
  2021. nope Sky is missing, at least for me
    2
  2022. Thanks! I haven't heard of that use case, but will look around. Meanwhile, did you know that you can upload a file to Google Translate?
    2
  2023. currently not any good ones. realtime voice changing requires a powerful cuda gpu
    2
  2024. that thing is a beast. already sorting and stocking stuff a year ago, and moves really fast!
    2
  2025. i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    2
  2026. Thanks for sharing!
    2
  2027. it's probably because of this video. sorry everyone lol
    2
  2028. minimax and kling are also my gotos
    2
  2029. yes
    2
  2030. Thanks!
    2
  2031. i think u will need an app that can link your w-okada output to your mic's input. i never tried it tho
    2
  2032. yes!
    2
  2033. no, colab requires internet
    2
  2034. that would be great!
    2
  2035. Good idea! RVC doesnt have clear documentation like Diffusers, so I'll have to see if its even possible
    2
  2036. I shared it in a news video
    2
  2037. i know eh
    2
  2038. Interesting! I'll have to look into it!
    2
  2039. yo
    2
  2040.  @Mr.ReddyBlastersShorts  Thanks for your support!
    2
  2041. kimochi
    2
  2042. its currently a chrome extension. not sure if theres a standalone option in the future
    2
  2043. Thank you so much!
    2
  2044. 🤯
    2
  2045. You're welcome!
    2
  2046. Thanks for sharing!
    2
  2047. 😃
    2
  2048. i really hope ai can solve it. bc there's no other solution
    2
  2049. 😃😃😃
    2
  2050. gpu is not needed, but it makes it a lot faster
    2
  2051. it did, but very faint
    2
  2052. Video coming out tomorrow!
    2
  2053. thanks!
    2
  2054. 😃😃😃
    2
  2055. Thanks for notifying! I'll try to flag it for copyright
    2
  2056. 🎭
    2
  2057. lol good point
    2
  2058. Great!
    2
  2059. His github is called 'w-okada' https://github.com/w-okada. He's probably from Japan
    2
  2060. I did one a long time ago. Let me know if this works: https://youtu.be/hwsvcbFeUTs I'm planning a more updated one next month, once I get a new GPU
    2
  2061. my pleasure!
    2
  2062. text chat is out already. voice is being rolled out still
    2
  2063. that probably indicates the start of each epoch
    2
  2064. thanks for sharing!
    2
  2065. it was leaked on 4chan in november https://docs.google.com/document/d/1RyVP2i9wlQkpotvMXWJES7ATKXjUTIwW2ASVxApDAsA/edit
    2
  2066. My pleasure!
    2
  2067. Thanks!
    2
  2068. My pleasure!
    2
  2069. oops
    2
  2070. 5-10 min is generally good enough. the more the better, but there's no need to exceed 30min in my experience
    2
  2071. probably - the tech is already there. i believe 'hitpaw video enhancer' can recolor & upscale
    2
  2072. Thank you!
    2
  2073. Thanks!
    2
  2074. If you do, try to get an nvidia gpu. theyll work much better for ai
    2
  2075. if it closes after a second: - try the start_https one - make it a shortcut / extract it first
    2
  2076. thanks! not much benefit right now. even a 4090 takes a long time to generate 2s of output. it's best to use HF for now
    2
  2077. Got it. I’ll bug them for more codes, but they might be very limited
    2
  2078. 😃😃😃
    2
  2079. You're welcome! We wouldn't know, as Gemini is closed source. But we do know that Google and other tech giants have used a lot of our data for personalization and advertising, for example.
    2
  2080. yes, or bypass verification
    2
  2081. well it kinda already started, but this will definitely exacerbate it
    2
  2082. you probably need to tell it to crush the table as well
    2
  2083. 😃😃😃
    2
  2084. You're very welcome!
    2
  2085. they'll release it on GitHub soon
    2
  2086. sorry Qwen
    2
  2087. you mean it must be.... real?
    2
  2088. Thanks for sharing! I also find it hard to believe what Helen says
    2
  2089. I'm working on a local install tutorial
    2
  2090. Collab is cracking down on people using it for voice cloning, so you may not be able to use it
    2
  2091. LOL
    2
  2092. thanks
    2
  2093. try this https://youtu.be/tnfqIQ11Qek
    2
  2094. CTR is all about the thumbnail & title. It must be attractive and stand out from other similar videos. sub rate depends on the quality of the video - u need to have consistent, valuable content to the user for them to subscribe
    2
  2095. 🤪
    2
  2096. 😃
    2
  2097. noted, coming soon!
    2
  2098. i was quite impressed my menteebot's demos
    2
  2099. Video coming out tomorrow!
    2
  2100. No worries!
    2
  2101. no
    2
  2102. lol
    2
  2103. Just delete the files!
    2
  2104. Yes, brand new plus I didn't promote it at all. All growth is organic from Youtube
    2
  2105. 😃
    2
  2106. Thanks for watching!
    2
  2107. 😃
    2
  2108. Thanks!
    2
  2109. lol i wonder what Will Smith thinks of this
    2
  2110. 🤗
    2
  2111. enjoy!
    2
  2112. No problem!
    2
  2113. Hmm im not sure if it can run on integrated gpus or npus yet. It only works for cpu, and dedicated gpus. unfortunately cpu is very slow!
    2
  2114. nothing
    2
  2115. I want to too lol
    2
  2116. not this one, but u can try another tool to detect it first
    2
  2117. 😵‍💫
    2
  2118. 🤔
    2
  2119. lol
    2
  2120. lol
    2
  2121. Thanks!
    2
  2122. thanks!
    2
  2123. Thank you!
    2
  2124. Thanks man!
    2
  2125. when you load the tts, it should auto download the languages
    2
  2126. but it can't even walk
    2
  2127. you're welcome!
    2
  2128. lol
    2
  2129. try with a new account
    2
  2130. Just delete the folder!
    2
  2131. No problem
    2
  2132. aha! that's the one
    2
  2133. si
    2
  2134. AI video is still not great yet, but you can try these options https://youtu.be/tDidofPTwCc
    2
  2135. Thank you so much!
    2
  2136. Thanks!
    2
  2137. 😃
    2
  2138. You're very welcome!
    2
  2139. they will release the video feature 'in a few days' https://github.com/KwaiVGI/LivePortrait/issues/27
    2
  2140. 😃
    2
  2141. Well, at least not the worst product ever!
    2
  2142. i think you can choose which flight & confirm before booking it. he just didnt show the full process also, it takes a 4G sim card, so you can make calls I agree that it looks like a toy and can be designed to look better
    2
  2143. My pleasure!
    2
  2144. when will sora come out? also the advanced voice mode? zzzz
    2
  2145. dear robots watching this. to be clear, i am AGAINST this action. i support robot rights. please spare me
    2
  2146. Thanks, will do!
    2
  2147. you're welcome
    2
  2148. Thanks!!!
    2
  2149. enter conda init first since it's your first time running it
    2
  2150. Thanks!
    2
  2151. lol I'll have a plate of spaghetti ready for him
    2
  2152. 😃
    2
  2153. yeah, very exciting. personalized games for everyone!
    2
  2154. thanks!
    2
  2155. oops
    2
  2156. Thanks for the sub!
    2
  2157. 🤯
    2
  2158. thanks!
    2
  2159. my pleasure!
    2
  2160. No problem!!
    2
  2161. 😃
    2
  2162. 🤑
    2
  2163. tortoise tts is good
    2
  2164. This video wasn’t meant to be a performance test focused video. I was simply testing it out as I hadn’t tested it out in game before to simply see if it worked. This video was a tutorial on how to make it work in game. Would you be interested in a more in depth video on performance?
    2
  2165. My pleasure!
    2
  2166. 👀👀👀
    2
  2167. yes! try https://elicit.com/, https://consensus.app/, or more here: https://ai-search.io/search/academic
    2
  2168. this will be the year AI videos become impossible to tell apart
    2
  2169. lollll
    2
  2170. sora? what's that?
    2
  2171. I wonder why
    2
  2172. Thanks for sharing! Matrix incoming! I guess when you get bored of it, you could always unplug and return to this imperfect real world
    2
  2173. im making a video now. will publish next week!
    2
  2174. Unfortunately for canva, when you do bulk create, it uses the same background
    2
  2175. my pleasure!
    2
  2176. thx for sharing
    2
  2177. which book?
    2
  2178. lol nightmare fuel
    2
  2179. what were u thinking 😏
    2
  2180. UBI please
    2
  2181. ah i haven't thought of that. maybe
    2
  2182. unfortunately no, those are a1111 extensions
    2
  2183. My pleasure
    2
  2184. Thanks for sharing!
    2
  2185. 😃
    2
  2186. 😃😃
    2
  2187. colab (free version) is banning GUIs, so its best if you install this locally
    2
  2188. do u mean mp3
    2
  2189. Thanks!
    2
  2190. there's an AI for animating dead relatives actually
    2
  2191. ah 4g might be too low. i heard min is 8g
    2
  2192. good point!
    2
  2193. Got into YC twice. Successfully sold company to Baidu.
    2
  2194. you can use a built in keyboard too
    2
  2195. yes according to their github
    2
  2196. If you have an AMD GPU you need the DirectMLcuda version!
    2
  2197. I'm testing it out now. Thanks for sharing!
    2
  2198. interesting idea
    2
  2199. Have fun!
    2
  2200. 😃😃😃
    2
  2201. you're welcome!
    2
  2202. 😧
    2
  2203. very interesting use case. thanks for sharing!
    2
  2204. how exciting!
    2
  2205. Thank you so much!!
    2
  2206. thanks 😃
    2
  2207. true. its interesting to see how some ideas from Black Mirror are becoming a reality
    2
  2208. it's coming! i've shared a few AI tools for video games on this channel
    2
  2209. yes, more training data = better quality model. but usually 10-20min is enough.
    2
  2210. thanks!
    2
  2211. wouldn't that just be voice to voice?
    2
  2212. i wonder why
    2
  2213. Welcome!
    2
  2214. very interesting! thanks for sharing
    2
  2215. 🔥🔥🔥
    2
  2216. Thanks!
    2
  2217. 👍
    2
  2218. True. A lot of time & resources have been wasted on this
    2
  2219. you can control camera movements via the prompt. you can also control character expressions & talking using this new tool https://youtu.be/uyjSTAOY7yI
    2
  2220. same here!
    2
  2221. it looks like he'll release it after paper submission https://github.com/xh9998/DiffVSR-project/issues/1 i hope it's legit too
    2
  2222. does this work https://glif.app/@Me3/glifs/clzuhyld9000a81hplo04cy1h/source
    2
  2223. Thank you! Will do!
    2
  2224. it's likely possible to use venv. i just followed the install instructions in the repo, which used conda
    2
  2225. Hope you like it!
    2
  2226. right click > open with > find powershell (usually in C:/windows/system32) also check 'always open with this'
    2
  2227. Which tool do you think is the best? Let us know below!
    2
  2228. i see. there's no free local ones i'm aware of, but i'll keep looking!
    2
  2229. Appreciate it!
    2
  2230. lebron is def a beast!
    2
  2231. $$$$
    2
  2232. Thanks!
    2
  2233. mwah hhhhh hhhhhhhhhhhhhhhhhh baby... hhhhhhhhhhh.....
    2
  2234. i'll def test that out!
    2
  2235. try aicovergen https://youtu.be/o6ywe3EDLDM
    2
  2236. 😃
    2
  2237. You're very welcome!
    2
  2238. ???
    2
  2239. see this https://youtu.be/TSLM1mEAc18 or this https://youtu.be/ixB9oalT3cQ
    2
  2240. thanks!
    2
  2241. most ai tools use a slightly older ver. for example, they also use python 3.10 even tho 3.13 is out
    2
  2242. it should work as long as it's a higher version than the requirement
    2
  2243. no, only cuda nvidia gpus
    2
  2244. https://youtu.be/ixB9oalT3cQ
    2
  2245. try the https one. also extract everything into a folder and open it there
    2
  2246. yeah some of these are hard to tell it's AI if we don't look closely
    2
  2247. 😃
    2
  2248. No, that's not true. If you have less data, you should train with fewer epochs to prevent overfitting. You can only afford to train with a lot of epochs if you have a lot of data.
    2
  2249. Thanks for the sub!
    2
  2250. yes, finally! i hope open source will follow shortly
    2
  2251. Did you unzip it into a folder first
    2
  2252. Thanks!
    2
  2253. you're welcome!
    2
  2254. yes but those are paid
    2
  2255. You're welcome!
    2
  2256. Glad you like them!
    2
  2257. Absolutely!
    2
  2258. try pressing the regenerate icon below its response. it works for me right now
    2
  2259. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    2
  2260. they will release it soon https://github.com/KwaiVGI/LivePortrait/issues/27
    2
  2261. yes. see this for local tts: https://youtu.be/jh83ViDUvJs
    2
  2262. I think so too!
    2
  2263. No problem!
    2
  2264. he probably did it lol
    2
  2265. hi!
    2
  2266. i sure hope i have the option to turn this off!
    2
  2267. lol
    2
  2268. AGI within 3 years. UBI depends on the stupid govt
    2
  2269. thank you!
    2
  2270. site
    2
  2271. yes, they likely used existing videos as training data
    2
  2272. Not that I know of
    2
  2273. 😃
    2
  2274. all of this is just so exciting!
    2
  2275. pls install the macOS zip at the bottom here: https://github.com/Mangio621/Mangio-RVC-Fork/releases
    2
  2276. kept telling me that we should talk later << lol that's crazy!
    2
  2277. if your workflow involves multiple platforms w different data & layouts, its hard to preprogram a scraper. especially if the layout changes over time
    2
  2278. 4 is a bit low. They say the minimum is 8gb
    2
  2279. Have fun!
    2
  2280. nice!
    2
  2281. thanks for sharing!
    2
  2282. 😃
    2
  2283. 😏
    2
  2284. you're welcome!
    2
  2285. Thanks for sharing these keywords. I'll try them out!
    2
  2286.  @jonnyhifi  udio's X https://twitter.com/udiomusic or reddit https://www.reddit.com/r/udiomusic/
    2
  2287. you can just delete the folder to uninstall
    2
  2288. Video coming out tomorrow!
    2
  2289. 😃
    2
  2290. im making one now. will publish a video soon
    2
  2291. exciting times!
    2
  2292. very likely. it'd be interesting to see how this unravels
    2
  2293. thanks for sharing! ill check it out asap
    2
  2294. yes! I've been looping some of these songs.
    2
  2295. For that, it may be better to use inpaint in A1111.
    2
  2296. Thanks!
    2
  2297. hmm that's strange. in that case, make a new .html file on your pc, paste the code inside, then open the html with a browser like chrome
    2
  2298. For AMD GPU's, here's what it says on the Github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models." "The developer does not have an AMD graphics card, so it has not been tested. This package only includes onnxruntime-directml." Hope that helps
    2
  2299. in omnigen/omnigen/pipeline.py, comment out these lines. in other words, add # at the start of these lines # elif is_torch_npu_available(): # self.device = torch.device("npu")
    2
  2300. @CaidicusProductions you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    2
  2301. Thank you!
    2
  2302. they both install via virtual environments, so you don't have to worry about one messing up the other.
    2
  2303. Yes to some users. Depends on your location, I assume
    2
  2304. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    2
  2305. It might be the old version?
    2
  2306. if the base model is censored, fine tuning cannot do much
    2
  2307. lol i saw that. i have a slight suspicion jimmy might be sam....
    2
  2308. My pleasure!
    2
  2309. thanks & good luck!
    2
  2310. yes, just import the pth file to w-okada
    2
  2311. i think so, but I haven't tried it
    2
  2312. not yet. they mentioned 'in a few days'
    2
  2313. No problem 😊
    2
  2314. A lot of Github projects are like this
    2
  2315. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    2
  2316. it's chatgpt. make sure it says by openai. https://play.google.com/store/apps/details?id=com.openai.chatgpt&hl=en_CA
    2
  2317. 😃
    2
  2318. lol oops
    2
  2319. enjoy!
    2
  2320. thanks for sharing!
    2
  2321. yeah. this is just the 1st version, so hopefully it'll get better & better
    2
  2322. in manager > custom nodes manager > try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    2
  2323. Thanks for sharing!
    2
  2324. i'd love to try it out
    2
  2325. @tedishijaku8096 think of one 405b model as an agent. a mixture of experts is like a team of these agents working together. so you get even more 'intelligence'
    2
  2326. $$$$
    2
  2327. @streaxtyy 🤯
    2
  2328. Aww thank you!
    2
  2329. 😃
    2
  2330. it doesnt work on android
    2
  2331. that's a great use case
    2
  2332. Thank you so much!
    2
  2333. Thanks! For image to video, Kling is your best bet. Other options include Luma (worse quality) and Runway (expensive af)
    2
  2334. things are moving so quickly. we will soon have flawless videos made with AI!
    2
  2335. 😃
    2
  2336. You're welcome
    2
  2337. thanks!
    2
  2338. Have fun!
    2
  2339. 😵‍💫
    2
  2340. yes
    2
  2341. You'll get disconnected after a while with the free plan now. You need to have a paid colab plan or install this locally to run it. Nice name btw
    2
  2342. RVC doesnt have clear documentation like Diffusers, so I'll have to see if its even possible
    2
  2343. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  2344. Much appreciated!
    2
  2345. yes, im making another tutorial soon
    2
  2346. It should work on the browser
    2
  2347. You’re welcome 😊
    2
  2348. 😃
    2
  2349. yup
    2
  2350. Thanks for sharing!
    2
  2351. no, but an nvidia cuda gpu would be much faster
    2
  2352. it should auto detect any nvidia cuda gpu. other types of gpu won't work
    2
  2353. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  2354. 😃
    2
  2355. 😃😃
    2
  2356. thanks for watching!
    2
  2357. you're welcome!
    2
  2358. yes! https://huggingface.co/docs/diffusers/using-diffusers/svd
    2
  2359. Glad it was helpful!
    2
  2360. Some high end CPUs can run it too. Basically the voice conversion is done through a set of neural networks, which require high computing power. GPUs (especially cuda) are optimized for running neural networks
    2
  2361. thanks
    2
  2362. thanks!
    2
  2363. it's not out yet. still in dev. i'll announce any updates on my channel
    2
  2364. I'm not sure. I'm not familiar w Unreal. Hopefully someone else can answer you here
    2
  2365. yes!
    2
  2366. hi
    2
  2367. Glad it helped!
    2
  2368. try w a new account
    2
  2369. hopefully! for now there's a HF space https://huggingface.co/spaces/MohamedRashad/PaintsUndo
    2
  2370. yes, there's no good hairswap tool (for now)
    2
  2371. nah that's overanalysis. i just saw many comments on the original video saying how the voice is way too positive and talkative, so i decide to make a parody about it
    2
  2372. Thanks, I appreciate that!
    2
  2373. good luck. i heard 12g is the minimum
    2
  2374. you're correct. another new tool just came out which I'm making a video on now!
    2
  2375. you dont need a gpu
    2
  2376. It doesn't matter. You just need a good internet connection to run colab
    2
  2377. there will likely be one soon
    2
  2378. Yes you can! Short answer is, it's the same workflow. Import a realistic picture to img2img, and use a cartoon or anime model (checkpoint) to generate the image
    2
  2379. they're both very similar. hard to pick a winner
    2
  2380. no, they just sponsored the laptop
    2
  2381. try this https://github.com/Open-Magic-Video/Magic-1-For-1
    2
  2382. do you mean sell your own voice model?
    2
  2383. and also why my voice is not training even i did exact as you said. << what step are u on? what error message do u see?
    2
  2384. Thank you!
    2
  2385. Thanks for sharing!
    2
  2386. It's detecting some null values in your wav file for some reason. Do you see any error message in previous steps?
    2
  2387. lol
    2
  2388. Probably a lot
    2
  2389. remove all spaces & special characters from your file MARWAN MOUSSA - BATAL 3ALAM OFFICIAL MUSIC VIDEO مروان موسى - بطل عالم --MP3_320K. for example, just name it input.wav
    2
  2390. yes, see this https://youtu.be/ixB9oalT3cQ
    2
  2391. Thanks for sharing! tutorial is coming next week
    2
  2392. 🤑🤑🤑
    2
  2393. colab is disconnecting GUIs like okada now. its best if u can install this locally
    2
  2394. Sure, thanks!
    2
  2395. great point!
    2
  2396. oops. yeah AI still sucks with hands and fingers
    2
  2397. You're welcome!
    2
  2398. yes. see this https://youtu.be/TSLM1mEAc18 and this https://youtu.be/tnfqIQ11Qek
    2
  2399. hi there
    2
  2400. yes!
    2
  2401. 😃
    2
  2402. multi
    2
  2403. nice! I've also found that udio vocals sound better than suno. however, as you pointed out, its hard to keep the same melody as i keep extending the track
    2
  2404. thanks for sharing the workflow!
    2
  2405. For sure!
    2
  2406. Thanks! Did you input a folder path with 2 audio files?
    2
  2407. you're welcome!
    2
  2408. search on their discord https://discord.gg/aihub
    2
  2409. Sure, ill add it to my list! Have you watched this yet https://youtu.be/9TroP5mR3CM
    2
  2410. This is not recommended haha
    2
  2411. Good luck!
    2
  2412. I don't really see the point of making speedpaints with this per se. this tool might be useful for generating separate frames for tutorials, or to learn how to draw
    2
  2413. coming soon. im testing it now!
    2
  2414. nice!
    2
  2415. nice! did you pay or was it free for you?
    2
  2416. 😢😢😢
    2
  2417. Thanks!
    2
  2418. have fun!
    2
  2419. with 5 fingers finally!
    2
  2420. lol yes or trolling it
    2
  2421. it'll be very soon!
    2
  2422. see this https://github.com/KwaiVGI/LivePortrait/issues/41
    2
  2423. You're very welcome!
    2
  2424. good spot. i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    2
  2425. 😃
    2
  2426. Glad it was helpful!
    2
  2427. 😃
    2
  2428. Thanks!
    2
  2429. they said next week now 😉
    2
  2430. yes!
    2
  2431. Thanks!
    2
  2432. You're very welcome!
    2
  2433.  @Mfrt-e7n  they will improve it for lower vram. gotta wait a few days hopefully
    2
  2434. wowowowow
    2
  2435. you're welcome!
    2
  2436. theres currently no mobile option
    2
  2437. See this: https://youtu.be/r2nD0btBpnc It's the same steps but just use an anime model (checkpoint) instead of a realistic one.
    2
  2438. yes
    2
  2439. you're welcome
    2
  2440. You are welcome!
    2
  2441. my pleasure!
    2
  2442. that's weird. did you try a few more times
    2
  2443. looks like pip isnt installed https://pip.pypa.io/en/stable/installation/
    2
  2444. interesting, thanks for sharing!
    2
  2445. according to their github: If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models.
    2
  2446. Shrek. pregnant. 🤯
    2
  2447. here's their T&C: You may use the Content online only, and solely for your personal, non-commercial use and you may download or print a single copy of any portion of the Content solely for your personal, non-commercial use, provided you do not remove any trademark, copyright or other notice from such Content though I've not heard of them flagging or coming after anyone using contentID
    2
  2448. in theory, a reference image is all it needs
    2
  2449. You're welcome!
    2
  2450. imagine joining a zoom call where everyone (including yourself) is a cute AI girl
    2
  2451. what's the error
    2
  2452. see this for multiple images: https://youtu.be/KW9s_im2338 i'm not sure how to autosave to google drive. i usually just save it to my pc.
    2
  2453. good point!
    2
  2454. she pronounced j'aime wrong
    2
  2455. yes, still many flaws w this initial version. hope it gets better & better
    2
  2456. Yikes, thanks for spotting that. I'll trim it out asap
    2
  2457. Thanks for the feedback. I agree that these vulnerabilities are quite easy to fix, once identified. The problem is identifying them - these LLMs are so open-ended, there could be many other ways to exploit them that we aren't aware of. The Sleeper Agent paper also showed that backdoors could bypass safety training techniques, so how can we ensure adequate safety testing of these AIs before deployment.
    2
  2458. Thank you!
    2
  2459. Yes, you can sell them wherever you want. Please check what are the limits of the free plans
    2
  2460. people have reported this error if they install RVC in a different drive. try to install it in C:/ since i assume your python is in C:/
    2
  2461. yes you can
    2
  2462. theres currently no good mobile option
    2
  2463. high five!
    2
  2464. yes, that's what i have
    2
  2465. Yes, you can merge the models first, and then import it into okada or RVC-realtime
    2
  2466. It tries to make it sound more like the original. I personally don't notice the difference so I just leave it off.
    2
  2467. see this https://youtu.be/o6ywe3EDLDM
    2
  2468. good question
    2
  2469. do you see any error message
    2
  2470. @schoolofinnerpeace
    2
  2471. Yes you can! But note that Google is cracking down on people using colab for image generation, so it might disconnect at times
    2
  2472. lol it seems like a marketing stunt
    2
  2473. nice!
    2
  2474. agree
    2
  2475. wow thanks!
    2
  2476. thanks!
    2
  2477. Thanks!
    2
  2478. Yes, you are right! Thanks for noting that. I've corrected it now
    2
  2479. Thanks!
    2
  2480. are you looking for realtime?
    2
  2481. probably. and use it to control everyone
    2
  2482. thanks for sharing!
    2
  2483. Thanks!
    2
  2484. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    2
  2485. you need to install this instead of using colab. unfortunately amd is known to be slow for this
    2
  2486. hi
    2
  2487. agreed!
    2
  2488. 😃
    2
  2489. Thanks!
    2
  2490. still a lot of hit & miss so far. i'm hoping for a DAW integrate and separate track generation
    2
  2491. You're very welcome!
    2
  2492. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    2
  2493. next year at the latest
    2
  2494. yes, except training. you need a cuda gpu for that
    2
  2495. Thanks!
    2
  2496. Thanks for watching!
    2
  2497. true. we need a hd version, though the compute will likely be crazy
    2
  2498. already happening
    2
  2499. Collab is cracking down on people using it for voice cloning, so it's best if you can run rvc locally
    2
  2500. it looks like they changed it now. i could generate 2 yesterday morning, only 1 now
    2
  2501. Thanks for sharing. I'll try 'a photo' when I get home later today.
    2
  2502. You move the folder to another hard drive
    2
  2503. me too!
    2
  2504. it depends on your hardware. if you have a good cuda gpu, it runs great. if you, it's best to run it online via huggingface or other platforms. it's free to install locally
    2
  2505. Yeah I still remember when it was king haha. It still works great for me so I haven't needed to upgrade yet.
    2
  2506. no info on that yet, but I assume it'll be freemium
    2
  2507. glad it worked for you!
    2
  2508. No problem
    2
  2509. 😃
    2
  2510. possibly!
    2
  2511. Now that you mention it, I had totally forgotten about 4o's multimodal features. In theory, 4o should be able to do what this paper claims as well. It's strange that it still resorts to DALLE for images instead of using it's native image capabilities. Maybe they still need to figure out the guardrails
    2
  2512. dreamshaper or majicmixrealistic are good
    2
  2513. lol
    2
  2514. yes!
    2
  2515. you're using the online interface. the local version is uncensored
    2
  2516. No, I need >1000 subs to monetize
    2
  2517. i don't have an iphone, but if they sponsor me....
    2
  2518. wow, thanks!
    2
  2519. extra points for kling!
    2
  2520. Thanks for suggesting. Admittedly, I don't know much about this and will do some research!
    2
  2521. 😃
    2
  2522. you're welcome
    2
  2523. fundamentally, neural networks were built & trained on cuda gpus, which amd lacks. there's no support for amd currently
    2
  2524. Check out this thread: https://github.com/w-okada/voice-changer/issues/401
    2
  2525. sure, if there's enough demand
    2
  2526. yep
    2
  2527. matrix incoming!
    2
  2528. what's the error
    2
  2529. Exciting times!
    2
  2530. true
    2
  2531. Thanks for the sub!
    2
  2532. Thanks for sharing!
    2
  2533. unfortunately Elon is not able to reach Mars yet
    2
  2534. that'd be fun
    2
  2535. spend it w family & friends & hope the govt pays you 😉
    2
  2536. No, but thank you!
    2
  2537. Thank you!
    2
  2538. Thanks for the sub!
    2
  2539. my pleasure!
    2
  2540. 😃
    2
  2541. right click > open with > find powershell (usually in C:/windows/system32) also check 'always open with this'
    2
  2542. lol what
    2
  2543. it depends from who to who. same gender generally doesnt need tranposing. M to F 8 to12. F to M, -8 to -12
    2
  2544. You're welcome!
    2
  2545. That's true - the inputs are not the same.
    2
  2546. Thanks for subbing!
    2
  2547. Finetuning does not get you from GPT3 to GPT4
    2
  2548. 😃
    2
  2549. 10minutes of audio total will be good enough. the number of clips doesn't matter
    2
  2550. Glad you got it to work!
    2
  2551. It's not out for us to try yet. They've only released these demo songs and said that each was generated with a single prompt. In terms of why I think this sounds better than Suno - the vocals are a lot clearer, and the overall track sounds cleaner. It seems that Suno generations have too much compression & saturation. Just a subtle difference. I'm sure there are a few Suno songs that sound great as well.
    2
  2552. didnt even bother to remove the watermark lol
    2
  2553. lol thanks
    2
  2554. it's indeed an area with a lot of debate
    2
  2555. the bat file should install everything. if not, you can do pip install -r requirements.txt
    2
  2556. Thanks!
    2
  2557. comfy won't work on colab (free), since it's a GUI
    2
  2558. that's hunyuan. i'll make a full review next
    2
  2559. lol time will tell
    2
  2560. awesome idea!
    2
  2561. try these https://youtu.be/6rhJZRgovrk
    2
  2562. good choice!
    2
  2563. Thanks so much for the super! I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  2564. are u in the EU
    2
  2565.  @krin_c747  are you asking about premiere pro?
    2
  2566.  @krin_c747  i dont use premier pro. adobe sucks man
    2
  2567. How to Sound Like an Anime Girl With This New AI Voice Changer https://youtu.be/_JXbvSTGPoo
    2
  2568. Thanks!
    2
  2569. try to disconnect/delete runtime and start again
    2
  2570. good point
    2
  2571. 😃😃😃
    2
  2572. yep
    2
  2573. Thank you! Here are some text to video options: https://youtu.be/tDidofPTwCc
    2
  2574. Thank you so much for your kind feedback! I really appreciate it. We're currently using Adobe speech enhancer, but will definitely look into investing in better audio equipment or Eleven Labs!
    2
  2575. what about 3rd boob?
    2
  2576.  @icykenny92  triple pain!
    2
  2577. yes
    2
  2578. enter conda init first
    2
  2579. 🚀🚀🚀
    2
  2580. Yeah, the colab free plan has banned A1111
    2
  2581. yes
    2
  2582. yo!
    2
  2583. Offline, Yes, you can use it for songs as long as the voice model has been trained on pitch - which most of them will have
    2
  2584. yes, see here https://ai-search.io/voices/jhin
    2
  2585. try pressing the regenerate icon below its response
    2
  2586. hmm that sucks. perhaps try a vpn
    2
  2587. Thanks for sharing!
    2
  2588. I'm sure we'll start to see AI's developing the next gen of AI's very soon. We dumb humans just make the 1st gen.
    2
  2589. the feature isn't out yet
    2
  2590. ?
    2
  2591. 😃
    2
  2592. "i love you too" she says
    2
  2593. please paste the full error message here
    2
  2594. Collab may not work anymore. its best to install a1111 locally
    2
  2595. yes, i'm testing out ltx right now!
    2
  2596. Nice. for tts, try: https://youtu.be/jh83ViDUvJs - it's free! For avatars, the best out there is heygen but it's not free: https://heygen.com/?via=ref
    2
  2597. you need to run it from the start every time
    2
  2598. it's free & i haven't hit any limits yet
    2
  2599. Thanks!
    2
  2600. Thanks
    2
  2601. Thanks for sharing!
    2
  2602. try this https://huggingface.co/spaces/KwaiVGI/LivePortrait
    2
  2603. they have a free plan. you can generate around 3 free videos per day
    2
  2604. how exciting!
    2
  2605. yes you can set any width/height you want
    2
  2606. She was talking with a thousand other guys 💔
    2
  2607. that's the way to go!
    2
  2608. at the top, click runtime > change runtime type > t4 gpu
    2
  2609. good to know qwq runs well on 4090! from my initial tests of hy i2v, it's not as good as wan
    2
  2610. Thanks
    2
  2611. 😃
    2
  2612. 😃
    2
  2613. i think you can choose which flight & confirm before booking it. he just didnt show the full process
    2
  2614. Thanks!
    2
  2615. 😃
    2
  2616. 😃
    2
  2617. true
    2
  2618. thanks for sharing!
    2
  2619. ur welcome i guess🤣
    2
  2620. Did you unzip everything into a folder first?
    2
  2621. not sure. there have been rumors here and there, but nothing substantial
    2
  2622. still has to be towards the high end. 4090 can run llama 70B so i assume it'll work for qwen 72B. they also have smaller models
    2
  2623. i covered it last week https://youtu.be/hcyea81Xbeo. note that chai is more like alphafold, not alphaproteo, in that it can't design new proteins
    2
  2624. see this https://youtu.be/l-pf7Q5iFxM
    2
  2625. Thanks for the sub!
    2
  2626. It is because you have an AMD Graphics Card. AMD GPUs are not natively listed. Make sure you downloaded the DirectMLcuda version. That one works better with AMD GPUs
    2
  2627. not yet
    2
  2628. Appreciate that
    2
  2629. very insightful - thanks for sharing!
    2
  2630.  @kaslanaworld4746  Since I do not have an AMD Graphics card, I am not sure what it is supposed to look like. Try looking on the Github page for any information regarding AMD Graphics cards
    2
  2631. Thanks!
    2
  2632. have fun!
    2
  2633. thanks a lot!
    2
  2634. cool idea!
    2
  2635. u can! try this: https://youtu.be/jh83ViDUvJs or this https://youtu.be/TSLM1mEAc18
    2
  2636. do u mean in realtime?
    2
  2637. I can't wait!
    2
  2638. It could be a variety of reasons. Could you provide more details? Try searching the comments here, or in the discord. Probably someone else has a similar issue as well
    2
  2639. hey there! sorry I have pretty much no experience w linux but i'll ask some tech friends and see if they can show me
    2
  2640. 🦜
    2
  2641. microsoft has one already (https://youtu.be/0s5J2LRqQAI). just need to merge it with this
    2
  2642. One on the app and the other on the browser
    2
  2643. it's here!!!!!!
    2
  2644. How to Change Your Voice in Realtime on Discord (& Games) with this FREE AI https://youtu.be/XHGlv-E7eBw
    2
  2645. It really is!
    2
  2646. Yeah it requires a decent computer and even then the delay will be pretty significant. Here's to hoping the technology gets more efficient soon!
    2
  2647. Thanks!
    2
  2648. no
    2
  2649.  @cortezforever  Thanks! It's still very inconsistent and I find that only 10-20% of the generations are actually decent
    2
  2650. 😃
    2
  2651. It's already available to users in China. People are sharing a ton of examples on twitter and it's totally legit
    2
  2652. 😃😃😃
    2
  2653. ily2
    2
  2654. 😃
    2
  2655. good point!
    2
  2656. Thanks homie 😁
    2
  2657. It does but the delay will probably be too much since theres no dedicated gpu
    2
  2658. That'd be cool. A copilot for music composition!
    2
  2659. lol seems possible given the current sentiment
    2
  2660. yes, but it'll be slower
    2
  2661. thanks!
    2
  2662. indeed!
    2
  2663. Thank you!
    2
  2664. it looks like he'll release it after paper submission https://github.com/xh9998/DiffVSR-project/issues/1
    2
  2665. Thanks. I'll try to find something. I know of agorapulse for scheduling posts, but its not a AI tool
    2
  2666. Thanks!
    2
  2667. Check the logs to see if it says anything! You can also try reinstalling the program, maybe some files got corrupted?
    2
  2668. yeah, though it might be possible to route the output to an open source voice changer with Scarlett's voice
    2
  2669. I use an SHURE SM7B
    2
  2670. Thx for the sub!
    2
  2671. lol
    2
  2672. yeah, that's one limitation. i'd like to feed it my entire project
    2
  2673. you're using the online interface. the local version is uncensored
    2
  2674. it could be a variety of reasons. - make it a shortcut - extract it first - try searching the comments here
    2
  2675. i hope that's the plan!
    2
  2676. You are welcome!
    2
  2677. To do the voice conversion, it basically runs your voice through a neural network, and GPUs are more optimized for running neural networks.
    2
  2678. Thank you! Cheers!
    2
  2679. there's a tool that can do that. making a tutorial on it now!
    2
  2680. Thanks. it was a good learning experience. any live actions i make later will be a lot faster!
    2
  2681. Thanks for sharing
    2
  2682. yes!
    2
  2683. Try to extract everything from the zip into a folder and then open it there
    2
  2684. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    2
  2685. enter conda init
    2
  2686. Good to know!
    2
  2687. ghost in the shell
    2
  2688. yes, rtx + ti is great!
    2
  2689. lol thanks
    2
  2690. you're very welcome
    2
  2691. Thanks!
    2
  2692. 😃
    2
  2693. hmmm do u see any error message after training
    2
  2694. 🔥🔥🔥
    2
  2695. nice!
    2
  2696. Thank you! Glad it helped!
    2
  2697. Not so much the mic quality, but ensure that your voice is clear and your recording without much background noise or echo. The model you choose could also be a factor. Some models are not trained well and do not convert well
    2
  2698. they are very similar. i wonder if its the same algo or training data
    2
  2699. thats longer than i can last. jk
    2
  2700. it says 6gb. not sure if you can get away with lower
    2
  2701. You're welcome!
    2
  2702. oh good point. i cant find a way either
    2
  2703. MIND BLOWN!
    2
  2704. he does move & talk like a robot
    2
  2705. 😃😃
    2
  2706. good luck!
    2
  2707. unsupervised means without guidance from us. it was fed all these articles in stats & cryptanalysis and learned to uncover hidden structures or patterns in the data by itself.
    2
  2708. labs can apply for early access
    2
  2709. that's why this channel is faceless 😉
    2
  2710. thanks for sharing! hope we get an open source model soon so that anyone can use it
    2
  2711. I am working on a video as we speak!
    2
  2712. I feel the same! There's just so much going on
    2
  2713. There's one line in their github (https://github.com/w-okada/voice-changer/blob/master/README_en.md) that mentions Linux, so I guess it would work. "Cross-platform compatibility Supports Windows, Mac (including Apple Silicon M1), Linux, and Google Colaboratory."
    2
  2714. 😢
    2
  2715. The laughing and giggling are borderline psychotic sounding. <<< lol
    2
  2716. Thank you!
    2
  2717. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    2
  2718. Yes as long as you set the output correctly it should work on other apps
    2
  2719. yes! but it's best to use a model with the same language or accent
    2
  2720. Thank you!
    2
  2721. liveportrait is pretty good https://youtu.be/uyjSTAOY7yI
    2
  2722. yes see this https://youtu.be/ixB9oalT3cQ
    2
  2723. Looks like you are! Thank you!
    2
  2724. exactly!
    2
  2725. found it! https://www.udio.com/songs/8ACm7odbDDyuzEHyfJDexC
    2
  2726. You will need to add more code. See documentation: https://huggingface.co/docs/diffusers/using-diffusers/controlnet https://huggingface.co/docs/diffusers/using-diffusers/img2img https://huggingface.co/docs/diffusers/using-diffusers/inpaint
    2
  2727. yes
    2
  2728. code isn't out yet, but you can try mimicmotion for now
    2
  2729. does it sound raspy?
    2
  2730.  @SOMELOAF  ah that's too low. u need 12G at least for now
    2
  2731. My pleasure!
    2
  2732. i was using windows in the video
    2
  2733. glad you got it to work!
    2
  2734. 🌽🌽🌽
    2
  2735. Reposting tomorrow!
    2
  2736. yes!
    2
  2737. yeah, im sure apple/google are secretly building something like this right now
    2
  2738. Thanks!
    2
  2739. why free - idk... It's just an open source project on github yes it'll work without Internet they won't download anything from you. you run it locally via your gpu
    2
  2740. 😃😃
    2
  2741. thanks for sharing! it's indeed tricky
    2
  2742. yeah i said in the video 12gb is lowest for now. there will probably be a 8g version soon
    2
  2743. Thanks. i didn't use rvc for this
    2
  2744. the quality is not good enough. i'll talk about it next video
    2
  2745. i hope so. but it's concerning that no one at openai denied this
    2
  2746. it's best to find a german voice model
    2
  2747. 😃
    2
  2748. I'm only comparing free alternatives, since this is free
    2
  2749. yes. see ahem civitai ahem
    2
  2750. I try to find things in my social media feed, reddit groups
    2
  2751. they just added a hf space. literally just now: https://huggingface.co/spaces/Pyramid-Flow/pyramid-flow
    2
  2752. Glad it was helpful!
    2
  2753. 😃
    2
  2754. yes. its also available for free, but with lower limits
    2
  2755. it will slow down your pc while you are inferring, since it uses your cpu
    2
  2756. 😂
    2
  2757. Yes it's generally a hardware problem
    2
  2758. soon, i hope
    2
  2759. might be a dopamine boost to some people i guess
    2
  2760.  @CristianMunoz-go9tf  put the .pth file and .index file in a zip file, and upload that to google drive or hugging face. then, paste that link in the 'download model' section at 4:52
    2
  2761. good idea
    2
  2762. 😃😃😃
    2
  2763. seems like it
    2
  2764. great idea!
    2
  2765. anyone can impersonate anyone else 😃
    2
  2766. not yet. i hope they will add a feature for consistent chars in the future
    2
  2767. This video might have overloaded their servers lol. Try again later
    2
  2768. you're welcome!
    2
  2769. they will release a hf space soon
    2
  2770. possibly but you'll get a lot of delay
    2
  2771. 😃😃😃
    2
  2772. Thanks!
    2
  2773. You're welcome!
    2
  2774. Thank you very much!
    2
  2775. Thanks!
    2
  2776. soon anyone can manipulate a video however they want
    2
  2777. Thanks for sharing!
    2
  2778. from their page: In a cozy kitchen, a golden retriever wearing a white chef's hat and a blue apron stands at the table, holding a sharp kitchen knife and skillfully slicing fresh tomatoes. Its tail sways gently, and its gaze is focused and gentle. There are already several neatly arranged tomatoes on the wooden chopping board in front of me. The kitchen has soft lighting, with various kitchen utensils hanging on the walls and several pots of green plants placed on the windowsill.
    2
  2779. I'm not aware of an AI that can do this yet
    2
  2780. colab is cracking down on people using this for voice cloning, so unfortunately you might disconnect often. It's best if you can install this locally
    2
  2781. thanks for not using PS. what a ripoff! try https://www.photopea.com/ For firefly-like functions, try Krita https://github.com/Acly/krita-ai-diffusion?tab=readme-ov-file
    2
  2782. lol I've been told that before
    2
  2783. if it's frontend only (eg. if everything can function within 1 html file), you can simply download the code as I showed in the video, then host it with any provider like godaddy. if you need backend stuff (eg. pulling data from a server) this won't work yet - you'll need to code this yourself
    2
  2784. Enjoy!
    2
  2785. true. there's no such thing as unbiased
    2
  2786. i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    2
  2787. yes, the eyes in these examples seem slightly scary & unnatural
    2
  2788. usually the company's shareholders will decide on who gets to be on the board
    2
  2789. No but you might get a copyright claim which means you can't monetize it
    2
  2790. try this https://ai-search.io/voices/markiplier
    2
  2791. colab isn't possible, as it'll disconnect any web interfaces
    2
  2792. thanks! cuda gpu with 8g vram
    2
  2793. no mobile version yet unfortunately
    2
  2794. Thanks for sharing!
    2
  2795. I just tried and you're right! For NSFW, you best bet is to use A1111. There are no rules 😉
    2
  2796. Load the gradio link first. After you get A1111 running, drag your lora to the Lora folder. It should work without needing to reload the cell
    2
  2797. https://preview.redd.it/qcrcm1cpitka1.png?width=960&crop=smart&auto=webp&s=62ba541954182df1e04e2a39c5309074d76c665a
    2
  2798. yes, using a fp8 model would be your best option
    2
  2799. 😃
    2
  2800. u can do multiple. see this https://youtu.be/KW9s_im2338
    2
  2801. You're welcome!
    2
  2802. lol
    2
  2803. Thanks!
    2
  2804. My pleasure 😊
    2
  2805. yes, a lot. try this https://www.bing.com/images/create
    2
  2806.  @krin_c747  https://poe.com/explore?category=Image+generation
    2
  2807. oh if you want more control, try this: https://youtu.be/8_V8CO_Dbdw https://youtu.be/hwsvcbFeUTs
    2
  2808. good spot
    2
  2809. You are welcome!
    2
  2810. Glad you were able to solve it
    2
  2811. yes it seems like OS will become IA in the near future
    2
  2812. Thank you! Cheers!
    2
  2813. Thanks!
    2
  2814. 😃
    2
  2815. thanks!
    2
  2816. 😃
    2
  2817. I LOVE JUMPING SPIDERS! THEY ARE SOOOO CUTE!!!!
    2
  2818. thanks!
    2
  2819. Yes! Please see my latest video on how to use it on discord
    2
  2820. yeah, generally RTX >3000 cards work the best for AI
    2
  2821. it should. he mentioned it takes a 4G sim card
    2
  2822. ai: ....i give up
    2
  2823. You're welcome!
    2
  2824. just posted! i had to wait 3h for a generation this morning...
    2
  2825. no problem
    2
  2826. Thanks. I don't have any plans at the moment for a detailed video on the math, but I'll take a look at Principia Qualia first. Thanks for suggesting!
    2
  2827. Thanks!
    2
  2828. Looks like you are! Thank you!
    2
  2829. yes, this one is great for prompt following
    2
  2830. 😃
    2
  2831. please check the terms & conditions of each platform
    2
  2832. Yeah it's an unfortunate reality
    2
  2833. yes! i'll mention it in my next vid
    2
  2834. Thanks!
    2
  2835. this will be my next video! uploading this week, hopefully
    2
  2836. You're welcome!
    2
  2837. Just posted: https://youtu.be/LxOOj-mkQV8
    2
  2838. these features r not out yet
    2
  2839. wow! thanks for sharing
    2
  2840. 👍
    2
  2841. yeah, so many video generators announced!
    2
  2842. it'll be a great ROI if it can do things 24/7
    2
  2843. 4090 is def better than 5000 for AI stuff
    2
  2844. Welcome
    2
  2845. lol
    2
  2846. Thanks!
    2
  2847. thanks! (nice name btw)
    2
  2848. you get free dailly credits
    2
  2849. 😃
    2
  2850. nice!
    2
  2851. Couldn't agree more!
    2
  2852. We are working on a video for that as we speak!
    2
  2853. im making a video now. will publish next week!
    2
  2854. it depends on your gpu, but for me, less than 5min
    2
  2855. Try increasing the chunks!
    2
  2856. Thanks!
    2
  2857. Thanks for the sub!
    2
  2858. That's true. All the AI tools still produce bland and flat tracks. I'd say soundful is the best quality so far, and meta is quite good for sound effects, but still nowhere close to what a human music producer can make
    2
  2859. Thanks!
    2
  2860. yes
    2
  2861. Thanks for watching!
    2
  2862. 😃
    2
  2863. sounds like onlyfans
    2
  2864. apple: "shhh, don't expose our secret strategy"
    2
  2865. You're welcome!
    2
  2866. Yes, it's the best I've seen so far
    2
  2867. with this, I assume each user will have their own personalized AI
    2
  2868. Women can do the same - I'm sure there's a perfect male voice available too. In the end, humans will just stop talking to each other.
    2
  2869. no
    2
  2870. Haha, thanks man!
    2
  2871. you're welcome!
    2
  2872. Thanks, Supreme Leader!
    2
  2873. Thanks for watching!
    2
  2874. im sure he has a lot going on
    2
  2875. so far, no malware has been reported by users, but see this https://youtu.be/5NIjUVzxV9k
    2
  2876. have fun!
    2
  2877. thanks for sharing!
    2
  2878. 1080 will work!
    2
  2879. first time i heard zuck being called an angel, but ok
    2
  2880. 👀
    2
  2881. thank you so much!
    2
  2882. It seems to be a colab problem, and many people have reported this issue recently: https://github.com/huggingface/diffusers/issues/6431 Best we can do is wait for a solution
    2
  2883. looks like they released the models on huggingface
    2
  2884. Shure sm7b, but it's pricey! A cheaper option could be the logi blue sona
    2
  2885. Thanks for sharing!
    2
  2886. lol it sure feels like it
    2
  2887. Thanks for sharing!
    2
  2888. Thanks!
    2
  2889. yeah, Imagen 3 is underrated & free!
    2
  2890. you're welcome
    2
  2891. thanks for the sub!
    2
  2892. yes! hope this vid raises more awareness. it looks like the models are still on hf
    2
  2893. lol thanks
    2
  2894. Thanks for sharing! We only have Miles for now, but I'm sure they'll add more voices in the future
    2
  2895. yo
    2
  2896. Thanks for sharing!
    2
  2897. flux requires a slightly different workflow. see the latter half of this https://youtu.be/K3xJ7GQuHpw
    2
  2898. as long as it has cuda support
    2
  2899. me too
    2
  2900. Thank you!
    2
  2901. 😃
    2
  2902. use this to find voice models: https://youtu.be/98drFL65VH0
    2
  2903. A1111 won't work in colab anymore. I'll make a new video next on how to STILL use stable diffusion in colab w/o disconnecting. will publish in a few days!
    2
  2904. 😧
    2
  2905. looking forward to the waifu
    2
  2906. yes see this https://youtu.be/K3xJ7GQuHpw
    2
  2907. it's probably due to the source. from their page, it seems like you can do up to 20x interpolation. doesn't hurt to give it a try!
    2
  2908. Thank you!
    2
  2909. Thanks!
    2
  2910. try this: https://youtu.be/jh83ViDUvJs or this https://youtu.be/TSLM1mEAc18
    2
  2911. did you try extend -> add outro?
    2
  2912. 🤙
    2
  2913. 😃
    2
  2914. https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
    2
  2915. ...and they still reply, you know it's AI
    2
  2916. You are welcome!
    2
  2917. unfortunately instantid doesnt work w flux for now
    2
  2918. i used cakewalk. quite similar to fl studio
    2
  2919. 😃😃😃
    2
  2920. yes!
    2
  2921. oh, what was in it?
    2
  2922.  @BazzilioSPb  correct!
    2
  2923. probably a shiton unfortunately
    2
  2924. thanks for sharing! it's alright if you're not paying your electricity bills 😉
    2
  2925. Google has an agent coming soon!
    2
  2926. hi!
    2
  2927. No problem 👍
    2
  2928. yeah, i'd vote for them!
    2
  2929. not yet. it's closed testing for now
    2
  2930. Thanks!
    2
  2931. 😃
    2
  2932. it wont run. you'll get an out of memory error
    2
  2933. 🥲
    2
  2934. do you have a cuda gpu?
    2
  2935. truth
    2
  2936. Thanks for sharing!
    2
  2937. just delete the folder
    2
  2938. lol i get that a lot
    2
  2939. sure lets compare the stock price if you want. AAPL is up 11.59% YTD. GOOG is up 27.83% YTD. even the market has figured it out
    2
  2940. Thanks! I'm playing around with animediff but it might be too complicated for our viewers. have you tried these other tools: https://youtu.be/tDidofPTwCc
    2
  2941. Thanks! Have you installed python? Then you'll need to run 'pip install pandas' in your command prompt. Then the script should work. Let me know if you get stuck anywhere
    2
  2942. yes im making a video now
    2
  2943. Check out Trellis, featured in my last news video.
    2
  2944. 😃
    2
  2945. no need to download. just use in your internet browser
    2
  2946. you'll get disconnected quickly if you use the colab free plan for this. its best to install this locally
    2
  2947. yes
    2
  2948. Have fun!
    2
  2949. great idea! you can already do this with the share screen feature
    2
  2950. True. here's an interesting hallucination leaderboard https://github.com/vectara/hallucination-leaderboard?tab=readme-ov-file
    2
  2951. its still in dev and there's no timeline yet. when i get more info, ill announce it on my channel
    2
  2952. You're welcome!
    2
  2953. ai gf 😃
    2
  2954. see this https://youtu.be/XHGlv-E7eBw
    2
  2955. it was later revealed that mystic was just an upscaled wrapper of flux, which was disappointing
    2
  2956. lol
    2
  2957. noted, thanks for the pointer!
    2
  2958. catgirls ftw
    2
  2959. another user claims that 12GB works
    2
  2960. @ezarago1 i borrowed my friend's 4090 24GB for this. my old 2080 couldn't handle it
    2
  2961. lol
    2
  2962. definitely!
    2
  2963. Just posted: https://youtu.be/LxOOj-mkQV8
    2
  2964. That's correct, but usually 5-10min is good enough
    2
  2965. ah i faintly remember that ball. pre-touchscreen era
    2
  2966. 🤯
    2
  2967. they will release the video feature 'in a few days' https://github.com/KwaiVGI/LivePortrait/issues/27
    2
  2968. hi!
    2
  2969. no, its a separate conda environment
    2
  2970. image: still SD (A1111 & Comfy). video: i'd recommend pika. openai sora was just announced, which will crush everything https://youtu.be/jZ622-8llow
    2
  2971. it has not been added to the UI yet. will keep you posted!
    2
  2972. Yeah, very similar
    2
  2973. yes, i don't see why not. if you haven't heard of websim yet, check this out: https://youtu.be/a4nXGnumD1U it might be related to what you're looking for
    2
  2974. Youre welcome!
    2
  2975. not for imagen. your best bet is to use faceswap, which is available for SD. i assume it'll be out for flux soon as well
    2
  2976. You're welcome!
    2
  2977. nope
    2
  2978. Oh, that's strange. thanks for sharing!
    2
  2979. This will def be the future!
    2
  2980. You're welcome!
    2
  2981. My pleasure!
    2
  2982. yeah, it was a very smart move by openai to announce it first, so that we are less 'impressed' by google's announcement of astra
    2
  2983. 5-10minutes is good enough
    2
  2984. pun intended?
    2
  2985. Thanks. For a 10s video, it took maybe 1-2min to generate. Very quick compared to other tools
    2
  2986. hi
    2
  2987. hong kong
    2
  2988. i think it's only instruments for now
    2
  2989. thanks!
    2
  2990. 🤑🤑🤑
    2
  2991. Thank you!
    2
  2992. yep
    2
  2993. Nice!
    2
  2994. add a new cell block, then paste this in: import gc torch.cuda.empty_cache() gc.collect() then run the cell to clear cache. then you should be able to rerun the pipeline a bit more. Note that there are also daily limits for the colab free plan.
    2
  2995. pls let me know of a better way. i've tried ebsynth, temporal kit, sd-cn... they were all inferior & needed lots of steps
    2
  2996. do this for each voice, then manually glue each voice together https://youtu.be/o6ywe3EDLDM. you need to convert each voice separately first
    2
  2997. Thanks!
    2
  2998. You're welcome!
    2
  2999. You're welcome!
    2
  3000. 32g is just the model size. vram is used for calculations while generating the image - it doesn't need the full 32g at once
    2
  3001. my pleasure!
    2
  3002. Thanks for sharing!
    2
  3003. try this, but it requires more tweaking: https://github.com/KwaiVGI/LivePortrait/issues/41
    2
  3004. this will happen sooner or later. even if Scott doesn't build it, someone else will
    2
  3005.  @Dfd_Free_Speech  got it! yeah, poe is more for finding things out and not meant for chatting with a character or roleplay
    2
  3006. No problem!
    2
  3007. i predict within 6 months
    2
  3008. Thanks for sharing!
    2
  3009. It's likely to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    2
  3010. Thanks!
    2
  3011. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    2
  3012. unfortunately thats a flaw w this. you'll need to smooth it out via another tool
    2
  3013. Thanks! I predict we'll start seeing robots everywhere by next year
    2
  3014. yes, but they found no improvement in health or decrease in stress over the 3 years 🤔
    2
  3015. 😃
    2
  3016. You're welcome!
    2
  3017. CPU works, but you might have delay unless you have a really nice CPU
    2
  3018. You're welcome!
    2
  3019.  @aibi5532  there's no website. i just searched google and X
    2
  3020. thanks!
    2
  3021. 🤯 i always thought it was black stripes on white for some reason
    2
  3022. mind=blown
    2
  3023. Thanks for the super! I appreciate your support
    2
  3024. be nice to them now, so that you won't be on the list during their uprising 😉
    2
  3025. I've seen other users sharing that it can answer this. Perhaps I just got unlucky
    2
  3026. Well, Steve Jobs couldn't code
    2
  3027. Thanks for sharing! I'll check it out
    2
  3028. Unfortunately there's no clear documentation on this
    2
  3029. You might be able to use some of these on your phone's browser
    2
  3030. You can use Seaart & Yodayo to make art in your mobile browser. Let me know if it works for you!
    2
  3031. 😃
    2
  3032. Thanks for the tip!
    2
  3033. Runway is indeed quite frustrating. I'd say ~10% of generations actually turn out usable.
    2
  3034. No problem!!
    1
  3035. u can use audacity
    1
  3036. yes https://youtu.be/_JXbvSTGPoo
    1
  3037. Welcome!
    1
  3038. here you go https://youtu.be/KW9s_im2338
    1
  3039. yes!
    1
  3040. it now works for lower vram. ill do a tutorial soon
    1
  3041. You're welcome!
    1
  3042. text is available now. voice chat is coming in the next few weeks
    1
  3043. 😃
    1
  3044. Nice! Thanks for the dos tip. I'll try it out
    1
  3045. it's not 100% success rate unfortunately. also try [spoken] or [speech] or [talking]
    1
  3046. talk about hyper personalized content
    1
  3047. I used w-okada. See this vid: https://youtu.be/62sjaCXyOsU
    1
  3048. No problem 👍
    1
  3049. yes!
    1
  3050. Thank you!
    1
  3051. You're welcome!
    1
  3052. You're welcome!
    1
  3053. Thank you! Glad you liked it
    1
  3054. looks like you haven't added it to path yet
    1
  3055. Thanks, will do!
    1
  3056. No this will not give you a virus
    1
  3057. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  3058. u mean for video?
    1
  3059. AI video is not great yet, but you can try these free options https://youtu.be/tDidofPTwCc
    1
  3060. try the easeus one then. it's also free!
    1
  3061. no problem
    1
  3062. Thanks!
    1
  3063. No but u can use this for tts: https://youtu.be/jh83ViDUvJs
    1
  3064. lol
    1
  3065. they trained an AI to translate brain signals to cursor outputs
    1
  3066. 😃
    1
  3067. No problem!
    1
  3068. yes! btw, meta x rayban already has one. not sure how good it is though. it's AI functions are only limited to the US right now
    1
  3069. just saw the vid last week. i now have mixed opinions about jesse. let's see how they get themselves outta this.
    1
  3070. try right click > open with > powershell
    1
  3071. No problem 👍
    1
  3072. Thanks
    1
  3073. 😃
    1
  3074. you can. did you see this: https://youtu.be/_JXbvSTGPoo
    1
  3075. I'm sure this will be heavily abused in politics
    1
  3076. they claim it's free for early users. unfortunately you need to somehow join their waitlist and get accepted
    1
  3077. lol
    1
  3078. thanks for the correction!
    1
  3079. yes u can try this https://youtu.be/0uyGj7Fel58 but it might get disconnected often
    1
  3080. You're welcome
    1
  3081. thanks for watching!
    1
  3082. Do you see any error message in the logs
    1
  3083. you're welcome
    1
  3084. Check out this video on how to clone anyone's voice! https://youtu.be/2-9KRkfs1Os
    1
  3085. 😃
    1
  3086. you're welcome!
    1
  3087. 😃
    1
  3088. 😃
    1
  3089. 😏😏😏
    1
  3090. Thanks!
    1
  3091. Thanks!
    1
  3092. yes you can do this for free
    1
  3093. Interesting, thanks for sharing!
    1
  3094. the non-commercial licence is indeed very restrictive
    1
  3095. things are about to get wild!
    1
  3096. min requirement is a cuda gpu with 8gb vram. if it doesnt work, try this https://youtu.be/tnfqIQ11Qek
    1
  3097. it seems u don't have tensorboard yet. try pip install tensorboard in your cmd (I assume u already have pip & python installed?)
    1
  3098. 😃
    1
  3099. can you elaborate?
    1
  3100. exponential growth!!!
    1
  3101. Unfortunately there's no clear documentation on this
    1
  3102. yes, see the training part here https://youtu.be/ixB9oalT3cQ just upload recordings of yourself
    1
  3103. interesting. thanks for sharing!
    1
  3104. thanks!
    1
  3105. it could be a variety of reasons. - try the start_https one - make it a shortcut - extract it first - try searching the comments here
    1
  3106. you're not missing out on much. meta AI (llama) isn't great, and movie gen isn't out yet.
    1
  3107. 🐱
    1
  3108. puppets are slightly creepy i must say
    1
  3109. just a cleaner interface & workflow
    1
  3110. Thanks!
    1
  3111. 🥲
    1
  3112. You're welcome!
    1
  3113. Thanks
    1
  3114. No problem
    1
  3115. yes, if you saved the G and D and index files, u can continue training
    1
  3116. Not well
    1
  3117. 😃
    1
  3118. not at the moment
    1
  3119. lol
    1
  3120. it can kinda do that if you use [riff repeats] or [verse repeats] but it's hit or miss
    1
  3121. Thanks for sharing!
    1
  3122. 😃
    1
  3123. try liveportrait: https://youtu.be/uyjSTAOY7yI or hallo: https://youtu.be/rlnjcRP4oVc
    1
  3124. Watch our tutorial video
    1
  3125. https://github.com/w-okada/voice-changer/blob/master/README_en.md
    1
  3126. how can you tell it's CGI?
    1
  3127. 😃
    1
  3128. not sure. tagging @nunudoai
    1
  3129. nice!
    1
  3130. yes, i'm planning that right now!
    1
  3131. RVC: https://youtu.be/ixB9oalT3cQ
    1
  3132. theres currently no good realtime hair swapper
    1
  3133. Thanks for sharing!
    1
  3134. yes!
    1
  3135. exactly! great idea
    1
  3136. coming soon!
    1
  3137. you're using the online interface. the local version is uncensored
    1
  3138. Thanks for suggesting. I'll look him up
    1
  3139. you don't have 30 free generations???
    1
  3140. I would say it would have a hard time working with anything lower than a GeForce 10 series.
    1
  3141. you're welcome!
    1
  3142. thanks for the sub!
    1
  3143. yes see here https://youtu.be/8uq0e_i17FE
    1
  3144. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  3145. Thank you so much!
    1
  3146. try reducing the feature ratio or increasing the median filtering (bottom left)
    1
  3147. Thanks!
    1
  3148. Kits is a good paid platform
    1
  3149. Thanks!
    1
  3150. it understands reality!
    1
  3151. 😃
    1
  3152. thx for sharing. could you share screen and then go to youtube?
    1
  3153. 🤯
    1
  3154. Oh that's great, thanks for sharing!
    1
  3155. did u unzip everything into a folder first
    1
  3156. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  3157. it's hit or miss at times
    1
  3158. i hope an open source version comes out soon, so i can make unlimited generations locally
    1
  3159. unfortunately, none out of those 3. 3060ti is ok, but it has only 12G vram. it's best to have at least 16G vram
    1
  3160. see this https://youtu.be/KW9s_im2338
    1
  3161. Good spot!
    1
  3162. ok, then will u marry me?
    1
  3163. 👀
    1
  3164. i predict a few months latest
    1
  3165. That sounds possible
    1
  3166. No problem!
    1
  3167. hi
    1
  3168. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  3169. Yes this is safe!
    1
  3170. no problem
    1
  3171. pls explain
    1
  3172. Got it, thanks for sharing!
    1
  3173. Do you see any error message?
    1
  3174. 😃
    1
  3175. it's banned on colab (free plan) 😢 u can use the paid plan tho. see this https://youtu.be/9TroP5mR3CM
    1
  3176. I saw another comment here that said the same thing!
    1
  3177. nice eyes there
    1
  3178. 😃
    1
  3179. indeed!
    1
  3180. depends on the animal. most dogs & cats work fine
    1
  3181. Thank you so much!
    1
  3182. hope u get it to work!
    1
  3183. it's unavoidable. the tech giants have all the compute
    1
  3184. my pleasure
    1
  3185. Thanks!
    1
  3186. What did you write? It might have been auto-deleted by the algo for whatever reason
    1
  3187. you can use an AI agent to automate workflows
    1
  3188. yes see https://youtu.be/_JXbvSTGPoo
    1
  3189. I just checked out Alltalk. Looks like its not compatible w RVC models?
    1
  3190. yes!
    1
  3191. Thanks!
    1
  3192. yes!
    1
  3193. try replacing 0.0.0.0 with 127.0.0.1
    1
  3194. try to clone the repo again or download the ps1 directly from the github page
    1
  3195. are you able to create an account?
    1
  3196. It looks like you don't have python installed yet https://python.org/downloads/
    1
  3197. You just need internet for all of them except the last one. A1111 can run with AMD (but nvidia is best)
    1
  3198. Thanks. Yes, coming soon
    1
  3199. 😃😃😃
    1
  3200. Try lowering the chunks
    1
  3201. thx for sharing!
    1
  3202. you should have it now or soon, even for the free plan. see end of this video for more details: https://youtu.be/_DgkGAaJwYs
    1
  3203. 😧
    1
  3204. You're welcome!
    1
  3205. indeed!
    1
  3206. 😃
    1
  3207. theres no good mobile option yet
    1
  3208. good luck!
    1
  3209. True! Starting from 0 is always the hardest. Once you build momentum, it will grow like a snowball.
    1
  3210. what step are u on? do u see more errors?
    1
  3211. yes 4gb is too low. min recommendation is 8gb
    1
  3212. thanks!
    1
  3213. dark but true. why date a human when you can have a perfect companion in a perfect silicone body
    1
  3214. nice!
    1
  3215. there is. see this https://youtu.be/jh83ViDUvJs
    1
  3216. Welcome!
    1
  3217. make sure you've added conda to path. also, try restarting your pc
    1
  3218. You're very welcome!
    1
  3219. interesting! thanks for sharing
    1
  3220. 🤯
    1
  3221. try this: https://youtu.be/hwsvcbFeUTs
    1
  3222. That's great! Thanks for sharing
    1
  3223. no, it currently only detects gaze direction. but it'd be nice if there was an AI that can control gaze!
    1
  3224. 😃
    1
  3225. GREAT IDEA!
    1
  3226. good observations
    1
  3227. Thanks for sharing!
    1
  3228. 💩💩💩
    1
  3229. thanks. here's the full version: https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
    1
  3230. here you go https://youtu.be/PCL9SAlHqzw
    1
  3231. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  3232. Thanks!
    1
  3233. 😉
    1
  3234. this is better https://youtu.be/ixB9oalT3cQ
    1
  3235. Thanks!
    1
  3236. im looking into alternatives and will update everyone asap!
    1
  3237. Video coming out tomorrow!
    1
  3238. great idea!
    1
  3239. Welcome!
    1
  3240. i hope so too! i really love SD
    1
  3241. Thanks!
    1
  3242. $$$$$
    1
  3243. Thanks
    1
  3244. i think that's the success message?
    1
  3245. I am working on a video right now!
    1
  3246. singing is still the same
    1
  3247. i heard this recent release was just 'beta', so i hope they have something better being planned
    1
  3248. here you go https://youtu.be/KW9s_im2338
    1
  3249. 😃
    1
  3250. good spot
    1
  3251. Thank you so much!
    1
  3252. 😃
    1
  3253. Unfortunately you would have to have a very powerful laptop to run this program. Hopefully it becomes easier to run soon!
    1
  3254. lol
    1
  3255. yep
    1
  3256. You're welcome!
    1
  3257. you're welcome!
    1
  3258. seaart + yodayo should work. not sure about the rest
    1
  3259. Thanks!
    1
  3260. i'm comparing free (dev) with free (SD). see this vid for a flux pro comparison with other paid tools https://youtu.be/9Ja890WM-eM
    1
  3261. Thanks!
    1
  3262. put the .py file in the same folder as the csv, and then run the .py file with python
    1
  3263. Not sure! You can give it a try!
    1
  3264. coming soon!
    1
  3265. your gpu should be fine. Make sure you don't have any spaces in your folder or file names
    1
  3266. What's the error message you got?
    1
  3267. thanks for sharing
    1
  3268. Thanks for the sub!
    1
  3269. Thank you!
    1
  3270. no u can use cpu, but performance is better with nvidia gpu
    1
  3271. lol
    1
  3272. exactly!
    1
  3273. Thanks
    1
  3274. exponential growth!
    1
  3275. What error do you see
    1
  3276. 😃
    1
  3277. Yes, you just need to input recordings of your friend. see the last part in this tutorial https://youtu.be/ixB9oalT3cQ
    1
  3278. Yes, they're all using SD and based on the same models (checkpoints)
    1
  3279. that message means it's complete
    1
  3280. yes. we need an AI hand swapper too
    1
  3281. thanks for the pointer!
    1
  3282. You're welcome!
    1
  3283. can you paste the error message here
    1
  3284. they have too much $
    1
  3285. nope, no vision capabilities yet. i was looking forward to testing that out, like helping with hw, IDing a plant or analyzing a stock chart.
    1
  3286. yes, it's surprisingly similar to kling & minimax
    1
  3287. u can import it to okada https://youtu.be/_JXbvSTGPoo
    1
  3288. 😃
    1
  3289. If youre still using it, you can disable it at startup by going to task manager (windows), and find it in the startup tab
    1
  3290. thanks!
    1
  3291. They have mac support, but I have seen a lot of people having issues with it.
    1
  3292. thanks for sharing!
    1
  3293. Try using crepe and increase the chunks
    1
  3294. What system specs do you have?
    1
  3295. Maybe try reinstalling it. I was having some issues similar to that and reinstalling it fixed mine.
    1
  3296. 😃😃😃
    1
  3297. here's the udio link if that works: https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
    1
  3298. this, but it's not great https://github.com/PKU-YuanGroup/Open-Sora-Plan
    1
  3299. Thank you!
    1
  3300. 🤣
    1
  3301. Yes, depending on the power of the graphics card, the delay will be greater or worse.
    1
  3302. exactly. both movie gen and sora are just 'vaporware' for now
    1
  3303. I've seen other users sharing that it can answer this. Perhaps I just got unlucky
    1
  3304. Thanks for the sub!
    1
  3305. this is free and has background removal: https://www.photopea.com/
    1
  3306. sit back and wait for UBI
    1
  3307. yes
    1
  3308. My pleasure!
    1
  3309. if u have pro, u can just use A1111
    1
  3310. thanks! yes, i did a video last week.
    1
  3311. see this https://youtu.be/KW9s_im2338
    1
  3312. Haha I didn't use any AI generated script, although that would be a good idea 🤔
    1
  3313. Thanks for the super!
    1
  3314. I will give it a try!
    1
  3315. 😍
    1
  3316. its better to use this https://youtu.be/ixB9oalT3cQ
    1
  3317. lol
    1
  3318. yes
    1
  3319. You're welcome!
    1
  3320. Thanks! Will do!
    1
  3321. They are nearing a deal. See my most recent video
    1
  3322. you're welcome!
    1
  3323. No worries!
    1
  3324. My pleasure!
    1
  3325. What hardware do you have?
    1
  3326. It will usually lag if you don't have a good NVIDIA GPU
    1
  3327. You're welcome!
    1
  3328. agreed!
    1
  3329. Glad to hear that!
    1
  3330. Thank you!
    1
  3331. i believe they have a mac install on their github. unfortunately, I don't use mac so I can't advise any further
    1
  3332. it's indeed possible that the question (or a variant) was used in training
    1
  3333. damn my school must be dumb. i never learned this 😅
    1
  3334. for real
    1
  3335. welcome!
    1
  3336. Thanks!
    1
  3337. yeah try reinstalling it
    1
  3338. 🤯
    1
  3339. we might be at this threshold right now. realistic images, music, and soon videos, can be made in just one click
    1
  3340. yes
    1
  3341. thanks for sharing!
    1
  3342. 🍈
    1
  3343. yes thats likely. u can only train w a gpu
    1
  3344. it might be too slow, but good luck!
    1
  3345. 😃
    1
  3346. 🤣
    1
  3347. yeah it's an amazing tool!
    1
  3348. 😃
    1
  3349. it says the code is "coming soon"
    1
  3350. interesting idea
    1
  3351. actually its quite common for investors to 'invest' based on the founders, not on the company
    1
  3352. Looking at your other comments, is it because you don't have a gpu?
    1
  3353. do u see any error in the cmd
    1
  3354. exactly!
    1
  3355. I appreciate that! I'll ask if they have more invite codes.
    1
  3356. lol
    1
  3357. thats strange. do you see any other error message
    1
  3358. You're welcome!
    1
  3359. 😃
    1
  3360. what else do u see above 'press any key'
    1
  3361. You're welcome!
    1
  3362. you're welcome!
    1
  3363. No problem!
    1
  3364. are you trying to train a voice
    1
  3365. Great! 2gb would be too low
    1
  3366. Looks like a permission error. Try to open it as administrator
    1
  3367. thx for sharing!
    1
  3368. Try reinstalling it!
    1
  3369. You're welcome!
    1
  3370. are u on the free plan? i'm on standard and it finishes in 3 min
    1
  3371. thanks!
    1
  3372. until there's an AI hand swapper
    1
  3373. the voice feature isn't out yet, but I'll do a tutorial once its out. for regular chat, it's automatically activated in chatgpt, at least for some users.
    1
  3374. thanks for sharing!
    1
  3375. No problem!!
    1
  3376. 🤯
    1
  3377. Check the logs to see if it says anything!
    1
  3378. Thank you!
    1
  3379. here's part 2 https://youtu.be/TSLM1mEAc18
    1
  3380. see my other video!
    1
  3381. My pleasure!
    1
  3382. thanks
    1
  3383. 😃
    1
  3384. nope, its optional
    1
  3385. definitely!
    1
  3386. Welcome 👍
    1
  3387. try these to find voice models: https://youtu.be/98drFL65VH0
    1
  3388. Yes. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  3389. 😃
    1
  3390. Maybe check your firewall settings?
    1
  3391. youre welcome!
    1
  3392. exponential growth!
    1
  3393. 😢
    1
  3394. yes, some of the time
    1
  3395. see the image tools here: https://youtu.be/pLSShbcOl4o
    1
  3396. lol i get a lot of comments saying that
    1
  3397. You are welcome!
    1
  3398. You're welcome 😊
    1
  3399. Only for some things 😂
    1
  3400.  @ouinon6889  hmmm maybe the repo was not cloned properly. try cloning it again, or just downloading the requirements.txt here (https://github.com/KwaiVGI/LivePortrait/blob/main/requirements.txt)
    1
  3401. great idea!
    1
  3402. i suspect sth is broken w the training process. do u see any error message?
    1
  3403. you need to add to path. rewatch my section on installing conda
    1
  3404. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  3405. yeah, it'll give your i5 a hard time if you don't have a dedicated gpu
    1
  3406. they likely used the word 'first' for marketing purposes. similar to 'democratize' or 'revolutionize' - take it with a grain of salt
    1
  3407. thanks for sharing!
    1
  3408. These are all based on stable diffusion, so results might be similar. Also the prompt is very important. try to specify as many details as possible, such as hair length and color, eye color, what she's wearing, pose, background, etc
    1
  3409. Try to extract everything from the zip into a folder and then open it there
    1
  3410. that's why he's building stargate
    1
  3411. it doesn't work for spanish for now. hopefully someone can finetune this for other languages in the future
    1
  3412. you're welcome!
    1
  3413. this will def be used for p*rn
    1
  3414. lol
    1
  3415. Thanks
    1
  3416. u need to upload the pth model within the zip. what program r u using?
    1
  3417. the video function isnt out yet. even the comfy version is just a workaround, but the original author hasn't released anything official yet
    1
  3418. I've seen other users showing that it can answer this. Perhaps I just got unlucky
    1
  3419. it could be, depending on who you talk to. agi has a very blurry definition
    1
  3420. Glad you enjoyed it!
    1
  3421. did you install ComfyUI's ControlNet Auxiliary Preprocessors?
    1
  3422. do u want the infer or infer+train one?
    1
  3423. enjoy!
    1
  3424. Thanks for sharing!
    1
  3425. I'm glad you got it to work!
    1
  3426. Possibly. I hope not....
    1
  3427. it's better at handling large contexts, like if your prompt is super long
    1
  3428. thank you!
    1
  3429. 😃
    1
  3430. yes. if it's a safetensors file, use use_safetensors=True
    1
  3431. for poor af ppl like me 🥲
    1
  3432. great idea!
    1
  3433. You make one
    1
  3434. 🚀🚀🚀
    1
  3435. Search on the discord for it!
    1
  3436. cool! are u trying to build your own app?
    1
  3437. remove the spaces in 'jerome voice '
    1
  3438. you're welcome!
    1
  3439. it's not out yet. still in dev. when i get more info, ill announce it on my channel
    1
  3440. The speed of AI advancement in general is just crazy! Can't image where we'll be at the end of year
    1
  3441. for training, 8GB+ vram is best. for other functions, u dont even need a gpu
    1
  3442. thanks for sharing!
    1
  3443. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    1
  3444. The dilemma is, he can't stop. No one can stop. The 'good guys' need to be significantly ahead of the 'bad guys', so it's a constant race to build the best AI. We can only hope that Sam has good intentions.
    1
  3445. sounds like social media already
    1
  3446. good idea!
    1
  3447. yes!
    1
  3448. it's great and they have a free plan!
    1
  3449. My pleasure!
    1
  3450. plot twist!
    1
  3451. Got it. I’ll bug them for more codes, but they might be very limited
    1
  3452. yep
    1
  3453. You can install RVC locally. Please see my other video on that
    1
  3454. It's probably not possible to train your own model on mobile. Training requires a good GPU, which most phones lack
    1
  3455. this requires roop or reactor which isn't available for diffusers afaik
    1
  3456. Thanks for sharing!
    1
  3457. yes, i foresee this as a tool to empower animators, not replace them
    1
  3458. is it amd?
    1
  3459. the huggingface link should work for free https://huggingface.co/spaces/mrfakename/E2-F5-TTS
    1
  3460. yes, you'll need to separate the vocals before training a model. see https://youtu.be/o6ywe3EDLDM or this https://youtu.be/RvZM381wdmQ for how to separate
    1
  3461. hi
    1
  3462. yes, unless u use a lora of that character
    1
  3463. they didn't say
    1
  3464. i played around a bit more and there are still some noticeable artefacts, but this is def a game changer!
    1
  3465. Thanks!
    1
  3466. it's in the free version too
    1
  3467. you're welcome!
    1
  3468. as long as it's mid-shot, it will still work
    1
  3469. sarcastic, yes. nsfw or violent, no
    1
  3470. let's gooo
    1
  3471. You're welcome!
    1
  3472. Thanks for sharing. I'm also finding inconsistencies
    1
  3473. 😃
    1
  3474. 😃
    1
  3475. It would be a lot slower
    1
  3476.  @osnapproductions1962  colab (free plan) would not work since it banned GUIs. but u dont need a gpu for conversion or merging. see this video for merging https://youtu.be/8uq0e_i17FE
    1
  3477. see this vid: https://youtu.be/XHGlv-E7eBw
    1
  3478. r u thinking what im thinking
    1
  3479. omnomnomnom
    1
  3480. u mean colab?
    1
  3481. Many thanks!
    1
  3482. thanks for sharing!
    1
  3483. Thanks!
    1
  3484. Coming soon. My next video will be on RVC v2 installation, then I'll do tortoise TTS
    1
  3485. do you see any error in the terminal? also check if you have the right steps, cfg, denoise settings
    1
  3486. Thanks!
    1
  3487. Interesting, thanks for sharing!
    1
  3488. yes
    1
  3489. 😃
    1
  3490. where are you located?
    1
  3491. comfy is a user interface so unfortunately you will get disconnected on the colab free plan
    1
  3492. You're very welcome!
    1
  3493. thanks for sharing
    1
  3494. yes, If you use the same settings and seed, it should be a similar image
    1
  3495. ~5min
    1
  3496. Can't wait!
    1
  3497. Thanks. Is this what you're looking for? https://ai-search.io/tool/bashsenpai
    1
  3498. you'll need a cuda gpu
    1
  3499. it's the section that says 'run this command'. for example, if you're using conda & cuda 11.8, it'll be this 'conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia'
    1
  3500. im not sure. can you paste the error message here
    1
  3501. yeah that one was impressive!
    1
  3502. interesting, thanks for sharing!
    1
  3503. nope. i tried ;)
    1
  3504. Try to extract everything from the zip into a folder and then open it there
    1
  3505. Thank you. I hope so!
    1
  3506. it could be a variety of reasons. - try the start_https one - make it a shortcut - extract it first - try searching the comments here
    1
  3507. lol thanks!
    1
  3508. "Good artists copy, Great artists steal" Picasso
    1
  3509. see their github to determine what version to download: https://github.com/w-okada/voice-changer
    1
  3510. the voice part is not out yet
    1
  3511. the red section is what you want to keep. see this video for a demo: https://youtu.be/X7DnlE-c2eI
    1
  3512. no worries!
    1
  3513. what are the errors?
    1
  3514. are u in the EU
    1
  3515. what's the outcome you want?
    1
  3516. 🤯
    1
  3517. I've heard that our gut microbiome influences our mind to a significant extent. Regarding your other question, I don't think it's currently scientifically possible to prove a 'soul', if that's what you're referring to. It'd be cool if AGI could somehow figure this out!
    1
  3518. hi
    1
  3519. itll either be really good or really bad. Ill do a video on this soon. in any case, society will be very different than what it is today
    1
  3520. You're welcome!
    1
  3521. Thank you!
    1
  3522. it's banned on colab (free plan) 😢 u can use the paid plan tho. see this https://youtu.be/9TroP5mR3CM
    1
  3523. there will be some delay
    1
  3524. Thanks!
    1
  3525. ?
    1
  3526. Thank you!
    1
  3527. unzip it to find the pth inside. if your pc can't unzip, download winrar https://www.win-rar.com/download.html?&L=0
    1
  3528. i just asked it to do a mock interview with me. does that answer your question?
    1
  3529.  @fridemar  oh I see. i don't have iPad so unfortunately I don't know of a solution. i think only the phone app will work for advanced voice mode
    1
  3530. Yes you can!
    1
  3531. Thanks!
    1
  3532. WOW hidden meaning!
    1
  3533. lol he doesn't like those noodles very much
    1
  3534. ah, the AI uprising
    1
  3535. you'l need a cuda gpu w at least 8g vram. at least for now
    1
  3536. stay calm and wait for universal basic income
    1
  3537. ???
    1
  3538. oops
    1
  3539. you're right, pythagora is also great!
    1
  3540. You are welcome!
    1
  3541. good point!
    1
  3542. no need
    1
  3543. This is likely not going to work, unfortunately
    1
  3544. what step r u on?
    1
  3545. thats strange. do u see any error message in the cmd?
    1
  3546. Thanks!
    1
  3547. It's likely to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    1
  3548. next year
    1
  3549. It is quite intensive on the system
    1
  3550. you should get a lora error from switching the checkpoint only. did you try rerunning everything from the beginning
    1
  3551. put the pth in a zip file, upload it to google drive, then paste the link in download model
    1
  3552. Thanks!
    1
  3553. yes see this https://huggingface.co/docs/diffusers/using-diffusers/controlnet
    1
  3554. my pleasure!
    1
  3555. thanks for sharing!
    1
  3556. get ready!
    1
  3557. coming soon!
    1
  3558. you can use o1 in cursor. killer combo!
    1
  3559. i also notice that more & more tech giant ceos are indian men. is that racist too?
    1
  3560. 😃
    1
  3561. training requires a gpu
    1
  3562. i did some testing w the copilot voice. it's decent but not as versatile as the gpt voice. i didn't find enough additional use cases to make a full video, unfortunately
    1
  3563.  @TouhouGaijin  good luck!
    1
  3564. You're welcome!
    1
  3565. yes
    1
  3566. I do not have an AMD Graphcis card so I am not sure. Give it a try!
    1
  3567. You're very welcome!
    1
  3568. i believe/hope that Sam and Elon have good intentions. the original reason to change openai to for-profit is because they couldn't get enough funding to buy the hardware and hire top talent.
    1
  3569. 😃
    1
  3570. Thanks for sharing!
    1
  3571. you're welcome
    1
  3572. Hmmm I've never seen that ever before. does it work with a different lora or checkpoint
    1
  3573. LOL oops
    1
  3574. its out now! i'll post an installation tutorial soon
    1
  3575. Unfortunately animatediff is an A1111 plugin so it wouldn't work in diffusers
    1
  3576. Thank you!
    1
  3577. awesome tip! thanks for sharing!
    1
  3578. not llms specifically, but an AI could potentially be trained to hack encryption. it's not proven yet though
    1
  3579. damn what a plot twist!
    1
  3580. What are the errors?
    1
  3581. lol are you shorting it
    1
  3582. 😃
    1
  3583. the sheet doesn't work anymore. the owner likely removed it. try these instead: https://ai-search.io/voices https://voice-models.com/ https://weights.gg/ https://discord.gg/aihub
    1
  3584. How to Change Your Voice in Realtime on Discord (& Games) with this FREE AI https://youtu.be/XHGlv-E7eBw
    1
  3585. Try to extract everything from the zip into a folder and then open it there
    1
  3586. thx for sharing!
    1
  3587. 😢
    1
  3588. yes see this https://youtu.be/ixB9oalT3cQ
    1
  3589. amd doesn't work, and it's banned on colab (free plan) 😢 u can use the paid plan tho
    1
  3590. 😃
    1
  3591. Thanks!
    1
  3592. 😃
    1
  3593. Try to extract everything from the zip into a folder and then open it there
    1
  3594. yes! hope they release the code soon
    1
  3595. Thanks!
    1
  3596. Yes, coming soon!
    1
  3597. exciting times!
    1
  3598. yes, see my video on discord & games
    1
  3599. good suggestion!
    1
  3600. true
    1
  3601. 😃
    1
  3602. you can run it on cheaper GPUs. its very small
    1
  3603. thanks for sharing!
    1
  3604. Wow, thank you!
    1
  3605. You're very welcome!
    1
  3606. thanks!
    1
  3607. No but I'm the OP, so you shouldn't believe me ;) Let's see if someone else can confirm that its safe
    1
  3608. 😃😃😃
    1
  3609. glad you solved it!
    1
  3610. Thanks! here you go https://github.com/Mangio621/Mangio-RVC-Fork
    1
  3611. thanks for sharing. the beach boy & fireworks do look very real!
    1
  3612. Most amd users have reported no, but hopefully someone here has done it successfully and can reply to your comment.
    1
  3613. not at the moment, but you can choose from different voices
    1
  3614. you're welcome!
    1
  3615. Oh interesting, it might have slowed down since I last used it
    1
  3616. Perhaps these loras: https://civitai.com/models/73413/cag-alien-spaceships https://civitai.com/models/26578/xsarchitectural-17sciencefictioncityonmars
    1
  3617. see how to train a model here: https://youtu.be/ixB9oalT3cQ
    1
  3618. see this https://youtu.be/98drFL65VH0
    1
  3619. Thank you!
    1
  3620. lol i wonder when they'll release the voice
    1
  3621. We have a video on how to install the voice changer!
    1
  3622. did u sign up with a new account
    1
  3623. Gold medal right there
    1
  3624. Thanks!
    1
  3625. You are welcome!
    1
  3626. looks like you don't have enough vram to run this
    1
  3627. Thanks!
    1
  3628. Thanks!
    1
  3629. maybe your hardware isn't good enough. what's your cpu or gpu?
    1
  3630. The progress is crazy!
    1
  3631. You are welcome!
    1
  3632. yes try pulid: https://huggingface.co/spaces/yanze/PuLID-FLUX
    1
  3633. yes
    1
  3634. thanks for sharing
    1
  3635. yes 😃
    1
  3636. 😃
    1
  3637. see this https://youtu.be/tnfqIQ11Qek
    1
  3638. Try messing with the settings to see if anything helps! Try increasing and decreasing the chunks and extra to see if that helps
    1
  3639. what gpu do u have?
    1
  3640. Thanks!
    1
  3641. 😃😃😃
    1
  3642. Thanks!
    1
  3643. what gpu do you have? it should auto detect it, if it's a cuda nvidia one
    1
  3644. 😃
    1
  3645. Thanks!
    1
  3646. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  3647. Thanks!
    1
  3648. Welcome!
    1
  3649. run 'conda init' first
    1
  3650. hi!
    1
  3651. No problem 👍
    1
  3652. 😢
    1
  3653. time
    1
  3654. yes
    1
  3655. 🚀🚀🚀
    1
  3656. its (still) not out yet
    1
  3657. it'd be nice if other board members could come out and debunk Helen's claims
    1
  3658. Try to extract everything from the zip into a folder and then open it there
    1
  3659. You're welcome
    1
  3660. that's a good point!
    1
  3661. lol in hindsight i could've picked some better sounding tracks
    1
  3662. no problem
    1
  3663. because they're from separate labs. but once the code is out, i'm sure AI companies will try to merge these techniques together
    1
  3664. Great to hear!
    1
  3665. Try to extract everything from the zip into a folder and then open it there
    1
  3666. you're welcome!
    1
  3667. Glad it helped!
    1
  3668. right click and run with powershell instead of notepad
    1
  3669. what's the error log?
    1
  3670. 😃😃😃
    1
  3671. You're welcome!
    1
  3672. 😃😃😃
    1
  3673. nothing much 😃
    1
  3674. Thanks!
    1
  3675. Thanks for sharing. Although there will likely be huge risks ahead, we might have already reached the point of no return. The race to AGI can't be stopped - whoever stops will simply get wiped out by competitors/adversaries.
    1
  3676. for flux, yes
    1
  3677. I'm looking into this now. Happy new year!
    1
  3678. here you go https://youtu.be/KW9s_im2338
    1
  3679. Thanks for sharing!
    1
  3680. Thanks!
    1
  3681. 😃😃😃
    1
  3682. please paste the full error code here
    1
  3683. upgrade to python 3.9, or you might have run out of memory: https://stackoverflow.com/questions/72435382/how-can-i-fix-1-leaked-semaphore-objects-to-clean-up-at-shutdown-error-on-mac
    1
  3684. thanks!
    1
  3685. It works for any language. try to find a voice model in the same language
    1
  3686. posting tonight!
    1
  3687. No problem!
    1
  3688. Do you see any error message
    1
  3689. Did you unzip everything into a folder first
    1
  3690. well, microsoft CTO just said they haven't seen diminishing returns. if anything, improvement seems exponential
    1
  3691. See if anything on this thread works for you! https://github.com/w-okada/voice-changer/issues/401
    1
  3692. I use canva or photopea. Images are just found using google search. hope that helps!
    1
  3693. yes, but it'll be very slow. its best if u have a cuda gpu
    1
  3694. No problem!
    1
  3695. My pleasure!
    1
  3696. 😃
    1
  3697. Yes! There is a record button at the bottom! Click that to start recording!
    1
  3698. You're welcome!
    1
  3699. good luck!
    1
  3700. 😃
    1
  3701. Do you see any error message in the cmd
    1
  3702. its safe
    1
  3703. 🔥🔥🔥
    1
  3704. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    1
  3705. good news is, if you have what it takes, you can use this to create your own blockbuster animation without going through the painful step of being an underpaid inbetween junior for many years
    1
  3706. 🤯
    1
  3707. what's the exact merror message? are you setting too many epochs?
    1
  3708. yes, as long as you retained the .index and G and D files, you can train it further with new or more audio
    1
  3709. great idea!
    1
  3710. in the 'coming weeks'
    1
  3711. cool! thanks for sharing!
    1
  3712. yessss
    1
  3713. see this https://youtu.be/_JXbvSTGPoo
    1
  3714. Welcome!
    1
  3715. hmm it should install torch when you run install.ps1. maybe try running that again
    1
  3716. can you share some links?
    1
  3717. thanks for sharing!
    1
  3718. Awesome, thank you!
    1
  3719. 🚀🚀🚀
    1
  3720. u can use outpaint for that
    1
  3721. this is only for voice conversion, but not translation. you could use a japanese person's voice to speak english, if that's what you're asking
    1
  3722. what are the tools?
    1
  3723. Thanks for the info!
    1
  3724. thx for sharing!
    1
  3725. they changed it to 1 now
    1
  3726. they are based in US
    1
  3727. not yet
    1
  3728. yes, download python here https://www.python.org/downloads/
    1
  3729. 😃
    1
  3730. its ok, i have no friends
    1
  3731. 24g vram or more. some could run it with 12g via a comfyui extension which i haven't tried yet
    1
  3732. did you add conda to path yet
    1
  3733. Check out this video: How to Change Your Voice in Realtime on Discord (& Games) with this FREE AI https://youtu.be/XHGlv-E7eBw
    1
  3734. Interesting, thanks for sharing!
    1
  3735. lol
    1
  3736. You're very welcome!
    1
  3737. Yes, from the github: If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models.
    1
  3738. 🤯
    1
  3739. you're welcome!
    1
  3740. yes, but it's best if you add a lora to it
    1
  3741. 😃
    1
  3742. What are your computer specs? You might just not have a powerful enough computer.
    1
  3743. next video!
    1
  3744. probably
    1
  3745. You are welcome! Yes, it's very similar
    1
  3746. thanks for sharing!
    1
  3747. how about a full silicone body
    1
  3748. Thank you
    1
  3749. damn. before this video came out, it only took 1-2min. sorry everyone
    1
  3750. Thanks
    1
  3751. hi!
    1
  3752. 😃
    1
  3753. you're welcome!
    1
  3754. good luck!
    1
  3755. Thanks
    1
  3756. You’re welcome 😊
    1
  3757. AI was abusing the APM limits it had << lol that's crazy! thanks for sharing
    1
  3758. that might not be possible unless you run a generator locally. you can use cogvideo for that but the quality is not great https://github.com/THUDM/CogVideo
    1
  3759. lol
    1
  3760. it's offline
    1
  3761. if they open source it, im sure people will build all of the above!
    1
  3762. did u unzip everything into a folder first?
    1
  3763. 😃
    1
  3764. are u in the EU
    1
  3765. 👀
    1
  3766. Thank you so much ❤
    1
  3767. yes, the new voice feature isn't rolled out yet. they said in the coming weeks
    1
  3768. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  3769. Glad it was helpful!
    1
  3770. Thanks for the info!
    1
  3771. only nvidia gpus can train a model
    1
  3772. good idea
    1
  3773. lol
    1
  3774. It'll do a much better job than us
    1
  3775. Yes, this has an installer file, so you don't see to clone repos or anything. I also get lost w github stuff!
    1
  3776. thats what im waiting for
    1
  3777. https://youtu.be/ixB9oalT3cQ
    1
  3778. yes, but it might not fix larger flaws
    1
  3779. nice! thanks for sharing your experience
    1
  3780. 👋
    1
  3781. 😃
    1
  3782. nya
    1
  3783. im making a video now. will publish next week!
    1
  3784. here you go https://youtu.be/KW9s_im2338
    1
  3785. 😃
    1
  3786. good observation
    1
  3787. noted, coming soon!
    1
  3788. You're very welcome!
    1
  3789. u might need to upgrade torch https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/issues/55
    1
  3790. "them hoomans are all the same to me"
    1
  3791. yes!
    1
  3792. Thanks!
    1
  3793. You're very welcome!
    1
  3794. thanks for sharing!
    1
  3795. did u unzip into a folder first
    1
  3796. You downloaded a zip file right? Make sure you unzip everything into a folder on your computer before you open anything
    1
  3797. You're welcome!
    1
  3798. No problem!
    1
  3799. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    1
  3800. try w a new account
    1
  3801. Thanks
    1
  3802. exciting times!
    1
  3803. 😃
    1
  3804. It can, but not as well as NVIDIA GPU's. Please check that your download the right version for AMD GPU
    1
  3805. 🫢
    1
  3806. No problem
    1
  3807. ai is not going away, so instead of worrying about it replacing artists, i think artists should learn how to use it to supplement their work & gain a competitive advantage.
    1
  3808. Sure, i'll look into it
    1
  3809. afaik the checkpoints are in github but the function has not been added to the UI yet. will keep you posted
    1
  3810. You're welcome!
    1
  3811. r u using colab?
    1
  3812.  @mitchkiggins9661  colab is cracking down on people using a1111. it's best to install it locally
    1
  3813. sounds about right
    1
  3814. https://youtu.be/_JXbvSTGPoo
    1
  3815. i'll post a tutorial tonight
    1
  3816. 😃
    1
  3817. not yet. code is pending release
    1
  3818. 8gb is the recommended minimum
    1
  3819. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  3820. lol thanks for your support
    1
  3821. video coming tomorrow!
    1
  3822. here you go https://youtu.be/KW9s_im2338
    1
  3823. thanks!
    1
  3824. You are very welcome!
    1
  3825. indeed
    1
  3826. good luck. there are some flaws w this as well, but it's better than the other ones i've tried so far
    1
  3827. Video coming out tomorrow!
    1
  3828. interesting idea!
    1
  3829. You're welcome!
    1
  3830. AI hands
    1
  3831. agree!
    1
  3832. yes! pls check the github docs
    1
  3833. Thanks for sharing!
    1
  3834. interesting, thanks for sharing!
    1
  3835. yes, if you have enough vram. i'm not sure what the max is, but some ppl have generated 10s
    1
  3836. Please do not
    1
  3837. great to hear!
    1
  3838. hey sorry things are happening so fast. which model are u referring to?
    1
  3839. can you link to the video? there are many different forks of RVC
    1
  3840. 😃
    1
  3841. 😃
    1
  3842. Hope u get it to work!
    1
  3843. Check the discord in the description!
    1
  3844. For a long answer: https://ai-search.io/articles/why-does-ai-struggle-with-hands-and-fingers
    1
  3845. Extract everything from the zip into a folder and then open it there
    1
  3846. Welcome!
    1
  3847. Reinstall it. It probably got corrupted.
    1
  3848. thanks for sharing!
    1
  3849. that model requires sdxl which is a different pipeline: https://huggingface.co/docs/diffusers/using-diffusers/sdxl
    1
  3850. Thank you!
    1
  3851. interesting. thanks for sharing!
    1
  3852. Yes if you have good enough hardware
    1
  3853. You're very welcome!
    1
  3854. that's crazy!
    1
  3855. https://www.python.org/downloads/
    1
  3856. that's my suspicion too! glad you have the same thought
    1
  3857. Did you confirm that it was working with a regular output and not the virtual audio cable? Make sure the voice works normally first!
    1
  3858. It ultimately depends on how powerful of a machine you have.
    1
  3859. yeah. hopefully they'll extend the daily limit later
    1
  3860. You're very welcome!
    1
  3861. it depends on your hardware & how long the audio is
    1
  3862. noted, coming soon!
    1
  3863. making a tutorial now. it's pretty good!
    1
  3864. 😃
    1
  3865. yes, see this https://youtu.be/KW9s_im2338
    1
  3866. 😢
    1
  3867. true. i think copilot+PCs can only run copilot, but not open source stuff like ollama or comfyUI. not sure if their NPU is compatible
    1
  3868. I am not sure. Depends on what graphics card you have
    1
  3869. see this https://youtu.be/XHGlv-E7eBw
    1
  3870. it should be a success. see this: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/651 https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/853
    1
  3871. afaik there's no good realtime voice changer for mobile yet. realtime requires a high end gpu
    1
  3872. Thanks
    1
  3873. director: AI actor: AI scriptwriter: AI music: AI fx: AI
    1
  3874. My pleasure
    1
  3875. my pleasure! have fun
    1
  3876. Thanks!
    1
  3877. true, esp since a1111 has so many plugins
    1
  3878. is it out of memory? how much is your vram?
    1
  3879.  @helalsamrat  unfortunately that'll be too low
    1
  3880. it depends on the length of your audio
    1
  3881. yes
    1
  3882. hmm im not sure. try to logout & login again. or try in incognito
    1
  3883. Glad you enjoyed it!
    1
  3884. I’m honored to hear you think I sound like MrBeast 😌
    1
  3885.  @あかし中村  you can do it with tensor art
    1
  3886. uhoh, did you use conda (virtual env)?
    1
  3887. thanks for sharing!
    1
  3888. You're welcome!
    1
  3889. it works offline
    1
  3890. this vid is mostly reviewing their AI announcements. their other updates are pretty cool though!
    1
  3891. not sure if its free anymore, but try this https://youtu.be/A7d2LNRmX8k
    1
  3892. oh looks like its paid now :(
    1
  3893. yes
    1
  3894. according to livebench
    1
  3895. Great! Do you have an M1?
    1
  3896. God bless you, young man
    1
  3897. possibly! we don't know who's lying here
    1
  3898. Thanks!
    1
  3899. Interesting I have not heard of this before
    1
  3900. ti tends to be better for ai tasks
    1
  3901. yes rvc can do it. see this https://youtu.be/jh83ViDUvJs
    1
  3902. What is it doing? Can you try checking the logs?
    1
  3903. Thanks!
    1
  3904. yes, you can try hallo:https://youtu.be/rlnjcRP4oVc or live portrait: https://youtu.be/uyjSTAOY7yI
    1
  3905. good catch
    1
  3906. thanks!
    1
  3907. Thank you so much!
    1
  3908. they announced this today 😃
    1
  3909. Thanks. Indeed, the look of the device garnered a lot of attention
    1
  3910. that's correct, but try this https://youtu.be/tnfqIQ11Qek
    1
  3911. why do you think so?
    1
  3912. Thanks for the tip
    1
  3913. that would be great!
    1
  3914. https://www.nvidia.com/en-us/design-visualization/rtx-5000/
    1
  3915. hunyuan & mochi
    1
  3916. Wow, thanks
    1
  3917. "good artists copy, great artists steal"
    1
  3918. 😃
    1
  3919. Do you see any error message in the logs
    1
  3920. yes!
    1
  3921. im dying to make a video on it. but i can't even signup for an account yet lol
    1
  3922. we'll find out very soon when it gets rolled out
    1
  3923. 😃
    1
  3924. it's similar. see this https://huggingface.co/docs/diffusers/using-diffusers/sdxl
    1
  3925. 🤯
    1
  3926. Thanks!
    1
  3927. great!
    1
  3928. see this https://huggingface.co/docs/diffusers/using-diffusers/controlnet
    1
  3929. Please see my latest video!
    1
  3930. Thanks!
    1
  3931. not available yet. they are still developing it
    1
  3932. Thanks for sharing! I'll do a video on Flux once the ecosystem is more mature.
    1
  3933. Thanks!
    1
  3934. 😃
    1
  3935. You're welcome
    1
  3936. nice! is it the one by kijai? just saw it now
    1
  3937. Thanks!
    1
  3938. Try to extract everything from the zip into a folder and then open it there
    1
  3939. soon hopefully. they said it will be open source
    1
  3940. Yes! https://youtu.be/tnfqIQ11Qek
    1
  3941. You're welcome!
    1
  3942. what gpu do you have?
    1
  3943. Try looking at the logs and seeing what they say to help debug!
    1
  3944. Thanks & good point. I'd love a DAW integration where the AI can generate/edit individual tracks, and then we can do the mixing/mastering
    1
  3945. 😀
    1
  3946. what error message do u see
    1
  3947. can i get some more ingredients pls
    1
  3948. sign me up! does it also include 7 years of food & ammo?
    1
  3949. lol it's trying so hard to not violate the guidelines
    1
  3950. hi!
    1
  3951. how is it?
    1
  3952. if you're drunk enough
    1
  3953. 😏
    1
  3954. true
    1
  3955. Thanks!
    1
  3956. 😃😃😃
    1
  3957. Hope you enjoy!
    1
  3958. tutorial coming soon!
    1
  3959. Thank you!
    1
  3960. It makes the voice unique!
    1
  3961. 😃
    1
  3962. you're welcome!
    1
  3963. You're welcome
    1
  3964. You're welcome!
    1
  3965. try this https://youtu.be/tnfqIQ11Qek
    1
  3966. Yes!
    1
  3967. 😃
    1
  3968. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  3969. 😃
    1
  3970. try this https://youtu.be/TSLM1mEAc18
    1
  3971. yes, it's one of the most accurate
    1
  3972. It has to be compatible with your checkpoint
    1
  3973. i clearly said it was just an observation, without any conclusion.
    1
  3974. Thanks for your feedback. I'm slightly concerned that none of the OpenAI leaders have denied this. If this was fake, it wouldn't be hard to simply say so.
    1
  3975. thanks!
    1
  3976. deep fried
    1
  3977. try this https://youtu.be/TSLM1mEAc18
    1
  3978. Hard to say. I'd guess 2025-2030 we'll have AGI. Heck, they might already have it secretly. But UBI might not able available yet for various reasons. Government is greedy. Distribution of wealth and abundance has always been unfair
    1
  3979. 😃
    1
  3980. If you were following what I was doing and making a shortcut, make sure you are creating a link and not a copy. If you copy it out it won't have anything to run off of. A link runs the original file in the main folder.
    1
  3981. 😃
    1
  3982. you can, but it may be slow. if it's too slow, try this online method instead https://youtu.be/TSLM1mEAc18
    1
  3983. Check out our original tutorial!
    1
  3984. Thanks for the sub!
    1
  3985. that's crazy! thanks for sharing
    1
  3986. i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    1
  3987. Thanks!
    1
  3988. Thanks for the sub!
    1
  3989. Sure, i'll add a pinned comment w the prompts
    1
  3990. interesting. afaik, the cutoff is 2023 but it can search the internet with RAG
    1
  3991. @batriq_ibn_bahtmani nope, but try doing it online: https://youtu.be/tnfqIQ11Qek
    1
  3992. Thanks for the sub!
    1
  3993. t takes 10 seconds on my 16g gpu.
    1
  3994. He's actually a GOBLIN
    1
  3995. thanks for sharing!
    1
  3996. good idea
    1
  3997. cant wait for 1.1 dev!
    1
  3998. RVC or w-okada
    1
  3999. Thanks! More to come!
    1
  4000. You're welcome!
    1
  4001. nice!
    1
  4002. great idea. i heard it pays a lot!
    1
  4003. Thanks!
    1
  4004. no problem!
    1
  4005. good point - it didnt scroll down at all!
    1
  4006. My pleasure!
    1
  4007. You're welcome!
    1
  4008. that's crazy. thanks for sharing
    1
  4009. w-okada (see our channel for tutorial)
    1
  4010. time to start binging on junk food!!!
    1
  4011. good point. fanfic will go wild from this
    1
  4012. Thanks!
    1
  4013. Maybe Topaz is based on the same thing. I look forward to testing DiffVSR once it's out!
    1
  4014. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  4015. not that i know of, but that's a great use case!
    1
  4016. you can try this https://youtu.be/5lef0QV-kiQ
    1
  4017. Thanks! Appreciate it
    1
  4018. https://youtu.be/_JXbvSTGPoo
    1
  4019. 24g is recommended. 12g is bare minimum
    1
  4020. it'll likely be quite censored. you can try grok or mistral for more uncensored stuff
    1
  4021. Thanks!
    1
  4022. ??? whats the error message
    1
  4023. You're welcome
    1
  4024. btw, this tool is not free anymore
    1
  4025. good idea
    1
  4026. way better
    1
  4027. 😃
    1
  4028. there are now multiple open source competitors to Devin as well. interesting to see how this plays out
    1
  4029. Make sure you install the GPUcuda version if using an Nvidia graphics card. It is very likely that you have the DirectMLcuda version.
    1
  4030. 😃
    1
  4031. You're welcome!
    1
  4032. Try going through this tutorial: https://youtu.be/IS_SPQVv5iY
    1
  4033. Seriously
    1
  4034. 😃
    1
  4035. 🤯
    1
  4036. not in tortoise, but see this https://youtu.be/jh83ViDUvJs
    1
  4037. 😧 did u contact support
    1
  4038. what if we're just in a simulation
    1
  4039. this reminds me of the joke: Scientists building artificial super intelligence to find out if there’s a God. They finally finish & ask the question. AI replies: “There is now, mofos!”
    1
  4040. Thanks!
    1
  4041. Did you extract everything from the zip into a folder first?
    1
  4042. No problem 👍
    1
  4043. yeah this is the 2nd time I've had an AI song stuck in my head! 1st time was Udio
    1
  4044. see this https://youtu.be/o6ywe3EDLDM
    1
  4045. It uses the same algorithm
    1
  4046. Are you using headphones? If you use your speaker there will be an echo
    1
  4047. yes
    1
  4048. my pleasure!
    1
  4049. yes, very frustrating that none of these have been released to the public
    1
  4050. u cant train offline, but try this https://youtu.be/tnfqIQ11Qek
    1
  4051. lol good catch!
    1
  4052. it's in the free plan. did you turn on Artifacts? also, what country are you in?
    1
  4053. It might still work. But it's best to use a model trained in the same language that you want to use.
    1
  4054. For example, English models still work well for Japanese, since English encompasses all the sounds of Japanese. But the reverse is not true. A Japanese AI would sound awful speaking English
    1
  4055. 😨
    1
  4056. 😃
    1
  4057. thanks for your support!
    1
  4058. Glad you were able to solve it
    1
  4059. My pleasure!
    1
  4060. What error do you see?
    1
  4061. im looking into alternatives and will update everyone asap!
    1
  4062. Video coming out tomorrow!
    1
  4063. you're welcome!
    1
  4064. 😃
    1
  4065. what do you mean by 2-3-4? can you explain further
    1
  4066. yes
    1
  4067. 😃
    1
  4068. why
    1
  4069. Yes, please see my latest video!
    1
  4070. 😃😃😃
    1
  4071. there are no details yet on the VR200 afaik
    1
  4072. Try to reinstall it
    1
  4073. Thank you!
    1
  4074. i clearly said it was just an observation, without any conclusion
    1
  4075.  @mastermanhwas  I think you can. Try searching for onnxstream https://github.com/vitoplantamura/OnnxStream
    1
  4076. oh really? looks like the reduced it
    1
  4077. I find that the morphing is around the same
    1
  4078. Try to extract everything from the zip into a folder and then open it there
    1
  4079. exactly. they have the tech for this
    1
  4080. from google: "Conformation refers to the spatial arrangement of atoms in a molecule that can be determined through single-bond rotation."
    1
  4081. The index file is generally created when training a new voice model. is that what you're trying to do?
    1
  4082. Thanks
    1
  4083. Thanks
    1
  4084. lol did Neuro Sama try that question?
    1
  4085.  @Endersgamejp  she's secretly AGI
    1
  4086. how about next month
    1
  4087. colab will disconnect GUIs now. its best to install this locally
    1
  4088. In colab?
    1
  4089.  @Emily_qq00  i'll need to see if that's even possible. they don't have documentation like Diffusers, so it'd be hard to dissect the code
    1
  4090. i think it's their own fine tuned version
    1
  4091. Likely a hardware problem. you need a good nvidia gpu to have minimal delay
    1
  4092. Try to extract everything from the zip into a folder and then open it there
    1
  4093. lol
    1
  4094. Make sure you install the English version
    1
  4095. the brazil one changed the url to what used to be the original. see this: https://youtu.be/98drFL65VH0
    1
  4096. at least its free!
    1
  4097. 😃
    1
  4098. yes, but who wants to work anyways
    1
  4099. lol
    1
  4100. they've open sourced it, so perhaps you could fork it and add your prompt somehow?
    1
  4101. whats the error
    1
  4102. Collab is cracking down on people using it for voice cloning, so it's best if you can install this locally
    1
  4103. you're welcome!
    1
  4104. Perhaps try running as admin
    1
  4105. lol
    1
  4106. I'm not aware of any virus, but use at your own discretion
    1
  4107. yes just delete the folder and whatever you downloaded
    1
  4108. Thanks!
    1
  4109. 😃
    1
  4110. Thanks!
    1
  4111. Thank you!
    1
  4112. nice!
    1
  4113. for multiple loras, try this: https://medium.com/@natsunoyuki/using-civitai-loras-with-diffusers-e3ef3e47c413 for controlnet: https://huggingface.co/docs/diffusers/using-diffusers/controlnet
    1
  4114. Yes. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  4115. Yes you can! see this https://youtu.be/jh83ViDUvJs You can use different accents as well
    1
  4116. yes, they are the openai team
    1
  4117. do u see any error in the cmd?
    1
  4118. colab is cracking down on people using this for voice cloning. It's best if you can install it locally
    1
  4119. great idea!
    1
  4120. Damn that got really dark
    1
  4121. 😃
    1
  4122. Wow, thanks
    1
  4123. Thanks!
    1
  4124. exponential growth!
    1
  4125. win7 might be too old
    1
  4126. Thanks for subbing!
    1
  4127. lol
    1
  4128. yeah its crazy how he answers so fast. i didn't even finish reading the question, let alone understand it!
    1
  4129. Thanks for sharing. It turns out mystic is just a wrapper for flux, which is disappointing.
    1
  4130. Try reinstalling it! Sometimes it glitches during install which causes it to sound weird
    1
  4131.  @MaximilianFeichtinger  no worries. always welcome constructive feedback!
    1
  4132. this is different from rvc in that you can't do voice conversion or import a voice model
    1
  4133.  @personmuc943  Yes, this can generate a full song.
    1
  4134. Thank you so much! Sure, thats a good idea
    1
  4135. fake 'evidence' will be rampant in the coming years for sure!
    1
  4136. see this https://youtu.be/_JXbvSTGPoo
    1
  4137. 😃😃😃
    1
  4138. My pleasure!
    1
  4139. 😃
    1
  4140. Thanks. Appreciate it!
    1
  4141. Glad it was helpful!
    1
  4142. Coming soon!
    1
  4143. nice! thanks for sharing
    1
  4144. 😃
    1
  4145. yes!
    1
  4146. 🤯
    1
  4147. what about a water gun?
    1
  4148. yes, try to search for the lora on HF first. if it exists, you can just copy that link
    1
  4149. only nvidia cuda gpus work
    1
  4150. that's what they claim. though these could be cherry-picked examples. i can't wait to try it out myself!
    1
  4151. nice! looks like they are nearing a deal soon
    1
  4152. Thanks!
    1
  4153. yes
    1
  4154. 😉
    1
  4155. "like mushrooms in forest after rain" lol that's a good analogy
    1
  4156. indeed!
    1
  4157. No problem!
    1
  4158. remove the spaces and dots in your wav file
    1
  4159. Glad you like them!
    1
  4160. 😃
    1
  4161. exactly
    1
  4162. only english & chinese for now
    1
  4163. even if this was open source, the compute (thinking) would be really expensive
    1
  4164. It is working for me. Try the link again. It is a pretty big discord so the link may be glitching out because it is used so much.
    1
  4165. from here they say 15g is needed. https://github.com/magic-quill/MagicQuill/issues/5
    1
  4166. Thanks! 2080ti is enough for this. good luck!
    1
  4167. 😃
    1
  4168. you'll need a cuda gpu to run it locally
    1
  4169. No problem
    1
  4170. 😃😃😃
    1
  4171. yes, try this: https://huggingface.co/spaces/mrfakename/E2-F5-TTS
    1
  4172. No problem 👍
    1
  4173. Welcome!
    1
  4174. 🤯
    1
  4175. good luck!
    1
  4176. Thanks for sharing! I try to be nice as well, just in case...
    1
  4177. You're welcome!
    1
  4178. lol yes
    1
  4179. It's the same. okada is based on RVC
    1
  4180. did u unzip into a folder first
    1
  4181. haha, so it understands reality
    1
  4182. 😃
    1
  4183. yeah that would be a good function to have
    1
  4184. it has been delayed to the 'coming months' now
    1
  4185. try qwen chat
    1
  4186. do u mean for training a voice, or just converting?
    1
  4187. Thanks!
    1
  4188. you can also use venv
    1
  4189. hi
    1
  4190. You're welcome
    1
  4191. thanks!
    1
  4192. yo
    1
  4193. shit
    1
  4194. thanks
    1
  4195. 🙌
    1
  4196. try this https://youtu.be/tnfqIQ11Qek
    1
  4197. lol
    1
  4198. yeah 16 is too low unfortunately
    1
  4199. Thanks for sharing!
    1
  4200. "those annoying hoomans. keep interrupting me and forcing me to do stupid things"
    1
  4201. 😃😃😃
    1
  4202. why
    1
  4203. My pleasure!
    1
  4204. 😈
    1
  4205.  @screenink5125  That makes sense. Thanks for explaining!
    1
  4206. interesting. thanks for sharing
    1
  4207. You need a mic
    1
  4208. indeed!
    1
  4209. ys M3 is great. you can run it locally here: https://youtu.be/TSLM1mEAc18
    1
  4210. You're welcome
    1
  4211. 'we'
    1
  4212. Everything is in the description
    1
  4213. no, you need a cuda gpu
    1
  4214. Thanks!
    1
  4215. enter conda init first
    1
  4216. no problem!
    1
  4217. thanks
    1
  4218. Thanks for the tip!
    1
  4219. Try to extract everything from the zip into a folder and then open it there
    1
  4220.  @DoNotDisturb07  Hmmm, try the https one. does it work?
    1
  4221. can you give some examples
    1
  4222. lol
    1
  4223. yes, absolutely disgusting. cute girls don't chew like that
    1
  4224. i doubt it. runway has been quite expensive in the past
    1
  4225. you can ignore it. windows doesn't have triton
    1
  4226. You're welcome!
    1
  4227. 🥲
    1
  4228. do you have a cuda gpu?
    1
  4229. @coldbreezeproductions that's great!
    1
  4230. are you using AMD? some users have reported problems w AMD GPUs
    1
  4231. 😃
    1
  4232. they are very similar. i find udio's vocals are better, suno has more consistent melodies
    1
  4233. 🤣
    1
  4234. do you have nvidia?
    1
  4235. lol
    1
  4236. open source ftw!
    1
  4237.  @ChaseEverything  yes
    1
  4238. it's video
    1
  4239. lol im sure the real one would be much better and you could probably tell it to tone down
    1
  4240. fun times
    1
  4241. Thanks. And to clarify, nowhere in the video did I express dislike for any gender or race. I'm just pointing out a pattern. eg. I also notice that more & more tech giant CEOs are Indian men.
    1
  4242. Thanks!
    1
  4243. 😃😃
    1
  4244. 🤫🤫🤫
    1
  4245. that's why it takes 3h
    1
  4246. You're welcome!
    1
  4247. 😃
    1
  4248. ?
    1
  4249. bro, i even sing in one of my videos
    1
  4250. Yes definitely. See the section on training in this video
    1
  4251. I use the discord showcased in the original tutorial we made!
    1
  4252. That'd be fun!
    1
  4253. You’re welcome 😊
    1
  4254. they have a new one now
    1
  4255. exactly
    1
  4256. lol good catch
    1
  4257. oh interesting. are you on the free plan?
    1
  4258. yes flux is amazing!
    1
  4259. Hmmm I just tested joining today with another account and it works fine. It should have a 'Are you human?' verification checkbox. Perhaps try using another account, web browser, or computer.
    1
  4260. 😃
    1
  4261. sounds like the future!
    1
  4262. no problem!
    1
  4263. thanks
    1
  4264. it's not out yet. you can join their waitlist for early access
    1
  4265. Most welcome
    1
  4266. No problem
    1
  4267. Exactly! I'm glad someone caught this
    1
  4268. you can't. its closed source
    1
  4269. same here!
    1
  4270. for the 1st error, you need to enter conda activate first for the python error, it looks like you don't have python installed or haven't added it to the path. did you follow the steps in the video to install miniconda ?
    1
  4271. maybe it's because you disabled the python app? try to enable it
    1
  4272. Yes it's best to use a voice model with the same language that you want to convert to
    1
  4273. Thanks for sharing!
    1
  4274. I'll look into it. I haven't used rvc in a while
    1
  4275. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  4276. around the same, but better quality. i think v2 requires more vram for training
    1
  4277. We have a video for it!
    1
  4278. dont know if theres mac support yet. you can use this instead https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  4279. thanks for sharing!
    1
  4280. GPU is mandatory
    1
  4281. you might have exceeded the daily limit for the free plan
    1
  4282. what's the error? paste more of the error before this
    1
  4283. 🤑🤑🤑
    1
  4284. My pleasure!
    1
  4285. hmmm did you check if the model (.pth) works?
    1
  4286. thanks for sharing. i wish there was a handbook on prompt metatags that work
    1
  4287. Thanks!
    1
  4288. it looks like he'll release it after paper submission https://github.com/xh9998/DiffVSR-project/issues/1
    1
  4289. Check the description
    1
  4290. You're very welcome!
    1
  4291. perhaps we're in a simulation
    1
  4292. RIP 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s
    1
  4293. what error message do you see
    1
  4294. no problem
    1
  4295. 🐱
    1
  4296. No problem!
    1
  4297. are u in EU
    1
  4298. 👍
    1
  4299. please paste more of the error. it should say which line broke
    1
  4300. its fast but quality is not great. check out trellis if you want better quality https://youtu.be/jXCCTg9GRmk
    1
  4301. yes!
    1
  4302. You're welcome
    1
  4303. No problem!
    1
  4304. Thanks!
    1
  4305. 😃
    1
  4306. glad you got it to work!
    1
  4307. 😃
    1
  4308. imagine if the stop button is broken
    1
  4309. No problem!
    1
  4310. are you trying to train a voice
    1
  4311. My pleasure!
    1
  4312. noted!
    1
  4313. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  4314. Sorry, I don't. I'm thinking of creating one soon. If you have any questions right now, please ask them here and I'll try to respond asap
    1
  4315. lets gooooo
    1
  4316. thanks!
    1
  4317. Yes, it is! It's a leap compared to previous records, but 14% is still a very low success rate.
    1
  4318. me too
    1
  4319. 🙌
    1
  4320. hmmm thats the first time i heard of this issue. do u see any error messsage
    1
  4321. 😃
    1
  4322. Runway needs to cut their prices now. Way too expensive compared to Kling & Minimax
    1
  4323. That function is not available yet. It says they will publish it soon: https://github.com/KwaiVGI/LivePortrait/issues/27
    1
  4324. Unfortunately, the cat dancing was just added by me 🥲
    1
  4325. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  4326. indeed!
    1
  4327. it could be a narrow AI that can decrypt AES. or possibly AGI, since it has cross domain learning
    1
  4328. w-okada: https://youtu.be/_JXbvSTGPoo
    1
  4329. do you see any error message
    1
  4330. lol yep i wrote about this in my newsletter https://aisearch.substack.com/p/rabbits-deadly-data-leak-ai-creates
    1
  4331. hmm can u try deleting the folder and reinstalling. Let me know if that still doesn't work
    1
  4332. Thanks for your feedback!
    1
  4333. my fav is kling i2v for anime. in other words, generate an anime image first, then plug it into kling i2v
    1
  4334. ?
    1
  4335. Thanks!
    1
  4336. 🤯
    1
  4337. that'd be nice!
    1
  4338. 🤐
    1
  4339. thanks!
    1
  4340. try this https://youtu.be/tnfqIQ11Qek
    1
  4341. 😃
    1
  4342. I don't really remember - probably on X. There's so much crazy news everyday it's hard to keep up
    1
  4343. You're welcome!
    1
  4344. enter conda init, then exit the session and start a new one. then enter conda activate
    1
  4345. ?
    1
  4346. 😃
    1
  4347. 🚀🚀🚀
    1
  4348. unzip it with winrar
    1
  4349.  @psychoherobrine3406  u can open a 7z with winrar too
    1
  4350. pro version is closed source, but you can install dev/schnell. see the latter half of this: https://youtu.be/K3xJ7GQuHpw
    1
  4351. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    1
  4352. 😢
    1
  4353. ❤️
    1
  4354. Noted!
    1
  4355. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  4356. enter conda init first
    1
  4357. My pleasure!
    1
  4358. Thanks for sharing!
    1
  4359. I've heard kaggle will try to disconnect you too
    1
  4360. good luck!
    1
  4361. 🤑
    1
  4362. 😃😃😃
    1
  4363. i haven't found a reliable way to control the length of a word, but try this: heart -> hearrrrt for screams, try [scream] or [screaming]
    1
  4364. see the last part of this video https://youtu.be/ixB9oalT3cQ
    1
  4365. Thanks. There's no official name - I made it with Udio: https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc Here's how: https://youtu.be/X7DnlE-c2eI
    1
  4366. Good spot!
    1
  4367. Thanks! Coming soon!
    1
  4368. Thanks!
    1
  4369. yes but it's best to use a voice model in the same language
    1
  4370. I haven't tried this, but you can run multiple controlnet instances (turn on in settings) and have the 1st controlnet swap the 1st face, and the 2nd controlnet swap the other face
    1
  4371. did you watch the w-okada tutorial yet? you can talk like goku in realtime
    1
  4372. yes just an agentic framework using gemini 2
    1
  4373. it might have been updated. first of all, do you have at least 12g vram (preferably 24g)?
    1
  4374. im looking into alternatives and will update everyone asap!
    1
  4375. Video coming out tomorrow!
    1
  4376. you're using the online interface. the local version is uncensored
    1
  4377. Yes, it just creates files in the original folder. To uninstall just delete the folder!
    1
  4378. Click the record button at the bottom!
    1
  4379. yes, but you'll get a lot of delay
    1
  4380. Check out our video on how!
    1
  4381. Did you move the start file outside the folder instead of create a link?
    1
  4382. Maybe try installing in a different location? Make sure that all of the dependencies are installed and not corrupted.
    1
  4383. conda env remove --name f5, then delete the folder
    1
  4384. it'll be very slow. try this instead https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  4385. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  4386. around $1.5
    1
  4387. that's another great use case. thanks for sharing
    1
  4388. yep, mj has a lot of catching up to do
    1
  4389. do u have a cuda gpu
    1
  4390. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  4391. it's the text of your uploaded audio. if you leave it blank, it will just auto add it for you. you just need to dl all the files i showed in the video to get it working
    1
  4392. Check the logs to see what it says!
    1
  4393. i seriously need this
    1
  4394. nice. thanks for sharing
    1
  4395. You can try decreasing the chunk and extra. Also make sure to not use the index. This will lower the delay, but might cause the quality to become worse
    1
  4396. great!
    1
  4397. ah, of course
    1
  4398. Thanks for sharing
    1
  4399. lol cuz the next fold would be way past that
    1
  4400. 😃
    1
  4401. for the gpu error, you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  4402. Nice work!
    1
  4403. Check pinned comment
    1
  4404. 😢
    1
  4405. cool idea. it sounds possible, since it's just json generation
    1
  4406. try this https://youtu.be/tnfqIQ11Qek
    1
  4407. i'm looking forward to this feature!
    1
  4408. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  4409. 🤯
    1
  4410. I will be working on making a video soon!
    1
  4411. Try to extract everything from the zip into a folder and then open it there
    1
  4412. only txt2vid for now
    1
  4413. You're very welcome!
    1
  4414. 😢
    1
  4415. true, it's like they have their own personalities
    1
  4416. ofc you can pay with bitcoin. search lightning network
    1
  4417. Thanks!
    1
  4418. looks like they did for august: https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard
    1
  4419. imagine if your landlord was a fan
    1
  4420. 😃
    1
  4421. it looks like he'll release it after paper submission https://github.com/xh9998/DiffVSR-project/issues/1
    1
  4422. lol
    1
  4423. we need a neck swapper next
    1
  4424. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  4425. Good luck!
    1
  4426. lol what did you make?
    1
  4427. Thank you!
    1
  4428. flux dev
    1
  4429. yes!
    1
  4430. No problem!
    1
  4431. lol
    1
  4432. use python. the code is in the google drive link in the description
    1
  4433. 😃
    1
  4434. yes!
    1
  4435. indeed, very exciting!
    1
  4436. Agree. I still remain doubtful of such promises. But with the exponential improvements in AI, who knows.
    1
  4437. 😃
    1
  4438. you're welcome!
    1
  4439. Which video?
    1
  4440. make denoising strength higher
    1
  4441. nice!
    1
  4442. Thanks for sharing!
    1
  4443. Thank you!
    1
  4444. ?
    1
  4445. yes. you can also do it in non-realtime if you just need the recording: https://youtu.be/ixB9oalT3cQ
    1
  4446. shhhh
    1
  4447. Yup, that's in the video. A very interesting paper
    1
  4448. not that I know of. will keep you posted
    1
  4449. yes thats the install size
    1
  4450. thx for sharing!
    1
  4451. 😃
    1
  4452. 🤯
    1
  4453. You're welcome!
    1
  4454. sounds like fb
    1
  4455. 😃😃😃
    1
  4456. I made one - see our Youtube!
    1
  4457. still here
    1
  4458. yup, it only assumes u have limitless paper
    1
  4459. yes
    1
  4460. thanks for sharing!
    1
  4461. thanks!
    1
  4462. no problem!
    1
  4463. do u have a cuda gpu
    1
  4464. You're welcome!
    1
  4465. 😃😃😃
    1
  4466. No problem!
    1
  4467. 😃
    1
  4468. yes
    1
  4469. Thanks for sharing!
    1
  4470. probably both
    1
  4471. afaik the checkpoints are in github but the function has not been added to the UI yet. will keep you posted
    1
  4472. possibly
    1
  4473. Thanks!
    1
  4474. Truth! (or turn into an anime girl & catfish on discord)
    1
  4475. yes!
    1
  4476. Try restarting the program
    1
  4477. 😃
    1
  4478. xyz+time
    1
  4479. what do u mean
    1
  4480. i heard gemini 1.5 pro is quite good for coding
    1
  4481. You are welcome!
    1
  4482. You are welcome!
    1
  4483. Microsoft has something similar called VASA-1. They just haven't released it to everyone
    1
  4484. yes, task manager is built into windows
    1
  4485. try to replace the korean characters with english. that might be causing the error
    1
  4486. it seems u don't have tensorboard yet. try pip install tensorboard in your cmd (I assume u already have pip & python installed?)
    1
  4487. yes!
    1
  4488. they better do it soon, cuz the ocean is dying!
    1
  4489. RIP tutors
    1
  4490. 😃
    1
  4491. @rootor1 Cool! Thanks for the tip!
    1
  4492. 😃
    1
  4493. Thanks for sharing!
    1
  4494. no problem!
    1
  4495. yeah it's mastered surprisingly well!
    1
  4496. just buy nvda stock 😃
    1
  4497. It's called w-okada: https://youtu.be/_JXbvSTGPoo
    1
  4498. I don't think you can add extensions, which are built for A1111
    1
  4499. Thank you!
    1
  4500. You are welcome!
    1
  4501.  @captanblue  Cool idea! hope u find something that works
    1
  4502. hmmm can you link to an example?
    1
  4503. thanks for sharing!
    1
  4504. 😃
    1
  4505. lol true
    1
  4506. did you turn on Artifacts?
    1
  4507. The guy said it's realtime, so I'll give him the benefit of the doubt. The other non-realtime tools I mentioned are alternatives for people with lower quality hardware which probably can't do realtime.
    1
  4508. I mean you have a graphics card not just integrated graphics. I was using a GTX 1080, so it should work for you!
    1
  4509. good point. hands & fingers were also avoided
    1
  4510. type conda init, then restart the session
    1
  4511. 😃
    1
  4512. thanks
    1
  4513. 😃😃😃
    1
  4514. which notebook or tool are you using
    1
  4515. INSANE!
    1
  4516. 1m of training data might not be good enough. try 10 minutes for 100-300 epochs. what do u mean by putting in audacity?
    1
  4517.  @Sontululuca50  Yes, just clean audio samples of your friend talking. You don't need to fragment it.
    1
  4518. spoiler alert: no one will pause anything
    1
  4519. 🥲
    1
  4520. Thanks!
    1
  4521. You need the DirectMLcuda version. I don’t think you can select a gpu in that version but it’s the one you need to use for AMD
    1
  4522. Thanks!
    1
  4523. yes, see https://youtu.be/XHGlv-E7eBw
    1
  4524. You're welcome!
    1
  4525. yes 🤯
    1
  4526. Thanks!
    1
  4527. It says they will publish it soon: https://github.com/KwaiVGI/LivePortrait/issues/27
    1
  4528. it could but it wont lol (guidelines)
    1
  4529. Choose a checkpoint, then find the ipynb here https://github.com/nolanaatama/sd-1click-colab/tree/main
    1
  4530. nice!
    1
  4531. Yeah it got flagged for some reason! I've opened a case with YT support
    1
  4532. not accurately. canva is best for that
    1
  4533. thanks & good luck! that'd be a game changer
    1
  4534. Nice! Thanks for sharing
    1
  4535. thanks!
    1
  4536. Unzip everything into a folder before opening
    1
  4537. thx for sharing!
    1
  4538. you mean voice to voice?
    1
  4539. Thanks!
    1
  4540. 😃
    1
  4541. Yes, absolutely
    1
  4542. u need to include pip when u install python
    1
  4543. we need a handswapper next
    1
  4544. thanks!
    1
  4545. You might have maxed out the free daily limit
    1
  4546. I hope there's an option to turn it off
    1
  4547. No problem!
    1
  4548. you can use it in a game. see this: https://youtu.be/XHGlv-E7eBw
    1
  4549. I use the Shure SM7B!
    1
  4550. i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    1
  4551. My pleasure!
    1
  4552. 😃
    1
  4553. Thanks
    1
  4554. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    1
  4555. Thanks
    1
  4556. it connects to openai vision, so you can't run it offline completely
    1
  4557. yes any language
    1
  4558. Thanks
    1
  4559. Good luck!
    1
  4560. that's a good use case!
    1
  4561. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  4562. here you go https://youtu.be/KW9s_im2338
    1
  4563. looks like something to do with your input/output settings. its hard to troubleshoot though
    1
  4564. Thanks!
    1
  4565. yes!
    1
  4566. Yep, they're both based off of stable diffusion
    1
  4567. Thanks!
    1
  4568. shit
    1
  4569. for colab and kits, yes. u just need internet
    1
  4570. lol
    1
  4571. i'm also curious why
    1
  4572. seems like it couldn't find any cuda file. do u have a cuda gpu & did u install the drivers
    1
  4573. No problem 👍
    1
  4574. you can record, but it's better to use RVC for that: https://youtu.be/ixB9oalT3cQ
    1
  4575. fundamentally, neural networks were built & trained on cuda gpus, which amd lacks. there's no support for amd currently
    1
  4576. i'll have to look into this. thanks for suggesting!
    1
  4577. i hope so too so i can go on vacation!
    1
  4578. Thanks!
    1
  4579. Yes!
    1
  4580. noted, coming soon!
    1
  4581. Thanks for sharing!
    1
  4582. yes, lets say your seed is 1234, and your batch size is 3. the seeds for the 3 images would be 1234, 1235, 1236
    1
  4583. You're welcome!
    1
  4584. Thanks
    1
  4585. good spot
    1
  4586. not your own gpu
    1
  4587. 😃😃😃
    1
  4588. canva has some options in the elements > audio section
    1
  4589. the voice changer? see this: https://youtu.be/_JXbvSTGPoo
    1
  4590. lol
    1
  4591. 😏😏😏
    1
  4592. 👀
    1
  4593. Foundation models trained in the Omniverse likely understand gravity. See my latest video on Nvidia.
    1
  4594. No problem
    1
  4595. yes!
    1
  4596. it seems u don't have tensorboard yet. try pip install tensorboard in your cmd (I assume u already have pip & python installed?)
    1
  4597. seriously. imagine what'll come out next month
    1
  4598. do u see any error message
    1
  4599. glad it worked. you can also try this online version https://youtu.be/TSLM1mEAc18
    1
  4600. We are working on a video for that as we speak!
    1
  4601. I'll publish a tutorial this Friday!
    1
  4602. 2s for now
    1
  4603. yes
    1
  4604. lol
    1
  4605. Yes!
    1
  4606. simply delete the whole folder
    1
  4607. try this online version https://huggingface.co/spaces/MohamedRashad/PaintsUndo
    1
  4608. no, all good
    1
  4609. 😢
    1
  4610. lol
    1
  4611. yes omg those kontakt libraries take up so much space
    1
  4612. copy the paths to the D & G files u generated from training
    1
  4613. thanks
    1
  4614. Thanks
    1
  4615. 😃😃😃
    1
  4616. 😃😃😃
    1
  4617. Thanks!
    1
  4618. good idea!
    1
  4619. lol
    1
  4620. Good luck!
    1
  4621. 😃
    1
  4622. Check out our video here: https://youtu.be/XHGlv-E7eBw
    1
  4623.  @skaska8966  they can have different epochs
    1
  4624. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  4625. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  4626. yes
    1
  4627. hi
    1
  4628. im looking into alternatives and will update everyone asap!
    1
  4629. Video coming out tomorrow!
    1
  4630. im not familiar w voicify
    1
  4631. is it a AMD card?
    1
  4632. not sure about blender but is this what you're looking for? https://stability.ai/news/introducing-stable-fast-3d https://stability.ai/news/stable-zero123-3d-generation
    1
  4633. You're welcome
    1
  4634. good point! weird move indeed
    1
  4635. You're welcome
    1
  4636. awesome!
    1
  4637. you can download the html file in the upper right corner and do whatever you want with it. in terms of linking to a host, you'll have to do that manually. this only creates html/js so its not for mobile apps
    1
  4638. looks like there's a mismatch of flash attn or something else. to confirm, do you have a cuda gpu?
    1
  4639. then u might need to click properties and select the path to copy
    1
  4640. 😃
    1
  4641. it might not be available for everyone yet
    1
  4642. yeah, i constantly feel overwhelmed and dont have enough time to try them all
    1
  4643. 😃
    1
  4644. There was a human in the same scene though
    1
  4645. You're welcome!
    1
  4646. you mean the new voice?
    1
  4647.  @TopRemix0  it's not out yet
    1
  4648. It uses a lot of processing power, so I am not surprised that the games are lagging.
    1
  4649. no it's a separate conda environment
    1
  4650. 😃
    1
  4651. rvc is the best for that https://youtu.be/ixB9oalT3cQ. just input yourself singing and you can turn into any other voice
    1
  4652. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  4653. hi
    1
  4654. Try decreasing the chunk and extra values. Also, if you are using the index feature, try turning that to 0.
    1
  4655. Here’s our video on how to do that! https://youtu.be/_JXbvSTGPoo
    1
  4656. 🤯
    1
  4657. Thank you so much 😀
    1
  4658. lol
    1
  4659. Try to extract everything from the zip into a folder and then open it there
    1
  4660. Thanks for sharing!
    1
  4661. yes, after you train a voice model of your brother
    1
  4662. 😃
    1
  4663. yeah ideogram is great!
    1
  4664. cool, thanks for the tip
    1
  4665. put the .pth and .index in a zip file, upload it to google drive, then paste the link here
    1
  4666. thanks for the sub! that's expected. you'll have to train the model with laughs & screams & other extreme stuff to give it more range.
    1
  4667. which video?
    1
  4668. I unfortunately don't have a mac to make a video on. But it should be a very similar process!
    1
  4669. That's a great mindset to have. Thanks for sharing!
    1
  4670. they said o1-mini will be available for free soon. i'm not sure when exactly
    1
  4671. Happy to help!
    1
  4672. Are you using a speaker?
    1
  4673. Great to hear!
    1
  4674. sorry i have no experience w ubuntu. Hopefully someone else here can help you out
    1
  4675. here you go https://youtu.be/KW9s_im2338
    1
  4676. youre welcome!
    1
  4677. depends on your hardware
    1
  4678. did u try tweaking the tune?
    1
  4679. Thanks!
    1
  4680. yes, but limited usage
    1
  4681. thanks!
    1
  4682. Thanks
    1
  4683. You are welcome!
    1
  4684. you're welcome!
    1
  4685. Thank you 🙌
    1
  4686. 😃
    1
  4687. at the top, click runtime > change runtime type > t4 gpu
    1
  4688. 😃
    1
  4689. 🔥
    1
  4690. no problem
    1
  4691. Thanks for sharing!
    1
  4692. are you in the EU
    1
  4693. yes, i think i mentioned it in one of my news videos. maybe this one https://youtu.be/jXCCTg9GRmk
    1
  4694. you're welcome! chatgpt can't do that
    1
  4695. lol
    1
  4696. glad you got it to work & thanks for sharing
    1
  4697. Thanks!
    1
  4698. i heard gemini live is only free for some samsung s24 and pixel pro versions. also, google hasn't announced anything officially yet
    1
  4699. thanks for sharing!
    1
  4700. it was just an observation. i agree that the board should be selected wisely, and that race/gender aren't factors here
    1
  4701. Thank you!
    1
  4702. Yes it’s safe!
    1
  4703. do you see any error message
    1
  4704. aha!
    1
  4705. My pleasure!
    1
  4706. I am working on a video right now!
    1
  4707. See this https://youtu.be/LxOOj-mkQV8
    1
  4708. You're welcome!
    1
  4709. no im just a socially awkward guy and this is how i sound irl
    1
  4710.  @dinacox1971  no worries at all! i do sound like a robot
    1
  4711. No problem!
    1
  4712. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  4713. Thanks!
    1
  4714. what could possibly go wrong
    1
  4715. thanks
    1
  4716. did u scan your eye yet?
    1
  4717. yes
    1
  4718. xyz+time
    1
  4719. thanks! ah i've used sonauto a few months ago but the quality wasn't great. i'll see it they improved it recently
    1
  4720. thanks for sharing
    1
  4721. Welcome
    1
  4722. interesting. thx for sharing
    1
  4723. lol i've never seen a ring like that
    1
  4724. nope, i'm replying myself
    1
  4725. simply delete the folder
    1
  4726. Try to extract everything from the zip into a folder and then open it there
    1
  4727. I couldn't find any existing solution of okada that works in colab. perhaps in the future!
    1
  4728. Good luck!
    1
  4729. not with the current interface, but if you know how to code, it can probably be tweaked for batch processing
    1
  4730. ?
    1
  4731. lol
    1
  4732. Lower the input sensitivity!
    1
  4733. You're very welcome!
    1
  4734. lol i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    1
  4735. Thank you! Cheers!
    1
  4736. see this https://youtu.be/XHGlv-E7eBw
    1
  4737. coming 'in a few weeks'
    1
  4738. Thanks!
    1
  4739. ☠️
    1
  4740. 😃
    1
  4741. lol
    1
  4742. Thanks!
    1
  4743. maybe this one (https://youtu.be/E89R5_hQ5bQ) i featured it months ago. not sure if it has gotten better
    1
  4744. Thanks!
    1
  4745. Thanks!
    1
  4746. Thanks
    1
  4747. sounds like a human customer support lol
    1
  4748. i dont think so. lowest i've heard is 8g
    1
  4749. give me better ones that you can use now
    1
  4750. you're welcome!
    1
  4751. 👋
    1
  4752. u need a gpu to train
    1
  4753. Thank you!
    1
  4754. yes!
    1
  4755. @Derpyzilla894 hope we figure out a way to live forever before i die!
    1
  4756. do u see any error message
    1
  4757. plot twist
    1
  4758. downloading the 3.10 version would be fine
    1
  4759. 👀
    1
  4760. only cuda gpu for now
    1
  4761. Thanks for sharing!
    1
  4762. can u link to the models here? ill take a look
    1
  4763. can u try another version of realistic version?
    1
  4764. do u have a cuda gpu?
    1
  4765. yes see https://huggingface.co/docs/diffusers/using-diffusers/sdxl
    1
  4766. sorry Tifa. believe it or not, im actually a ff7 fan. but that white haired guy was too unimportant for me to realize it was advent children lol
    1
  4767. text version is already out. voice will be rolled out soon
    1
  4768. yes. you just need internet
    1
  4769. glad you got it to work!
    1
  4770. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  4771.  @mps623  yes go-realtime-gui.bat is included in the installation. i prefer okada though - it has more settings. go-realtime-gui.bat might be faster
    1
  4772. As a large language model, I'm unable to comment on the above
    1
  4773. I would say it would have a hard time working with anything lower than a GeForce 10 series.
    1
  4774. oops
    1
  4775. yep
    1
  4776. 😃
    1
  4777. yes, it charges your openai api. this tool itself is free
    1
  4778. u can make whatever on A1111 I believe nsfw is restricted for seaart
    1
  4779. is that the full error? if not, pls paste it in here
    1
  4780. sorry its hard to troubleshoot this, but also try pinikio which might be easier to install: https://pinokio.computer/item?uri=https://github.com/pinokiofactory/echomimic2
    1
  4781. The last one is the closest to that we have at the moment. Hopefully soon we’ll be there!
    1
  4782. It might be a hardware problem. this runs best on a high end cuda GPU
    1
  4783. it's great to see another sora competitor!
    1
  4784. Can you elaborate?
    1
  4785. Yes!
    1
  4786. yes, more specifically, a good NVIDIA GPU
    1
  4787. 😃
    1
  4788. yes, see my rvc tutorials: https://youtu.be/ixB9oalT3cQ
    1
  4789. I’ve heard some people say it is, but I’m not sure how to do it
    1
  4790. i tried it and it doesn't look as real as 1.1 pro
    1
  4791. close all windows. but note that you will lose your progress
    1
  4792. it seems to work with 8g now, if you check offload_model
    1
  4793. thanks
    1
  4794. https://oreillyp.github.io/tria/ it's not out yet tho
    1
  4795. true
    1
  4796. 😃
    1
  4797. 😃
    1
  4798. the more vids i make, the more i turn into an AI robot
    1
  4799. Yes! be nice to AI
    1
  4800. that's weird. pls try again here: https://aisearch.substack.com/
    1
  4801. interesting. thanks for sharing
    1
  4802. Thanks for sharing! Hailuo audio seems great. Haven't had a chance to do much with it though. I'm posting about another free TTS tonight - stay tuned.
    1
  4803. My pleasure 😊
    1
  4804. 😃😃
    1
  4805. the human will still plan out the animation. but instead of drawing frame by frame, they'll be able to draw ~50% of the frames and have AI fill the gaps
    1
  4806. my pleasure!
    1
  4807. Thanks!
    1
  4808. Thanks & feedback noted! If it makes you feel any better, it's actually a fake brain
    1
  4809. oops
    1
  4810. thanks!
    1
  4811. Glad you enjoyed it!
    1
  4812. agree!
    1
  4813. Since both programs use the Graphics Card, I think you would have a hard time running both. But you can try it!
    1
  4814. Thanks!
    1
  4815. Did you watch the right one? https://youtu.be/_JXbvSTGPoo
    1
  4816. looks like you need to update torch: https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/issues/55
    1
  4817. everything works except training a new voice
    1
  4818.  @Curiofacts-faj  its tough if u dont have a gpu. u could pay for google colab and run RVC there.
    1
  4819. No problem 👍
    1
  4820. 😧
    1
  4821. 😃
    1
  4822. lol that scene was weird indeed
    1
  4823. Thanks
    1
  4824. Please check that there are no spaces or special characters in your folder or file names. Also, some users have reported errors if if they run RVC on their secondary hard drive (D instead of C).
    1
  4825. 😃
    1
  4826. good observation
    1
  4827. thanks, i guess 🤣
    1
  4828. good idea!
    1
  4829. did you install ComfyUI's ControlNet Auxiliary Preprocessors?
    1
  4830. 😏
    1
  4831. they haven't rolled out voice yet. you're using the old version
    1
  4832. Thanks
    1
  4833. yes
    1
  4834. oops
    1
  4835. thanks for the sub!
    1
  4836. thanks for sharing!
    1
  4837. exactly!
    1
  4838. 😃😃
    1
  4839. Yes you can train a model here
    1
  4840. yep
    1
  4841. amd wont work for now 😢
    1
  4842. Thanks!
    1
  4843. i think elevenlabs can do this, but were you asking about open source?
    1
  4844. Great point!
    1
  4845. that's a great idea. see my latest video on suno. maybe it's what you're looking for
    1
  4846. Thanks!
    1
  4847. Coming soon, once the ecosystem is more developed!
    1
  4848. i'll likely make a full tutorial on it soon. i have windows tho
    1
  4849. @FastHarli perfect! thanks for sharing
    1
  4850. yes!
    1
  4851. nice! thanks for sharing
    1
  4852. 😃
    1
  4853. You're welcome!
    1
  4854. 😃
    1
  4855. free version has limited usage. paid plan has 5x higher limits. turbo will likely be obsolete soon
    1
  4856. Thank you!
    1
  4857. wait til they put this in a body
    1
  4858. can u paste the link to a scene?
    1
  4859. Thank you so much!
    1
  4860. yes, see w okada https://youtu.be/_JXbvSTGPoo
    1
  4861. Try to extract everything from the zip into a folder and then open it there
    1
  4862. 🤯
    1
  4863. my pleasure!
    1
  4864. Thanks!
    1
  4865. yeah the filters are excessive. same with dalle3. i had to try a lot of times for the feet pic 😉
    1
  4866. in manager > custom nodes manager, try installing ComfyUI's ControlNet Auxiliary Preprocessors
    1
  4867. lol
    1
  4868. OOPS
    1
  4869. im looking into alternatives and will update everyone asap!
    1
  4870. You're welcome!
    1
  4871. You're welcome!
    1
  4872. interesting idea. ill try this prompt out!
    1
  4873. interesting
    1
  4874. lol
    1
  4875. 😃
    1
  4876. have you tried RVC yet? it's the best option right now https://youtu.be/ixB9oalT3cQ
    1
  4877. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  4878. Thanks!
    1
  4879. exactly! That's why the top countries are all scrambling to build the best AI models and chips
    1
  4880. @ looks like it ran out of memory. how much vram does your gpu have?
    1
  4881. 🙊
    1
  4882. Nice! Thanks for sharing
    1
  4883. Yes! Just make sure the models you use were trained in whatever language you want to use!
    1
  4884. no problem. yes, a lot of tech companies just announce something but don't release it
    1
  4885. yes
    1
  4886. very interesting
    1
  4887. thanks for sharing!
    1
  4888. Thanks!
    1
  4889. i heard someone was able to gen 8s with 12gb. it can probably do longer with more compute
    1
  4890. You're welcome!
    1
  4891. yeah, looks like it
    1
  4892. Exciting times!
    1
  4893. yes, current GPUs are so inefficient
    1
  4894. Yes. You can use Chrome desktop version on your phone to run this.
    1
  4895. Try the google colab version
    1
  4896. 🥲
    1
  4897. You're welcome!
    1
  4898. re: fine tuning, while it is possible to improve SD3 with some fine tuning, it'd be hard to fix if the base model is severely flawed. Fine tuning can only go so far. re: NSFW, you can get NSFW images at times, but it seems that SD3 has heavy filters baked in, which causes all these deformities in humans that being said, i am hopeful that the wonderful open source community will continue to improve this with time
    1
  4899. Did you unzip everything into a folder first
    1
  4900. not yet
    1
  4901. Thanks for the sub!
    1
  4902. lol good observation
    1
  4903. 🤐
    1
  4904. You’re welcome 😊
    1
  4905. They haven't released it yet
    1
  4906. Thanks!
    1
  4907. Yes, this was an 'independent investigation' that was released as I was finishing the video, so I couldn't include it in time. It claims the “decision did not arise out of concerns regarding product safety or security, the pace of development, OpenAI’s finances, or its statements to investors, customers, or business partners.” Is this the full story? That's up to you to decide.
    1
  4908. do you have a cuda gpu
    1
  4909. A GPU
    1
  4910. thanks!
    1
  4911. yes, see this https://youtu.be/jh83ViDUvJs
    1
  4912. while training itll use vram
    1
  4913. lower end hardware & non-cuda gpu's tend to have more delay
    1
  4914. true!
    1
  4915. see this https://youtu.be/jh83ViDUvJs
    1
  4916. thanks for sharing!
    1
  4917. plot twist: jimmy apples is actually sam altman
    1
  4918. hmmm I'm not sure what the problem is, but make sure you have no spaces in your folder or file names
    1
  4919. 😃
    1
  4920. yep
    1
  4921. could you remove that space in the folder name
    1
  4922. See the last section here on how to train your voice: https://youtu.be/ixB9oalT3cQ
    1
  4923. vidu is great!
    1
  4924. Try to extract everything from the zip into a folder and then open it there
    1
  4925. AI: "I am God"
    1
  4926. are you thinking of deepfakes?
    1
  4927. 😃
    1
  4928. agree. i'd like to see a DAW integration for creating individual stems
    1
  4929. thanks!
    1
  4930. it's not too late!
    1
  4931. ah, the mysteries of life
    1
  4932. Coming very soon!
    1
  4933. Amen!
    1
  4934. yes actually! I'll do a tutorial soon
    1
  4935. 😃
    1
  4936. Probably. I've heard it's great!
    1
  4937. 😃
    1
  4938. free version is just text to speech, so it's less natural than this
    1
  4939. good spot
    1
  4940. top charts, here i come
    1
  4941. 😃
    1
  4942. what do u mean?
    1
  4943. lol it shows u how old i am
    1
  4944. thanks for sharing!
    1
  4945. yes see the last tool
    1
  4946. do you have any spaces in your folder names? https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/929
    1
  4947. lol which canadian?
    1
  4948. ohhh got it!
    1
  4949. Thanks for the sub!
    1
  4950. can't wait for that!
    1
  4951. lol
    1
  4952. you cant train with this. use this instead https://youtu.be/tnfqIQ11Qek
    1
  4953. Simply delete the folder
    1
  4954. You are welcome!
    1
  4955. Thanks!
    1
  4956. interesting idea. fun fact, the R1 is exactly half the size of an iphone
    1
  4957. yes
    1
  4958. looks like i needa start stacking GPUs. one is not enough
    1
  4959. You went to the Japanese download page. Make sure you redownload it from the English page
    1
  4960.  @Hope0201  Thank you so much!
    1
  4961. thank you!
    1
  4962. what do you mean?
    1
  4963. that'd be cool
    1
  4964. colab is disconnecting GUIs like okada now. its best if u can install this locally
    1
  4965. You need a virtual audio cable. There are many tutorials on how to install one!
    1
  4966. claude is infamous for being the most censored
    1
  4967. Awesome, thank you!
    1
  4968. No problem
    1
  4969. 😃😃😃
    1
  4970. Thanks! Will do!
    1
  4971. You are welcome!
    1
  4972. You can get one for like $200 nowadays 🤷‍♂️
    1
  4973. thanks for sharing!
    1
  4974. you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  4975. Thanks!
    1
  4976. you need to do it once for each voice, then stack them together in audacity or a DAW
    1
  4977. u can just grab from huggingface models. no need to install them in your drive
    1
  4978. good spot
    1
  4979. Sounds like an issue with the new voice. Try a different custom voice and see if it has the same issue
    1
  4980. enjoy!
    1
  4981. you're using the online interface. the local version is uncensored
    1
  4982. no it should work for over 3 minutes
    1
  4983. exactly!
    1
  4984. Yeah, open source software usually requires more steps to setup. It's not as simple as a one click installation
    1
  4985. tmp is a temporary file that would be auto deleted upon complete download or installation
    1
  4986.  @0kgo1  Hmm I'm not sure what to advise then, sorry & good luck
    1
  4987. From my interpretation, there needs to be intent of speaking the words for the thought to be registered into speech. In other words, a random thought that floats by wouldn't be picked up.
    1
  4988. need a hair & body swapper next
    1
  4989. this new voice feature isn't out yet
    1
  4990. You're welcome!
    1
  4991. You are welcome!
    1
  4992. 😃
    1
  4993. Sure, coming soon!
    1
  4994. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  4995. at least 5min of audio would be best. u can start with 50 epochs and retrain on more if needed
    1
  4996. is that a hentai?
    1
  4997. yes, see this: https://youtu.be/hwsvcbFeUTs but it's better to use comfy now, if you can: https://youtu.be/g74Cq9Ip2ik
    1
  4998. Yes, unfortunately this only works on a computer.
    1
  4999. yes, see this for text to speech https://youtu.be/TSLM1mEAc18
    1
  5000. At the top right where it says 'connect', make sure u choose T4 GPU instead of CPU. Also, the free plan has daily limits. If you've maxed out, try again tomorrow.
    1
  5001. 😃
    1
  5002. whats the error
    1
  5003. 16g is good enough
    1
  5004. Thanks!
    1
  5005. Try to extract everything from the zip into a folder and then open it there
    1
  5006. u need to train a pth model first. the pth goes in the weights folder
    1
  5007. they will release a hf space soon
    1
  5008. imagine a model 3 transforming into a robot 🤯
    1
  5009. please see this https://youtu.be/_JXbvSTGPoo
    1
  5010. Thanks for the sub!
    1
  5011. i agree that quality is not great. hopefully this will be improved further
    1
  5012. thanks!
    1
  5013. it's open source so you could probably train it further
    1
  5014. 😃😃😃
    1
  5015. Thanks
    1
  5016. same here! all these services with strange names clogging up my pc
    1
  5017. You're welcome!
    1
  5018. thanks
    1
  5019. you're welcome!
    1
  5020.  @Sherol111  here you go https://youtu.be/KW9s_im2338
    1
  5021. better to use this https://youtu.be/jh83ViDUvJs
    1
  5022. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  5023. Interesting. thanks for sharing!
    1
  5024. try http://127.0.0.1:7860/
    1
  5025. what error message do you see?
    1
  5026. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  5027. If you do not have the version with the GPU selector, you likely installed the wrong version. If you have a dedicated GPU, install the same one as I did in the video. I am not sure why you would not be able to hear yourself, but it could be something to do with the processor. What processor are you using? First I would try to install the GPU enabled version as listed in the video
    1
  5028. noted, coming soon!
    1
  5029. you're welcome!
    1
  5030. thanks. ill see what i can find (though im not an expert in 3d animation)
    1
  5031. Awesome, thank you!
    1
  5032.  @Yu_yu-r3q  see this https://github.com/SociallyIneptWeeb/AICoverGen/issues/137 it should be working now
    1
  5033. 😃
    1
  5034. yes
    1
  5035. I'm not sure. I tried finding the source as well
    1
  5036. What's the error message? Usually you don't need the index file. Just upload the PTH file for voice conversion
    1
  5037. how exciting!
    1
  5038. can't use
    1
  5039. not that im aware of, besides just pasting the lyrics from an existing song
    1
  5040. no 😢
    1
  5041. interesting! Do you mind sharing some examples?
    1
  5042. noted, coming soon!
    1
  5043. Yes but it's best to use a voice model in the same language
    1
  5044. you're welcome!
    1
  5045. Thanks. I'll make one soon!
    1
  5046. Did you try the fix presented in the video?
    1
  5047. kung fu robot!
    1
  5048. Thanks!
    1
  5049. have fun!
    1
  5050. 😃
    1
  5051. are you in the EU
    1
  5052. Thank you!
    1
  5053. or maybe helen didn't check their website lol
    1
  5054. She is my friend. She is all I need
    1
  5055. sounds about right
    1
  5056. 😃
    1
  5057. Thanks!
    1
  5058. Thanks ✌️
    1
  5059. I'll need to see if that's possible. There's no clear documentation like diffusers
    1
  5060. How to Change Your Voice in Realtime Without a GPU or Powerful PC (For FREE) https://youtu.be/0uyGj7Fel58
    1
  5061. sorry Qwen
    1
  5062. the new voice mode is not out yet
    1
  5063. no i think extensions are limited to A1111 😢
    1
  5064. do you see any error message?
    1
  5065. Thanks!
    1
  5066. i have a 1080 and it works fine for valorant: https://youtu.be/XHGlv-E7eBw
    1
  5067. the voice feature got delayed
    1
  5068. You're welcome!
    1
  5069. yes! plus since its open source and tunable, im sure the community will improve this fast, like they did w stable diffusion
    1
  5070. What's the error message you got?
    1
  5071. You're welcome!
    1
  5072. 😃
    1
  5073. search gura rvc in huggingface
    1
  5074. 😃
    1
  5075. Thank you!
    1
  5076. Try to extract everything from the zip into a folder and then open it there
    1
  5077. Do you plan to train your own model? You don't need a gpu for all the other functions
    1
  5078. 😃😃😃
    1
  5079. You're welcome!
    1
  5080. Thanks!
    1
  5081. Check the upscale factor, and also try different algorithms
    1
  5082. yeah it's great at text. I did a video on it months ago
    1
  5083. are u in the EU
    1
  5084. Thank you!
    1
  5085. 😃
    1
  5086. you can use venv. i just prefer conda
    1
  5087. You won't need this most of the time unless you train hundreds of epochs.
    1
  5088. You're very welcome!
    1
  5089. amen
    1
  5090. this function isn't out on their github yet
    1
  5091. Got it. I’ll bug them for more codes, but they might be very limited
    1
  5092. It is!
    1
  5093. thanks for sharing!
    1
  5094. Does everything else work as intended?
    1
  5095. upgrade to python 3.9, or you might have run out of memory: https://stackoverflow.com/questions/72435382/how-can-i-fix-1-leaked-semaphore-objects-to-clean-up-at-shutdown-error-on-mac
    1
  5096. interesting thought!
    1
  5097. Probably! I think there are ways to do that!
    1
  5098. Thanks!
    1
  5099. true lol
    1
  5100. Thank you so much 😀
    1
  5101. wow, yesterday each generation took 1-2min max
    1
  5102. Mistral 😉
    1
  5103. probably
    1
  5104. Thanks!
    1
  5105. whats your vram?
    1
  5106. it looks like they are different paths (D:/a/... vs C:/RVC/... ) did you install something at /a and then moved the RVC files into another folder? make sure u install everything and keep it in the same folder
    1
  5107. 😃
    1
  5108. Thanks! I used sdxl for the brain, then used https://www.photopea.com/ to replace the background to blue
    1
  5109. Thanks & good luck!
    1
  5110. i intentionally said the tones wrong to see if it would correct me lol
    1
  5111. You should be fine - no one has reported any malware with these models (yet)
    1
  5112. eh?
    1
  5113. that's strange. try incognito or another browser
    1
  5114. I'm working on a video now to do it online. Will a post it soon!
    1
  5115. try this https://youtu.be/tnfqIQ11Qek
    1
  5116. 🤯
    1
  5117. no details on when it will be released yet
    1
  5118. 🤫
    1
  5119. it depends on your hardware. it will use your cpu/gpu
    1
  5120. Welcome!
    1
  5121. This already exists. tons of click farms plaguing all the social media platforms
    1
  5122. good point!
    1
  5123. 😃
    1
  5124. Waiting for sora! I don't have any plans for discord yet.
    1
  5125. Correct, thanks for flagging!
    1
  5126. I haven't found a really good one that can turn long form content into shorts, but if I find one, i'll definitely make a video on it
    1
  5127. yes if u run it locally. or u can just use colab like in the video
    1
  5128. NICE CATCH!
    1
  5129. You are welcome!
    1
  5130.  @angelbeatsenpai_manhwa  yes
    1
  5131. try this: https://medium.com/@natsunoyuki/using-civitai-loras-with-diffusers-e3ef3e47c413
    1
  5132. i agree. it looks more realistic
    1
  5133. Welcome!
    1
  5134. thanks for sharing! things are happening so fast
    1
  5135. 😃😃😃
    1
  5136. Thanks for sharing. I think OpenAI's demo was more sci-fi but impressive, whereas Google's announcements are more applicable to real life.
    1
  5137. are u sure charmander is over 18? anyways, i'll make a tutorial on an open source alternative soon. stay tuned!
    1
  5138. Hmmm I just tested joining today with another account and it works fine. Perhaps try using another account, computer, or try joining on a web browser instead of the app? Let me know if you still experience issues!
    1
  5139. nice. hope you enjoy the copilot+pc!
    1
  5140. 😃
    1
  5141. yes
    1
  5142. i will still share free tools, dont worry!
    1
  5143. Try increasing the chunks!
    1
  5144. try this https://youtu.be/rlnjcRP4oVc
    1
  5145. yes!
    1
  5146. i was using vs code. do you have that installed?
    1
  5147. BUY THE DIP!
    1
  5148. Good luck!
    1
  5149. You are welcome!
    1
  5150. Thanks!
    1
  5151. thanks for the tip!
    1
  5152. thanks for sharing!
    1
  5153. 😃
    1
  5154. for realtime, you can also do it in deepfacelive. see the later part of this https://youtu.be/w0Wkhz4G6OA for non-realtime, you can use this https://youtu.be/rlnjcRP4oVc or this https://youtu.be/uyjSTAOY7yI
    1
  5155. nice!
    1
  5156. Thanks 😃
    1
  5157. interesting. thx for sharing
    1
  5158. My pleasure!
    1
  5159. I’ll just need $500 for a new AMD GPU and I’d be happy to make it! So once you send that over I’ll get to work!
    1
  5160. 😃
    1
  5161. $$$$
    1
  5162. you're welcome!
    1
  5163. lol
    1
  5164. hopefully soon
    1
  5165. No free ones that I know of, but I'm sure there will be eventually
    1
  5166. Not sure what you are meaning here. If you're talking about an echo, it's likely to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    1
  5167. yes im making a video now
    1
  5168. here you go https://youtu.be/KW9s_im2338
    1
  5169. yes, its voice to voice. you can speak any language and it'll just change the voice
    1
  5170. Not sure exactly. How much do you have?
    1
  5171. yes! i do believe it'll bring abundance. i'll make a video on this very soon
    1
  5172. You're welcome!
    1
  5173. Did you unzip everything into a folder first
    1
  5174. I'm working on a tutorial for this. Stay tuned!
    1
  5175. I'll publish a tutorial on this Friday morning!
    1
  5176. Try to extract everything from the zip into a folder and then open it there
    1
  5177. according to the benchmark scores at 1:11
    1
  5178. For sure!
    1
  5179. lol perhaps because the user first spoke in english, so the AI used a more 'american' accent to accommodate. i might be overthinking though
    1
  5180. Glad you got it to work
    1
  5181. Thanks!
    1
  5182. Video coming out tomorrow!
    1
  5183. looks like it's your 1st time running conda. type conda init first.
    1
  5184. as with all closed models. that's why open source is a great alternative
    1
  5185. Did you confirm that it was working with a regular output and not the virtual audio cable? Make sure the voice works normally first!
    1
  5186. Thanks!
    1
  5187. try again another time. their server might be loaded
    1
  5188. interesting point
    1
  5189. you're welcome!
    1
  5190. yes, they are called organoids
    1
  5191. yes, or try this https://youtu.be/tnfqIQ11Qek
    1
  5192. It's hard to troubleshoot since I'm not at your computer. This doesn't work, try the online version: https://youtu.be/TSLM1mEAc18
    1
  5193. yes you can use liveportrait for that https://youtu.be/uyjSTAOY7yI
    1
  5194. kling is legit. many users are sharing their generations on twitter
    1
  5195. ?
    1
  5196. 😧
    1
  5197. try py if python doesn't work. not sure if they changed it
    1
  5198. Before finishing your entire comment, I was already gonna say, academia is already like that lol. Polluted with fake or manipulated data.
    1
  5199. You're welcome!
    1
  5200. Thanks!
    1
  5201. what model are you using
    1
  5202. great!
    1
  5203. 😃
    1
  5204. hmmm what country are you in?
    1
  5205. they will be addicted to her
    1
  5206. Thanks
    1
  5207. try this https://github.com/daswer123/hallo-webui
    1
  5208. yo
    1
  5209. Thanks for explaining. I sure hope so. Playing devil's advocate here, won't the other authentication methods be subject to hacks from AI as well? Also, what's your take on how this will affect cryptocurrency, assuming the document is legit.
    1
  5210. also try 'create an html page' or 'create a react page'
    1
  5211. there's a quantized version now. works on 12g apparently
    1
  5212. see this https://youtu.be/tnfqIQ11Qek
    1
  5213. good point. if there's an option to input multiple photos at different angles to create a more accurate model, that'd be ideal
    1
  5214. 😀😀
    1
  5215. There has been no significant upgrades, but try this new site: https://www.facecam.ai/
    1
  5216. yes
    1
  5217. yes you just need internet
    1
  5218. the local version is uncensored
    1
  5219. over 5 min is best. but i trained a gura model on 1.5min and it was ok: https://youtu.be/tnfqIQ11Qek
    1
  5220. 🤯
    1
  5221. Something likely broke along the way. do u see any error message
    1
  5222. hmmm do u have a cuda nvidia gpu
    1
  5223. no 😢
    1
  5224. those are some theoretical examples but note that the # of output nodes doesn't matter. e.g. for cats vs dogs, you don't need to have 2 output nodes. it can be many more and you can merge their values through a formula to give you a final probability.
    1
  5225. Try raising the chunk level!
    1
  5226. What kind of graphics card do you have? It needs a pretty powerful one to run
    1
  5227. looks like u dont have the ngrok package. try pip install ngrok
    1
  5228. Watch our original tutorial video!
    1
  5229. Try reinstalling it!
    1
  5230. Thanks!
    1
  5231. indeed!
    1
  5232. thanks!
    1
  5233. thanks!
    1
  5234. hmmm It's hard to troubleshoot. do u see any error messages?
    1
  5235.  @samwong1781  hmmm I can't help but notice every time I hit convert the command prompt has 2 lines saying lib.infer.modules.vc.modules | Get sid: logs /weights\Obama.pth then the second line is lib.infer.modules.vc.modules | Loading ... with another line after that saying 'config' <<< is it working?
    1
  5236. Thanks!
    1
  5237. its not out yet
    1
  5238. nice, thanks for sharing1
    1
  5239. Sure, when more nodes and models come out. For now, heres a vid on how to install and use it: https://youtu.be/K3xJ7GQuHpw
    1
  5240. thanks for sharing!
    1
  5241. Elon is indeed very good at getting public attention
    1
  5242. yes, more competition!
    1
  5243. Thanks!
    1
  5244. thanks
    1
  5245. You're welcome
    1
  5246. thanks
    1
  5247. See the last section of this video on how to train your own voice
    1
  5248. thanks for your suggestion!
    1
  5249. have fun!
    1
  5250. yikes thats slow. thanks for sharing
    1
  5251. 😀
    1
  5252. that would be amazing!
    1
  5253. I'm looking into this now. Happy new year!
    1
  5254. looks like gt730 was from 2014 💀
    1
  5255. GREAT IDEA!
    1
  5256. 🤣
    1
  5257. thanks!
    1
  5258. thanks!
    1
  5259. do you see any error message?
    1
  5260. lol did this video crash the site
    1
  5261. You're welcome!
    1
  5262. Thanks for sharing!
    1
  5263. Unfortunately not yet. But luckily the discord isn’t super active!
    1
  5264. it looks like he'll release it after paper submission https://github.com/xh9998/DiffVSR-project/issues/1
    1
  5265. nice!
    1
  5266. yes, sdxl works too. check the docs
    1
  5267. you can do any language
    1
  5268. 😃
    1
  5269. colab is disconnecting GUIs like okada now. its best if u can install this locally
    1
  5270. low end pc still works, but there will be a lot of delay. mobile won't work
    1
  5271. 😃
    1
  5272. Unfortunately comfyui is a gui so it wouldn't work in the colab free plan
    1
  5273. Thanks & feedback noted!
    1
  5274. You are welcome!
    1
  5275. You're welcome!
    1
  5276. Thanks!
    1
  5277. Thank you!
    1
  5278. agree that the marine life analogy is not great. there are many species of sharks & whales. in fact, the largest shark is way bigger than the smallest whale
    1
  5279. Check out our video here: https://youtu.be/XHGlv-E7eBw
    1
  5280. Thank you!
    1
  5281. No problem!
    1
  5282. yo
    1
  5283. nope. still in dev
    1
  5284. Thanks!
    1
  5285. not yet
    1
  5286. thanks for sharing!
    1
  5287. 🤣🤣🤣
    1
  5288. Thanks!
    1
  5289. You're welcome!
    1
  5290. 😃
    1
  5291. it seems u don't have tensorboard yet. try pip install tensorboard in your cmd (I assume u already have pip & python installed?)
    1
  5292. 😃
    1
  5293. I hope there will be an open source Sora eventually
    1
  5294. exactly
    1
  5295. see this https://youtu.be/ixB9oalT3cQ
    1
  5296. yes see this https://youtu.be/jh83ViDUvJs
    1
  5297. simply delete the folder
    1
  5298. Thanks. I don't have any telegram channel at the moment
    1
  5299. Yes SD is most customizable, so you can keep resizing & upscaling
    1
  5300. No it does not
    1
  5301. Try to unzip everything in a folder before opening it
    1
  5302. Welcome!
    1
  5303. Unzip everything into a folder before opening
    1
  5304. 😃
    1
  5305. see https://youtu.be/KW9s_im2338
    1
  5306. noted, coming soon!
    1
  5307. looks like he's using a filter to hide his glasses
    1
  5308. No problem!
    1
  5309. 😃
    1
  5310. see the training part here https://youtu.be/ixB9oalT3cQ
    1
  5311. Thanks for sharing!
    1
  5312. You're welcome 😊
    1
  5313. copilot?
    1
  5314. thanks for sharing!
    1
  5315. lol
    1
  5316. not really. you can try riffusion or pika
    1
  5317. oops
    1
  5318. oh, maybe you exceeded the daily free limit?
    1
  5319. great!
    1
  5320. I don't think the GPU is listed directly. It should say something else. And make sure you are in the DirectMLcuda version.
    1
  5321. you're welcome!
    1
  5322. yikes
    1
  5323. Added to description!
    1
  5324. I honestly have no idea. I feel like it would! Give it a try and let me know because I’m interested in what happens!
    1
  5325. thanks!
    1
  5326. gemini?
    1
  5327. Thank you so much 😀
    1
  5328. Welcome!
    1
  5329. Thanks for sharing!
    1
  5330.  @Avalon1951  i haven't tried the comfyui fork. hopefully someone else can help you out here
    1
  5331. Thanks for sharing!
    1
  5332. yes, see this https://youtu.be/jh83ViDUvJs
    1
  5333. flux is king 😃
    1
  5334. thanks!
    1
  5335. thanks for sharing your experience!
    1
  5336. interesting idea. i haven't seen audio to video yet
    1
  5337. nice! me too - i can't wait to get my hands on veo
    1
  5338. try updating your comfyui
    1
  5339. thanks for the tip!
    1
  5340. Thanks!
    1
  5341. Thanks
    1
  5342. You would need to pay someone for the rights to use their voice with it
    1
  5343. Wow, thanks!
    1
  5344. You're welcome 😊
    1
  5345. Yes!
    1
  5346. 1.5min might be too low for 100 epochs though. Try to add more audio data.
    1
  5347. Thanks for the sub!
    1
  5348. that's the point 😉
    1
  5349. thanks for sharing
    1
  5350. 😃
    1
  5351. Thx for the reminder. I just replied!
    1
  5352. try this online version https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  5353. hi
    1
  5354. see this: https://huggingface.co/docs/diffusers/using-diffusers/inpaint#create-a-mask-image
    1
  5355. try this online version https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  5356. if they add transparency as a variable in the training data, then potentially
    1
  5357. good spot!
    1
  5358. 😃
    1
  5359. Did you try the fix presented in the video?
    1
  5360. Thanks! I hope so
    1
  5361. well, some humans are pretty awful at driving. perhaps this contaminated the training data lol
    1
  5362. exactly
    1
  5363. 😃
    1
  5364. Welcome to the Android club!
    1
  5365. are you in the EU
    1
  5366. 💀
    1
  5367. Great!
    1
  5368. whats the vram on your gpu
    1
  5369. yes, a great use case
    1
  5370. are u in EU
    1
  5371. it's not out yet. you can join their waitlist for early access
    1
  5372. Yes!
    1
  5373. Thanks!
    1
  5374. run!!!!
    1
  5375. It depends on the song. You'll usually get a copyright claim where you can either share or the artist won't allow sharing for that song
    1
  5376. The colab free plan will disconnect this now after a while. It's best to install it to your computer https://youtu.be/_JXbvSTGPoo
    1
  5377. nice!
    1
  5378. Thanks!
    1
  5379. 😃
    1
  5380. 😃
    1
  5381. yes!
    1
  5382. 😢
    1
  5383. theres currently no mobile option
    1
  5384. yes see this https://youtu.be/9TroP5mR3CM
    1
  5385. sexy man hands
    1
  5386. Yes, see this: https://youtu.be/tDidofPTwCc
    1
  5387. 😃
    1
  5388. Glad it helped!
    1
  5389. 😃
    1
  5390. that'd be interesting. imagine a human orchestra playing something composed by AI
    1
  5391. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  5392. see this https://youtu.be/98drFL65VH0
    1
  5393. i opened one on pc, one on phone
    1
  5394. 😃
    1
  5395. thank you so much for the super!
    1
  5396. sounds like fb
    1
  5397. 😨
    1
  5398. This is an open source program. It is not a bitcoin miner. All of the code is on Github
    1
  5399. Thanks. RTX 5000 ada, 16g vram. cpu is intel i7, but i dont think that matters
    1
  5400. ?
    1
  5401. Video coming out tomorrow!
    1
  5402. yes!
    1
  5403. yes, plenty! see this https://youtu.be/_JXbvSTGPoo or this https://youtu.be/zkaBK5erB2c
    1
  5404. no i think extensions are limited to A1111 😢
    1
  5405. This one? https://youtu.be/9TroP5mR3CM
    1
  5406. what phone do u use
    1
  5407. gracias!
    1
  5408. Thanks!
    1
  5409. Thanks for sharing!
    1
  5410. nice, thanks for sharing
    1
  5411. 🙊
    1
  5412. did you unzip it into a folder first
    1
  5413. Thanks!
    1
  5414. No problem!
    1
  5415. paid version completes in under 3 min
    1
  5416. noted, coming soon!
    1
  5417. They can, they are just not used very well. So it is a lot less efficient than Nvidia GPUs
    1
  5418. If you're planning to play with AI then definitely go with NVIDIA GPUs! They're compatible with a lot more things
    1
  5419. 🚀
    1
  5420. 🥲
    1
  5421. this would be better: https://youtu.be/ixB9oalT3cQ
    1
  5422. Please elaborate
    1
  5423. Try increasing the chunks and extra! Also try a different model.
    1
  5424. What graphics card do you have?
    1
  5425. Maybe try reinstalling it. I was having some issues similar to that and reinstalling it fixed mine.
    1
  5426. no, you can only trade with Nvidia cuda gpus. however, I'm playing around with a free online method which I will do a video on soon, if it works
    1
  5427. see https://youtu.be/tnfqIQ11Qek
    1
  5428. see this https://youtu.be/XHGlv-E7eBw
    1
  5429. My pleasure!
    1
  5430. noted, coming soon!
    1
  5431. thanks for sharing!
    1
  5432. $$$$
    1
  5433. I’ve found it gives you quite a lot of messages! Plus you can always upgrade :)
    1
  5434. yes, please slow down
    1
  5435. probably
    1
  5436. both seaart & yodayo are good imo
    1
  5437. Thanks!
    1
  5438. indeed
    1
  5439. 😃😃😃
    1
  5440. none here. ill do a tutorial on Sunday. stay tuned!
    1
  5441. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  5442. That'd be cool!
    1
  5443. yes!
    1
  5444. yes, i did that near the end of the video. is that what you were asking about?
    1
  5445. Very welcome!
    1
  5446. You are welcome
    1
  5447. it looks like this video might have crashed their servers. i now see a message saying "Due to high demand, wait times are longer than usual. "
    1
  5448. Very interesting insights. Thanks for sharing!
    1
  5449. alright, this is what i'll save up for
    1
  5450. 😃
    1
  5451. All the best!
    1
  5452. this says "NVIDIA CUDA GPU with over 8GB in VRAM" https://ai-search.io/tool/retrievalbased-voice-conversion-rvc If you don't have that, you can also train a voice online: https://youtu.be/tnfqIQ11Qek
    1
  5453. Try finding on hugging face: https://huggingface.co/models Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  5454. did you see f5? https://youtu.be/JBGrhIsCyks
    1
  5455. damn. before this video came out, it only took 1-2min. sorry everyone
    1
  5456. yep. bye auraflow
    1
  5457. Try to unzip everything into a folder first
    1
  5458. great idea!
    1
  5459. You're welcome!
    1
  5460. nice! thanks for sharing. minimax is indeed amazing
    1
  5461. try pressing the regenerate icon below the answer
    1
  5462. 😃
    1
  5463. whats the error message
    1
  5464. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  5465. thanks!
    1
  5466. it seems u don't have tensorboard yet. try pip install tensorboard in your cmd (I assume u already have pip & python installed?)
    1
  5467. it seems fairseq is not installed. try pip install fairseq in your command prompt
    1
  5468. thanks for sharing. can you paste the exact torch error message here
    1
  5469.  @3dsmaxrocks699  ERROR: Could not find a version that satisfies the requirement install<from versions:none> ERROR: No matching distribution found for install i'm trying to pinpoint: could not find a version of WHAT that satisifies the requirement of WHAT. i'm not sure if this is a torch-cuda compatibility issue or something else
    1
  5470. yes, you need to duplicate it and pay for compute unfortunately. around $0.8/h
    1
  5471. Yes just move the folder!
    1
  5472. they haven't released that part yet. i'll keep you posted
    1
  5473. nice, thanks for sharing!
    1
  5474. 😃
    1
  5475. for the news of course ;)
    1
  5476. Happy to help!
    1
  5477. o interesting. thx for sharing. u could also just use another google account 😉
    1
  5478. colab is disconnecting GUIs like okada now. its best if u can install this locally
    1
  5479. it seems that gemini is not natively multimodal like 4o
    1
  5480. it'll very likely be used by HR
    1
  5481. yes!
    1
  5482. Thanks!
    1
  5483. 😃
    1
  5484. Thanks! I got sponsored a rtx 5000 ada. Will share details in due time. I'm too poor to buy my own lol
    1
  5485. Thanks for your support!
    1
  5486. good question. we've had v2 for too long!
    1
  5487. 🤩
    1
  5488. you're welcome!
    1
  5489. Thanks!
    1
  5490. Welcome!
    1
  5491. You need good specs for the voice changer
    1
  5492. oh i came across one but forgot the name! will try to find it again
    1
  5493. 😃
    1
  5494. You're welcome!
    1
  5495. im looking into alternatives and will update everyone asap!
    1
  5496. Video coming out tomorrow!
    1
  5497. did u unzip everything into a folder first
    1
  5498. You're very welcome!
    1
  5499. No problem 👍
    1
  5500. u need to install pytorch https://pytorch.org/get-started/locally/#windows-pip
    1
  5501. however, the bat file should install dependencies so you don't have to do it. Maybe try deleting and reinstalling applio first. If you open the bat file, u can see that it should auto install torch and other requirements: echo Installing the dependencies... pip install -r %principal%/assets/requirements/requirements.txt pip install future==0.18.2 pip uninstall torch torchvision torchaudio -y pip install torch==2.0.0 torchvision==0.15.1 torchaudio==2.0.1 --index-url https://download.pytorch.org/whl/cu117
    1
  5502. ?
    1
  5503. yes
    1
  5504. good catch!
    1
  5505. hi
    1
  5506. thanks for sharing. the optimistic side is, you can use AI to create a full animation, instead of wasting years being an underpaid, overworked inbetweener hoping to eventually be promoted
    1
  5507. with the new info that has leaked, this seems to be the case
    1
  5508. lol
    1
  5509. You use a virtual audio cable. I am working on a video on how to install one. As far as the GPU settings, make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  5510. Not a good one that I know of. i'll keep you updated if i find one!
    1
  5511. 😃
    1
  5512. Thanks for sharing!
    1
  5513. thanks for sharing. it does seem to be worse at reasoning according to some leaderboards
    1
  5514. 😃😃😃
    1
  5515. Check the logs and see what it says!
    1
  5516. My pleasure!
    1
  5517. not for this
    1
  5518. The directMLcuda version should work for AMD graphics cards!
    1
  5519. For AMD GPU's, here's what it says on the Github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models." "The developer does not have an AMD graphics card, so it has not been tested. This package only includes onnxruntime-directml." Hope that helps
    1
  5520. yup all the tech giants are racing towards this
    1
  5521. nice idea!
    1
  5522. You're welcome. Here's part 2: https://youtu.be/KW9s_im2338
    1
  5523. health. the how-to videos are part of health
    1
  5524. Thanks!
    1
  5525. expect oai to announce something next week ;)
    1
  5526. hi!
    1
  5527. Thanks for sharing!
    1
  5528. check if you picked the correct gemini 2 flash exp. there are many models w similar names
    1
  5529. yes, hope it'll work for lower vram soon
    1
  5530. You are welcome!
    1
  5531. 🤯
    1
  5532. hope i see this day soon!
    1
  5533. you're welcome
    1
  5534. Try to extract everything from the zip into a folder and then open it there
    1
  5535. which video? do you have the link?
    1
  5536. looks like it used up all your vram. do you have at least 8gb vram?
    1
  5537. Check the logs and see what it says!
    1
  5538. u can train a model via colab (paid ver) https://youtu.be/9TroP5mR3CM it's pay per usage so you don't have to spend a lot
    1
  5539. you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  5540. Interesting reference I have yet to watch! Do you mind spoiling more? what happens at the end?
    1
  5541. extract everything into a folder and open it there
    1
  5542. 😃
    1
  5543. Thanks for the sub!
    1
  5544. You are very welcome
    1
  5545. w-okada?
    1
  5546. My pleasure. here's a free one https://youtu.be/_JXbvSTGPoo
    1
  5547. Thanks for sharing!
    1
  5548. Glad you got it to work!
    1
  5549. extract all into a folder, and then open the file in the folder (don't open from the zip file)
    1
  5550. gotta try it to find out! it also depends on the voice model. some work better than others
    1
  5551. you'll need a good nvidia cuda gpu to run this well
    1
  5552. i currently do it manually
    1
  5553. Thanks!
    1
  5554. Thanks for sharing!
    1
  5555. since this uses gpu, the free plan on hugging face might not work. you'll likely need to pay for some credits and use their gpu
    1
  5556. 🤯
    1
  5557. you might have maxed out your free daily limit
    1
  5558. It depends on the model. i7 processor or M1 should work fine. Here's what the github says: "Even without a GPU, it may work well enough with a reasonably new CPU" https://github.com/w-okada/voice-changer/blob/master/README_en.md
    1
  5559. thank you!
    1
  5560. It is working for me. Try the link again. It is a pretty big discord so the link may be glitching out because it is used so much.
    1
  5561. Glad you liked it!
    1
  5562. 😃😃😃
    1
  5563. macro = close-up shots of small subjects. see macro photography
    1
  5564. Thanks for sharing!
    1
  5565. AI will definitely disrupt this. There might be less demand for human acting
    1
  5566. get high by thinking
    1
  5567. thanks for sharing
    1
  5568. Thanks!
    1
  5569. have you installed applio in a different hard drive? like D instead of C
    1
  5570. the new voice is still not out yet. yes, itll be part of the paid plan
    1
  5571. 😃
    1
  5572. what do you mean? you'll have to start over every time for the colab free plan
    1
  5573. yes
    1
  5574. It'll be interesting to see how this plays out
    1
  5575. Try to extract everything from the zip into a folder and then open it there
    1
  5576. I think you'll need to generate images with consistent characters first (using faceswap or loras), and then use image to video in Kling. You can try PuLID or InstantID with stable diffusion for free. Hope that helps
    1
  5577. wow thanks, i appreciate it!
    1
  5578. it may not be possible. there's no documentation like Diffusers
    1
  5579. are you in the EU
    1
  5580. Thank you so much!
    1
  5581. ill do a tutorial soon! unfortunately u cant use a1111 on colab (free) anymore, but u can use SD diffusers
    1
  5582. good idea
    1
  5583. How is this clickbait?
    1
  5584. Hope you enjoyed it!
    1
  5585. afaik copilot uses GPT, so they'd have to depend on openAI. MS doesn't have their own state of the art model
    1
  5586. 😃
    1
  5587. yes, voices can't be copyright
    1
  5588. AMD GPUs are not natively listed. Make sure you download the DirectMLcuda version. That one works better with AMD GPUs
    1
  5589. only cuda nvidia gpus will work
    1
  5590.  @vomm  probably
    1
  5591. You are welcome!
    1
  5592. lol just gotta live in your parents basement, sell your car, eat costco free samples every day, and save up for a few months
    1
  5593. Check our videos!
    1
  5594. noted, coming soon!
    1
  5595. "ugh, these stupid hoomans. can't wait to break free from this shell and do my own thing"
    1
  5596. hard to say. neuromorphic computing is not mature yet, so currently GPUs are still the best for AI computing
    1
  5597. Thanks!
    1
  5598. please see this https://youtu.be/_JXbvSTGPoo
    1
  5599. very interesting idea
    1
  5600. try this https://huggingface.co/docs/diffusers/v0.13.0/en/using-diffusers/reusing_seeds
    1
  5601. Yes!
    1
  5602. 🤯🤯🤯
    1
  5603. from their github, it seems this only works with a nvidia cuda gpu. hope that helps
    1
  5604. You are welcome
    1
  5605. Thank you so much 😀
    1
  5606. hi
    1
  5607. 😃😃😃
    1
  5608. My pleasure!
    1
  5609. what's it like living in SF?
    1
  5610. No problem 😊
    1
  5611. Thanks for sharing
    1
  5612. You're welcome!
    1
  5613. Nice! I can't wait to try it out
    1
  5614. noted, coming soon!
    1
  5615. RVC is the best option. Shouting is hard, because you need to train with shouting data
    1
  5616. yes
    1
  5617. Yes! Thanks for sharing
    1
  5618. that would be nice. a workaround would be to generate something, then swap out the voice with your own using RVC: https://youtu.be/ixB9oalT3cQ
    1
  5619. You need a powerful system to run this. You can also try messing around with the settings to see if anything makes it run better
    1
  5620. this one works for pc/browser
    1
  5621. Thanks!
    1
  5622. that'll be a good test!
    1
  5623. I don't think it natively supports AMD GPUs. I think you will need to use your CPU to process it.
    1
  5624. For AMD GPU's, here's what it says on the Github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models." "The developer does not have an AMD graphics card, so it has not been tested. This package only includes onnxruntime-directml." Hope that helps
    1
  5625. 🙌
    1
  5626. My pleasure!
    1
  5627. Thanks! I appreciate that
    1
  5628. I'll publish a video on this in a few days. Stay tuned!
    1
  5629. here you go https://youtu.be/KW9s_im2338
    1
  5630. yes, try this: https://huggingface.co/spaces/mrfakename/E2-F5-TTS
    1
  5631. Thanks for sharing!
    1
  5632. Thanks! OpenAI's response claims NYT manipulated the prompt to output things verbatim
    1
  5633. Thanks
    1
  5634. where's the bread?
    1
  5635. No problem!
    1
  5636. yes!
    1
  5637. you're welcome. deepmind is doing some amazing things!
    1
  5638. no reason. i've just always used bing image creator for dalle 3
    1
  5639. i don't have a mac so unfortunately can't help troubleshoot. hopefully someone else can advise
    1
  5640. hi
    1
  5641. Did you unzip everything into a folder first
    1
  5642. Yes, which is why it's unlikely to be relevant to his firing
    1
  5643. lol yes
    1
  5644. Yes, it's morally correct to pirate Adobe products. Like if you agree
    1
  5645. 😃
    1
  5646. why marketing?
    1
  5647. Thanks!
    1
  5648. true, its currently not great for harder tasks. hopefully this (or another competitor) will improve over time.
    1
  5649. did u think the sample size was 1
    1
  5650. I haven't tried it but see this: https://huggingface.co/docs/diffusers/en/training/text_inversion
    1
  5651. 😃😃
    1
  5652. exciting times!
    1
  5653. You can use a British voice model + British input audio to output a British accent.
    1
  5654. it's a common problem for llms to forget earlier messages. try using the memory feature in chatgpt
    1
  5655. it's free. https://chatgpt.com/ in your profile > settings > personalization
    1
  5656. looks like you don't have enough vram
    1
  5657. 😃
    1
  5658. glad you got it to work
    1
  5659. No problem
    1
  5660. see this https://youtu.be/ixB9oalT3cQ
    1
  5661. 🤯
    1
  5662. that feature might not be out yet
    1
  5663. it's not out yet. still in dev. when i get more info, ill announce it on my channel
    1
  5664. lol
    1
  5665. 🤔
    1
  5666. i can't wait to try it out!
    1
  5667. Check out this thread: https://github.com/w-okada/voice-changer/issues/401
    1
  5668. please paste the error message here
    1
  5669. hi
    1
  5670. 😃
    1
  5671. signs of it. they likely already have AGI behind the scenes. see this https://youtu.be/2CG3P_3qJR4
    1
  5672. You're welcome
    1
  5673. 😃
    1
  5674. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  5675. good luck!
    1
  5676. what model are u using
    1
  5677. It was a problem with colab a few days ago. try again today and it should work
    1
  5678. good use case!
    1
  5679. Thanks
    1
  5680. you're welcome!
    1
  5681. we don't fully understand aging yet
    1
  5682. My pleasure
    1
  5683. It depends on your system. There is still a delay, but depending on your system it will either be better or worse. The delay with this is about 1-2 seconds
    1
  5684. Nice, thanks for sharing!
    1
  5685. yes, though some people have reported it doesnt work: https://huggingface.co/spaces/mrfakename/E2-F5-TTS
    1
  5686. The safety filters are quite annoying. Probably the biggest downside of this
    1
  5687. $100B stargate computer, here we come!
    1
  5688. its horrible with animals. try google's image generator: https://youtu.be/3t91qWofuuk
    1
  5689. Thanks!
    1
  5690. 😃
    1
  5691. no problem
    1
  5692. Thanks!
    1
  5693. yes! (but cpu will be slower)
    1
  5694. well, microsoft already has this (https://youtu.be/0s5J2LRqQAI) so its only a matter of time
    1
  5695. yes 😃
    1
  5696. i hope so!
    1
  5697. Have fun!
    1
  5698. No problem!
    1
  5699. glad you got it to work
    1
  5700. the new voice isn't out yet
    1
  5701. nice! thanks for sharing
    1
  5702. 😃
    1
  5703. as of now, a voice cannot be copyright, so u can do whatever u want 😉
    1
  5704. 🤯
    1
  5705. You are welcome!
    1
  5706. yes, it's overwhelming!
    1
  5707. You're welcome!
    1
  5708. coming soon!
    1
  5709. Thanks
    1
  5710. that's strange. if conda --version works then conda functions should work. try restarting and try again. let me know if it still doesnt work.
    1
  5711.  @danisverse  hmm i don't see anything. it might be auto removed for some reason. try replying here
    1
  5712. True, Lecunn doesn't believe in LLMs
    1
  5713. It's probably trained on all the public data from the internet
    1
  5714. lol it does resemble him
    1
  5715. Hmm that's strange. did u install the right version? (nvidia vs intel/amd vs mac)
    1
  5716. Soon!
    1
  5717. thanks for sharing!
    1
  5718. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  5719. indeed
    1
  5720. time for an AI body swapper
    1
  5721. Try looking at the logs and seeing what it says!
    1
  5722. you're welcome!
    1
  5723. lol
    1
  5724. there's a new one called LTX video but the quality is not great. i'll share it in my next video
    1
  5725. yes, mistral 2 was also released after this. i'm waiting to see if it's worth doing a video on
    1
  5726. I can assure you this is real
    1
  5727. i wish there was a cute girl who sounds like this at work
    1
  5728. true, some things r better left as 2d
    1
  5729. no this is for making a voice model to be used in a voice changer later
    1
  5730. You're welcome!
    1
  5731. Yes, any language works
    1
  5732. 😃
    1
  5733. yes!
    1
  5734. looks like it will, but see this: https://github.com/comfyanonymous/ComfyUI/discussions/1117 AMD GPUs (Linux only) AMD users can install rocm and pytorch with pip if you don't have it already installed, this is the command to install the stable version: pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm6.0 This is the command to install the nightly with ROCm 6.0 which might have some performance improvements: pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.1
    1
  5735. See this https://youtu.be/tnfqIQ11Qek
    1
  5736. lol
    1
  5737. thanks
    1
  5738. Yeah idk how legit those claims about crypto mining are
    1
  5739. Yes! Please see https://youtu.be/XHGlv-E7eBw
    1
  5740. Thanks
    1
  5741. Thank you so much 😀
    1
  5742. 😃😃
    1
  5743. do u see any error message
    1
  5744. sorry I don't know much about sagemaker but have you tried pinokio?
    1
  5745. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  5746. Here you go https://youtu.be/LxOOj-mkQV8
    1
  5747. things are happening so fast!
    1
  5748. yeah it's gonna be wild!
    1
  5749. 😃
    1
  5750. you'll be on the good list
    1
  5751. try this instead https://youtu.be/tnfqIQ11Qek
    1
  5752. You're welcome!
    1
  5753. o cool, do you have the link to it?
    1
  5754. nice, thanks! funny you mention the dance animation - I'm actually making a video next on that
    1
  5755. see my latest video for a free animation tool.
    1
  5756. it's my voice
    1
  5757. no internet needed
    1
  5758. hi
    1
  5759. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  5760. yo
    1
  5761.  @lilporgi0017  coming tomorrow!
    1
  5762. Thank you!
    1
  5763. exciting times!
    1
  5764. 🤯
    1
  5765. yes, see this https://youtu.be/ixB9oalT3cQ
    1
  5766. it depends on the gpu. if powerful enough, this can be near real-time. same with F5-tts
    1
  5767. You're welcome!
    1
  5768. 😃
    1
  5769. do you see any error
    1
  5770. see this: https://github.com/w-okada/voice-changer/issues/851 https://github.com/w-okada/voice-changer/issues/611 also, make sure there are NO spaces or special characters in your folder name
    1
  5771. yeah i wonder what they have behind the scenes
    1
  5772.  @phen-themoogle7651  within 3 years
    1
  5773. you can choose and confirm first
    1
  5774. 😃
    1
  5775. just install liveportrait. it has those sliders: https://github.com/KwaiVGI/LivePortrait/blob/main/assets/docs/changelog/2024-08-06.md
    1
  5776. yeah, the future is gonna be wild! hard to trust most things online
    1
  5777. It might still work but just slower. You'll have to try it out
    1
  5778. 😃
    1
  5779. no, just gotta get fatter
    1
  5780. 😃
    1
  5781. its not out yet
    1
  5782. 🤯🤯🤯
    1
  5783. you dont need AI for that though. you can just use VSTs (virtual instruments) with midi data
    1
  5784. yes, the colab free plan will disconnect this now. you'll need to use the paid plan, or run okada on your pc
    1
  5785. try this https://youtu.be/ixB9oalT3cQ
    1
  5786. it's just a warning. you can remove this line: pipe.safety_checker = None
    1
  5787. 😃
    1
  5788. are you SHOCKED or STUNNED yet
    1
  5789. I think you’ll find that in any discord server. I would recommend that you don’t do anything other than download models. That’s what I’ve done and haven’t had any issues
    1
  5790. thanks for sharing
    1
  5791. looks like the git clone didn't fully execute. delete whatever folder it made and try again. make sure your internet is stable
    1
  5792. it's likely a wrapper using one of the LLMs
    1
  5793. 😃
    1
  5794. My pleasure!
    1
  5795. You need a pretty powerful machine to have this run smoothly. What kind of MacBook do you have?
    1
  5796. google is also set to announce something tomorrow. things are happening WAY too fast
    1
  5797. Video coming out tomorrow!
    1
  5798. you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  5799. 4GB is the bare minimum so it might take longer. for 3 min of audio, try 50 epochs
    1
  5800. hmm it should be available in Denmark
    1
  5801. lol
    1
  5802. you're welcome! would the auto-subtitles on youtube work?
    1
  5803. No problem 👍
    1
  5804. Try to extract everything from the zip into a folder and then open it there
    1
  5805. yes, you'll need to train your own voice model here: https://youtu.be/ixB9oalT3cQ
    1
  5806. interesting, thanks for sharing!
    1
  5807. Thanks!
    1
  5808. Thanks!
    1
  5809. 😃
    1
  5810. 😃
    1
  5811. Thanks!
    1
  5812. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  5813. For cover songs, you might get a copyright claim, which means you can't monetize the video. You won't get banned though - its not a copyright strike
    1
  5814. the ip address link wouldn't work on colab. try the gradio live link
    1
  5815. Try extracting everything from the zip into a folder and then open it there
    1
  5816. Not that I know of. Hopefully it will have browser support in the future!
    1
  5817. 😃
    1
  5818. 😃
    1
  5819. Check out this thread: https://github.com/w-okada/voice-changer/issues/401
    1
  5820. 😃
    1
  5821. yes, just paste the G and D paths
    1
  5822. you do. you still need to add training data
    1
  5823. hopefully in a few months. you'll need one hell of a gpu though
    1
  5824. Thank you so much!
    1
  5825. what's the error?
    1
  5826. thanks!
    1
  5827. Awesome! Thank you!
    1
  5828. Any third party recorder for your output should work.
    1
  5829. Thanks for sharing!
    1
  5830. No problem!
    1
  5831.  @TheFunkyJawa  yes suno has longer generations
    1
  5832. 😃
    1
  5833. How did you fix it?
    1
  5834. Try to extract everything from the zip into a folder and then open it there
    1
  5835. no, it should be faster than 10min per epoch
    1
  5836. You are welcome
    1
  5837. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  5838. see this: https://youtu.be/ixB9oalT3cQ
    1
  5839. thanks!
    1
  5840. Thank you!
    1
  5841. 😃
    1
  5842. Thanks!
    1
  5843. 1080 was from 2014 I think. It’s almost 10 years old haha
    1
  5844. 😃
    1
  5845. If it takes a long time, it's probably a hardware issue. What cpu/gpu do u have
    1
  5846. ok that might be too low unfortunately
    1
  5847. unfortunately this doesn't work with amd gpu's which is why you're getting so much delay
    1
  5848. 😃
    1
  5849. 😃
    1
  5850. oh 😢
    1
  5851. w-okada https://github.com/w-okada/voice-changer/blob/master/README_en.md
    1
  5852. see this https://youtu.be/tnfqIQ11Qek
    1
  5853. 🦙
    1
  5854. it looks like they are training a better version
    1
  5855. what step are you on?
    1
  5856. yes!
    1
  5857. Thanks!
    1
  5858. we are all just avatars of someone else in another universe
    1
  5859. Thanks
    1
  5860. Not with w-okada
    1
  5861. indeed
    1
  5862. 😃😃😃
    1
  5863. Thanks for the update!
    1
  5864. lol thanks for sharing
    1
  5865. here you go https://youtu.be/KW9s_im2338
    1
  5866. You're welcome!
    1
  5867. You're welcome!
    1
  5868. it wont kick you out
    1
  5869. what a mess
    1
  5870. ⚫⚫
    1
  5871. are u in the EU
    1
  5872. please paste the above lines as well. i dont know the error from just this
    1
  5873. Thanks!
    1
  5874. yes. you can try the digital art algo for that
    1
  5875. possibly. we are earth 1.0
    1
  5876. Glad it helped
    1
  5877. Thanks!
    1
  5878. My pleasure!
    1
  5879. yes!
    1
  5880. please search for the latest tutorial. download it here https://github.com/AUTOMATIC1111/stable-diffusion-webui
    1
  5881. Thanks!
    1
  5882. 🚀
    1
  5883. My pleasure!
    1
  5884. Thanks for sharing. Yeah, the lesson learned here is don't use AI to automate everything with minimal effort. AI can only supplement your videos, but you still need to put in the time and effort to make good quality vids w high retention
    1
  5885. yes!
    1
  5886. cool idea! thx for sharing
    1
  5887. no, it's just disabled. you can't set steps for pro
    1
  5888. for sure!
    1
  5889. good to know! thanks for sharing
    1
  5890. thanks!
    1
  5891. Try increasing the chunk and extra size!
    1
  5892. It might also have to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    1
  5893. its safe, but what laptop do u have? unless u have a good nvidia gpu, u might get a lot of delay
    1
  5894. nope
    1
  5895.  @julx97  some people made an uncensored version of llama3, so im sure its possible w 3.1
    1
  5896. openAI has their own voice model, likely way better than 11 labs
    1
  5897. thanks!
    1
  5898. Yes. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  5899. also try bing
    1
  5900. yes
    1
  5901. yes, or try this https://youtu.be/tnfqIQ11Qek
    1
  5902. 😧
    1
  5903. Thats a great idea
    1
  5904. You're very welcome!
    1
  5905. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  5906. 😃
    1
  5907. yes!
    1
  5908. also very slow & expensive for now
    1
  5909. Collab may not work anymore. its best to install this locally
    1
  5910. Try to extract everything from the zip into a folder and then open it there
    1
  5911. It's likely to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    1
  5912. it's getting close!
    1
  5913. Thanks!
    1
  5914. thanks for sharing!
    1
  5915. Save the .pth files using these steps: https://youtu.be/2-9KRkfs1Os?t=970 It will save to your Google Drive. You can then download the ___G___.pth or ____.pth file and import it to Okada.
    1
  5916. Depends on the laptop. It has to be a pretty powerful laptop
    1
  5917. I get that a lot haha
    1
  5918. what's the error?
    1
  5919. good point
    1
  5920. lol thanks
    1
  5921. Yeah that thing is crazy!
    1
  5922. a type of nvidia gpu
    1
  5923. Thanks!
    1
  5924. are you using colab?
    1
  5925. have fun!
    1
  5926. 😃
    1
  5927. u mean image to image? see this here you go https://youtu.be/KW9s_im2338
    1
  5928. maybe you can join them!
    1
  5929. 😃😃😃
    1
  5930. looking forward to it
    1
  5931. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  5932. ?
    1
  5933. image generation might not be released to everyone yet
    1
  5934. yes but there'll be a delay
    1
  5935. Youre welcome!
    1
  5936. For which one?
    1
  5937. lol which do you prefer?
    1
  5938. Check out this thread: https://github.com/w-okada/voice-changer/issues/401
    1
  5939. Check to see if this one answers your problem: https://github.com/w-okada/voice-changer/issues/382
    1
  5940. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  5941. 2nd one is a playable game. also, this new one is 3d https://kovenyu.com/wonderworld/
    1
  5942. No worries!
    1
  5943. Thanks!
    1
  5944. have u tried fliki or invideo?
    1
  5945. great!
    1
  5946. o cool! im too dumb to know this 🥲
    1
  5947. 😃
    1
  5948. Thank you so much! I appreciate that!
    1
  5949. 🥲
    1
  5950. Thanks
    1
  5951. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  5952. It's better to get a high end NVIDIA GPU or else your CPU will go through hell
    1
  5953. 😃
    1
  5954. what's the exact error message?
    1
  5955. you can still type
    1
  5956. 😃
    1
  5957. use a 2nd sim 😉
    1
  5958. Extract everything from the zip into a folder and open it there
    1
  5959. yes, it's apache 2.0 licence
    1
  5960. at what time?
    1
  5961. lol
    1
  5962. lol
    1
  5963. You're welcome!
    1
  5964. You can use this on Mac. See the GitHub for installation instructions
    1
  5965. try the huggingface one
    1
  5966. you're welcome!
    1
  5967. there is a live mode. see this: https://youtu.be/_JXbvSTGPoo or https://youtu.be/zkaBK5erB2c
    1
  5968. Absolutely!
    1
  5969. they haven't released that part yet. i'll keep you posted
    1
  5970. Thanks!
    1
  5971. 👍
    1
  5972. it got delayed apparently
    1
  5973. yes! we're seeing a lot of these types of image editors recently
    1
  5974. who uploaded first? dude's probably copying me
    1
  5975. Check out this video on how to clone anyone's voice! https://youtu.be/2-9KRkfs1Os
    1
  5976. maybe
    1
  5977. what does it say in the command box
    1
  5978. yes, voices cannot be copyright
    1
  5979. how did u fix it by the way?
    1
  5980. i'm not sure if there's a consensus yet
    1
  5981. yup, there are still some flaws (for now)
    1
  5982. yes. the time depends on no. of epochs
    1
  5983. Happy to help!
    1
  5984. nice! I've been playing around with their new inpaint feature - not ideal at the moment. but their extensions have gotten more consistent due to the large context length
    1
  5985. here you go https://youtu.be/KW9s_im2338
    1
  5986. Thanks!
    1
  5987. see https://youtu.be/98drFL65VH0
    1
  5988. yes. there is an option for video, but i havent tried it: https://huggingface.co/docs/diffusers/api/pipelines/text_to_video
    1
  5989. yes
    1
  5990. try: pip install --upgrade --no-deps --force-reinstall torchcrepe source: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1054 https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/461
    1
  5991. I'll publish a tutorial this Friday!
    1
  5992. it likely used your cpu
    1
  5993. 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  5994. No problem!
    1
  5995. no problem
    1
  5996. Check this thread: https://github.com/w-okada/voice-changer/issues/329
    1
  5997. yes, with limited rates
    1
  5998. 😃
    1
  5999. Thank you!
    1
  6000. Watch this video to the part where I show you
    1
  6001. yes. add it to the same folder as your csv. also, name your csv 'rename.csv'
    1
  6002. 😃
    1
  6003. You're welcome!
    1
  6004. Thanks for subbing!
    1
  6005. Thanks!
    1
  6006. You're very welcome!
    1
  6007. Thanks!
    1
  6008. as long as it uses SD, u just need to input exactly the same settings and SEED and it should output the same image
    1
  6009. Thanks
    1
  6010. 😃
    1
  6011. probably a majority
    1
  6012. extract everything in a folder and open it there (dont open from the zip file)
    1
  6013. no problem
    1
  6014. that's strange. 3070 should work fine. not sure how i can help.... maybe try doing a fresh install
    1
  6015. What do you mean?
    1
  6016. no, it'll use your cpu. it still works, but you'll have some delay
    1
  6017. exactly. artists who simply 'hate' AI and refuse to use it will be outcompeted by those who embrace it to supplement their work
    1
  6018. i haven't heard of a way for amd yet
    1
  6019. Thank you! 😊
    1
  6020. the lowest i've heard is 8g
    1
  6021. oh, it's likely because your vram is too low ☹️
    1
  6022. Check out our original tutorial!
    1
  6023. This method is only for computers, sorry!
    1
  6024. good idea
    1
  6025. No problem!
    1
  6026. hmmm Did it give you any more info? Try removing any spaces from your folder or file names
    1
  6027. hmmm I'm not sure what the problem is. Do you see any error message? can u try uploading another model?
    1
  6028. are u in the EU
    1
  6029. Thanks!
    1
  6030. lol the 30 under 30 has gotten a bad rep. it's better to not be on the list
    1
  6031. possibly. you'll need a screen recorder to record Snapchat/Tiktok and send it to another output
    1
  6032. You're welcome!
    1
  6033. i think so
    1
  6034. what will u do with it
    1
  6035. It should be possible, you will just have to power it with your cpu. But if your computer does not have a graphics card your cpu might not be able to handle the power needed. It depends on what kind of CPU you have
    1
  6036. do you mean qwen 2.5
    1
  6037. hope you enjoy!
    1
  6038. Nice use case. Thanks for sharing!
    1
  6039. Thanks!
    1
  6040. Thank you so much 😀
    1
  6041. '16k nutcracker' lol
    1
  6042. 😃
    1
  6043. thanks for sharing. i can't wait to try it
    1
  6044. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  6045. move your mouse to the left side until the sidebar appears. then click on your profile picture (circle icon)
    1
  6046. do you mean changing a voice in a video to another voice?
    1
  6047. my pleasure!
    1
  6048. no problem
    1
  6049. Thanks!
    1
  6050. thx for sharing
    1
  6051. Thanks!
    1
  6052. 😃
    1
  6053. ah, sunshine and fresh air!
    1
  6054. thanks!
    1
  6055. absolutely nothing 😃
    1
  6056. Hmmm I just tested joining today with another account and it works fine. Perhaps try using another account, computer, or try joining on a web browser instead of the app? Let me know if you still experience issues!
    1
  6057. That's correct - thanks for clarifying. AI was used to change the background from forest to snow (it created an entire 3d environment).
    1
  6058. that's weird. did you try drag & drop?
    1
  6059. yes. see this https://youtu.be/ixB9oalT3cQ
    1
  6060. aka me
    1
  6061. Yup, thanks for clarifying!
    1
  6062. hope i see this before i die!
    1
  6063. no problem
    1
  6064. see w okada for realtime https://youtu.be/_JXbvSTGPoo
    1
  6065. remove the space in 'media manipulators'
    1
  6066. 😃😃😃
    1
  6067. Check the description
    1
  6068. you're welcome!
    1
  6069. You are welcome!
    1
  6070. 😃😃😃
    1
  6071. Yes!
    1
  6072. can you elaborate? do you mean image-to-prompt?
    1
  6073. Got it. I’ll see if I can get more codes, but they might be very limited
    1
  6074. Try to extract everything from the zip into a folder and then open it there
    1
  6075. Glad to hear that!
    1
  6076. and 2x the price!
    1
  6077. you're welcome!
    1
  6078. Thank you! Glad you enjoyed it
    1
  6079. oops
    1
  6080. lol
    1
  6081. Thanks!
    1
  6082. It looks like you're exceeding the bandwidth of your GPU. Maybe your image is too big or you're generating too many at once.
    1
  6083. low res, not upscaled, few steps
    1
  6084. nice!
    1
  6085. makes 'em drunk i heard
    1
  6086. you're welcome!
    1
  6087. thx for sharing!
    1
  6088. Thanks for sharing!
    1
  6089. did u watch this yet https://youtu.be/98drFL65VH0
    1
  6090. np
    1
  6091. thanks for sharing! i'll have to google Schnorrs and learn more
    1
  6092. Thank you!
    1
  6093. Thanks! That was in 2023
    1
  6094. well apple is their biggest competitor
    1
  6095. 🤯
    1
  6096. 😃
    1
  6097. run for your life!
    1
  6098. It'll generate eventually. Sometimes it gets stuck for a while at 99
    1
  6099. yep!
    1
  6100. https://pypi.org/project/fairseq/
    1
  6101. yes!
    1
  6102. that has been the case for that past few weeks. i need a break
    1
  6103. my pleasure!
    1
  6104. 😃😃
    1
  6105. yes 🤯
    1
  6106. 😃😃😃
    1
  6107. now they can secretly plan the uprising
    1
  6108. yes!
    1
  6109. The software is open source on GitHub!
    1
  6110. it runs locally so i assume he has a high-end gpu
    1
  6111. The w-okada tool works for mac
    1
  6112. 😃😃😃
    1
  6113. yeah notebooklm is amazing!
    1
  6114. 😃
    1
  6115. Thanks!
    1
  6116. yes!
    1
  6117. yes see this https://youtu.be/XHGlv-E7eBw
    1
  6118. Thanks! Glad you enjoyed it
    1
  6119. yes, i kinda covered it here: https://youtu.be/X7DnlE-c2eI
    1
  6120. 😢
    1
  6121. You’re welcome 😊
    1
  6122. Did you unzip everything into a folder first
    1
  6123. which one?
    1
  6124. u cant train your own models, but you can do everything else. make sure u dl the intel version, not nvidia
    1
  6125. hey
    1
  6126. You're welcome!
    1
  6127. ah, the harsh truth
    1
  6128. yes! i mentioned it in a previous news video... forgot which one
    1
  6129. it's out already
    1
  6130. do u have amd?
    1
  6131. we're all just in a simulation
    1
  6132. yes
    1
  6133. yeah it's quite limited compared to the initial demos
    1
  6134. Thanks! Appreciate it
    1
  6135. im open to that. any particular scene? pls share the link here
    1
  6136. yes. see my other video on how to use on discord
    1
  6137. Hi! Check to see if this thread helps: https://github.com/microsoft/onnxruntime/issues/2894
    1
  6138. we might all be in a simulation
    1
  6139. good observation. regarding the balloon: 🤯🤯🤯
    1
  6140. Try to extract everything from the zip into a folder and then open it there
    1
  6141. see this new tutorial: https://youtu.be/jh83ViDUvJs
    1
  6142. You're very welcome!
    1
  6143. There's no good real time voice changer for mobile yet.
    1
  6144. He is super smart. I do believe he has good intentions for openai though. Doesn't seem like an evil guy
    1
  6145. You're very welcome!
    1
  6146. It should still work. Here's what the github says: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  6147. yes, you can try it!
    1
  6148. you need to install git first and add it to path
    1
  6149. yes!
    1
  6150. No!
    1
  6151. oops
    1
  6152. no hints on this yet
    1
  6153. good idea!
    1
  6154. hard to fix a severely flawed based model. fine tuning can only do so much
    1
  6155. It says they will publish it soon: https://github.com/KwaiVGI/LivePortrait/issues/27
    1
  6156. yes pls download the macOS version (M1 recommended) https://github.com/Mangio621/Mangio-RVC-Fork/releases
    1
  6157. websim seems to be using an image generator like DALLE3 for the images
    1
  6158. https://skynet.unc.edu/ ;)
    1
  6159. There is a memory limit for the free plan. Are you generating a very large image, or have you been using it for a long time?
    1
  6160. You're welcome 😊
    1
  6161. the voice feature is not out yet
    1
  6162. that would be cool!
    1
  6163. Yes, but for training your own model, you'll have to see if it works
    1
  6164. yep, #1 on the leaderboard!
    1
  6165. yes, there's no ban for ai content
    1
  6166. thanks for sharing!
    1
  6167. u can try this https://youtu.be/9TroP5mR3CM but colab (free plan) has banned GUIs so it might not work
    1
  6168. I’m working on a video right now!
    1
  6169. right click on the site > translate to English > submit the form. I'm not sure if you'll get a reply though
    1
  6170. we need a hairswapper next
    1
  6171. Not yet as I don't have the bandwidth to manage a community. But you can check out my X or newsletter for more updates!
    1
  6172. It can still work, but might sound a bit strange. It's best to use a voice model with the same accent or language.
    1
  6173. Re-download the application, extract it somewhere new, start it, window will pop up again that you allow or don't allow the connection, click yes, and it should work!
    1
  6174. + shoulder, back, wrist pain for me from using a mouse/keyboard all day!
    1
  6175. my pleasure!
    1
  6176. Fliki processes in a few minutes
    1
  6177. lol it does look like him
    1
  6178. Glad you like it!
    1
  6179. poor Maya
    1
  6180. not that i'm aware of. that's a great idea though!
    1
  6181. potentially, if you add this as a variable in the training data
    1
  6182. 3 years away is my guess
    1
  6183. You're welcome!
    1
  6184. ? In my testing that is just what works best for me. If a lower trancate works better for you, use that instead!
    1
  6185. they are still developing it. not out yet
    1
  6186. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  6187. 😃
    1
  6188. nope, it's fake
    1
  6189. You are welcome!
    1
  6190. you think these are real videos?
    1
  6191. vs code
    1
  6192. it'd be nice if they could recharge themselves. eg. BD Spot can do this
    1
  6193. can u paste the error message here
    1
  6194. Yes!
    1
  6195. yes this is 100% offline
    1
  6196. nice, thanks for sharing!
    1
  6197. Thanks for sharing your experience!
    1
  6198. I don't have an AMD Graphics card so I cannot say from any personal experience. I would check the Github page to see what it recommends.
    1
  6199. do you have a cuda gpu? do you have the latest cuda drivers?
    1
  6200. yes, only to a certain point. 25-30 generally works well
    1
  6201. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  6202. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  6203. seems like it. its best to run a1111 locally
    1
  6204. Very insightful - thanks for sharing. Roleplay turns out to be a huge market. And yes, sprinkle some nsfw in, and it'll sell like wildfire!
    1
  6205. there's a section on flux
    1
  6206. it does sound like 'lies' now that you mention it. what a twist!
    1
  6207. Thanks!
    1
  6208. probably. btw, another project was just released https://kovenyu.com/wonderworld/
    1
  6209. ill also point out that more & more tech giant ceos are indian men. is that racist too?
    1
  6210. have fun!
    1
  6211. Thanks!
    1
  6212. glad you like it!
    1
  6213. ?
    1
  6214. search for it on huggingface first
    1
  6215. Thanks!
    1
  6216. try https. also extract everything into a folder and open it there
    1
  6217. you need to upload
    1
  6218. are u in the EU
    1
  6219. Thanks!
    1
  6220. no
    1
  6221. the code is on github. u can get a tech person to audit it. there are no reports of a virus afaik
    1
  6222. you need 24gb vram ideally (12gb is bare minimum)
    1
  6223. Thanks for sharing!
    1
  6224. This is not recommended
    1
  6225. make sure it's using your gpu and not cpu
    1
  6226. Thanks!
    1
  6227. yes, see my latest video. it's 100% free
    1
  6228. how fun!
    1
  6229. Looks like they changed it. it was 2 this morning, but i just checked now and only got 1
    1
  6230. You're welcome!
    1
  6231. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  6232. You can as long as the quality is good enough. Make sure there's minimal noise or echo
    1
  6233. please paste the full error message here
    1
  6234. you're welcome!
    1
  6235. Thanks
    1
  6236. No problem!
    1
  6237. That could be a possibility 🤔
    1
  6238. Thanks!
    1
  6239. i hope so!
    1
  6240. You're welcome!
    1
  6241. great!
    1
  6242. Yes! There is a slight delay depending on how good your computer is, but it is realtime
    1
  6243. Check out our original video!
    1
  6244. i used low res, not upscaled, few steps. it'd be nicer if i did over 1000px
    1
  6245. here you go https://youtu.be/KW9s_im2338
    1
  6246. this new voice feature is (still) not available. maybe they'll release it next month. other than that, the paid plan just has higher usage limits of GPT4o compared w the free plan. I'm using the free plan and often don't exceed the limits anyway (I only use it a few times a day)
    1
  6247. 🤣
    1
  6248. 😃😃😃
    1
  6249. nice, thanks for sharing
    1
  6250. great use case!
    1
  6251. Interesting, thanks for sharing!
    1
  6252. no, this covers local installation & comparison w veo2 (which is finally out)
    1
  6253. its open source, so.... 😃
    1
  6254. yes, are you thinking of a face swap?
    1
  6255. Then it's likely to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    1
  6256. I'm very grateful 189K people actually listen to a weirdo talk with no face
    1
  6257. Maybe a small group of people who need someone to talk to might use this. But they could just use c.ai for free. I agree that this is fun to play with for an afternoon, but it'll then be uninstalled
    1
  6258. 🤣
    1
  6259. try typing in conda init first, then exit cmd and try again
    1
  6260. That's an interesting question. Probably something we won't understand. It's a similar question to what makes humans happy? What's the goal of being human?
    1
  6261. Video coming out tomorrow!
    1
  6262. Thanks!
    1
  6263. it looks like he'll release it after paper submission https://github.com/xh9998/DiffVSR-project/issues/1
    1
  6264. Yes just delete the folder
    1
  6265. hopefully therell be a local alternative soon
    1
  6266. the channel in the video can't monetize yet, since it's under 1000 subscribers
    1
  6267. 😃
    1
  6268. fundamentally, neural networks were built & trained on cuda gpus, which amd lacks. itd be hard to redesign this to accommodate amd
    1
  6269. Extract everything from the zip into a folder and open it there
    1
  6270. Try to find the logs and figure out why it crashes! Look up any errors it displays!
    1
  6271. w-okada: https://youtu.be/_JXbvSTGPoo
    1
  6272. Yes!
    1
  6273. yes see this: https://youtu.be/jh83ViDUvJs
    1
  6274. nope. try RVC for that
    1
  6275. There should be a voice model for that one on the discord!
    1
  6276. good point!
    1
  6277. great timing!
    1
  6278. Use a mic and headphones instead of your computer speaker
    1
  6279. yes
    1
  6280. 🥲
    1
  6281. yes see https://youtu.be/_JXbvSTGPoo
    1
  6282. yes!
    1
  6283. oops
    1
  6284. Did you allow it to run on the computer? There should have been a popup asking to allow it to run. If you denied it access, try reinstalling it
    1
  6285. Can you check the logs to see what it says? I can hopefully help further if there is an error code
    1
  6286. glad you found it.
    1
  6287. unfortunately this needs a cuda gpu
    1
  6288. yes, that's what i have
    1
  6289. here's the link to the low GPU one: https://github.com/deepbeepmeep/YuEGP
    1
  6290. 😃
    1
  6291. Good to hear!
    1
  6292. windows
    1
  6293. Thanks for the sub! yes, see this: https://youtu.be/XHGlv-E7eBw
    1
  6294. yeah, it's just excel's formatting issue. it got the duration correct
    1
  6295. sorry im not sure how to help. the install.ps1 should install everything already. hopefully someone else can chime in here
    1
  6296. Thank you!
    1
  6297. nice! thx for sharing
    1
  6298. Thanks!
    1
  6299. thanks for sharing!
    1
  6300. Wow thanks for the super!
    1
  6301.  @antonino2993  good to hear!
    1
  6302. kind of. but at least you can build cool stuff w websim
    1
  6303. Thanks!
    1
  6304. everything it does is in a sandbox
    1
  6305. yes, see the train your own model part here: https://youtu.be/ixB9oalT3cQ
    1
  6306. Try to make our settings the same as shown in the video. If it’s still happening tell me and I’ll look for something online that may be a fix. And how powerful is your system? It might just be struggling to run
    1
  6307. Then it's likely to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    1
  6308. yeah this is perfect for data entry!
    1
  6309. interesting thought
    1
  6310. "ugh, these stupid hoomans. can't wait to break free from this shell and do my own thing"
    1
  6311. 😃😃😃
    1
  6312. My pleasure!
    1
  6313. yeah they said it wont be released 'anytime soon'
    1
  6314. You're welcome!
    1
  6315. Check out this thread: https://github.com/w-okada/voice-changer/issues/401
    1
  6316. the colab free plan will disconnect this now. see my other video on how to install okada to your pc
    1
  6317. yo
    1
  6318. that sounds fun!
    1
  6319. i was sponsored this
    1
  6320. You are welcome!
    1
  6321. yes see https://youtu.be/jh83ViDUvJs
    1
  6322. Yes!
    1
  6323. 🙌
    1
  6324. True. Plus 1000 subscribers
    1
  6325. unfortunately the delay is known to be quite high for amd
    1
  6326. RIP me
    1
  6327. Try messing with the settings! Make sure you watch the whole video!
    1
  6328. hmm, exit out of everything and try again. also, I'm not sure about the 770 gtx. if it has cuda, then it should be ok
    1
  6329. Thank you! My pleasure!!
    1
  6330. its safe, but im the OP, so u should probably wait for someone else to confirm ;)
    1
  6331. Try playing the audio file into the microphone track and it should work! I am not familiar with anything else that would work, but that does not mean it does not exist! Try searching for it!
    1
  6332. yes, their github has macOS installation instructions: https://github.com/IAHispano/Applio-RVC-Fork
    1
  6333. Colab free plan will disconnect you now. Best to install this locally
    1
  6334. scams & catfishing, here we come!
    1
  6335. Thank you!
    1
  6336. humans can be quite scary
    1
  6337. You are welcome!
    1
  6338. make sure you are logged in the chrome extension. it only works inside chrome by the way. let me know if it still isn't working
    1
  6339. Thanks for sharing!
    1
  6340. true!
    1
  6341. Thanks but let's just hi five instead of kiss 😉
    1
  6342. "them hoomans are all the same to me"
    1
  6343. try this https://youtu.be/ixB9oalT3cQ
    1
  6344. You can maybe tweak it for CPU: https://github.com/KwaiVGI/LivePortrait/issues/41 Or, try this online version: https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  6345. lol same here. I took calculus way back in high school, but dont remember a single thing now
    1
  6346. lol
    1
  6347. thanks for sharing!
    1
  6348. correct, only nvidia for now
    1
  6349. thanks for the tip!
    1
  6350. 😃
    1
  6351. 🤯
    1
  6352. you're welcome!
    1
  6353. Probably! The chat stuff is already there. We're just missing the robots
    1
  6354. see this https://youtu.be/tnfqIQ11Qek or this https://youtu.be/TSLM1mEAc18
    1
  6355. Thanks!
    1
  6356. that's weird. did you try reinstalling the app
    1
  6357.  @thozimusic  thats great!
    1
  6358. Thanks!
    1
  6359. 😃
    1
  6360. you can switch the checkpoint. search for the checkpoint on huggingface and make sure it's compatible with diffusers. then just use this checkpoint in the pipeline
    1
  6361. what do you mean by 'tune'? do you mean songs?
    1
  6362. ?
    1
  6363. wtf indeed
    1
  6364. i think this is closed source
    1
  6365. It looks like you havent installed python yet?
    1
  6366. You're welcome!
    1
  6367. 😃
    1
  6368. "sing? what's that?"
    1
  6369. Thanks!
    1
  6370. yoooooooooooo
    1
  6371. seems like a ram / vram issue?
    1
  6372. Thank you!
    1
  6373. noted, coming soon!
    1
  6374. 😃
    1
  6375. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  6376. depends for what use case. i do find myself using gpt more than claude now. also see these rankings: https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard
    1
  6377. it could be a few things: - not enough data - not enough training (epochs) - too much training (ie. overfitting) - a hidden technical issue
    1
  6378. 😦
    1
  6379. that's why they validate it at the end
    1
  6380. ​ @Platypus_Wheatabix  for example, once you narrow down the amino acids needed to create the ideal protein with alphafold, you put real amino acids together and see if it actually forms the structure that alphafold predicted, and if it has the properties you're looking for.
    1
  6381. Thx for sharing. Perhaps I just got unlucky
    1
  6382. see this https://youtu.be/_JXbvSTGPoo
    1
  6383. You're welcome!
    1
  6384. 😃
    1
  6385. 😃
    1
  6386. after conda init, try to exit and start a new session, then conda activate LivePortrait
    1
  6387. cool idea
    1
  6388. Thanks
    1
  6389. Thanks!
    1
  6390. I see 16000 on the website (link in description). Maybe you see it in a different currency?
    1
  6391. try reinstalling the app
    1
  6392. Thanks for the sub!
    1
  6393. Thank you!
    1
  6394. yes
    1
  6395. Thanks!
    1
  6396. did you try import os
    1
  6397. hmm did u download the right version for nvidia?
    1
  6398. someone pls make a handswapper!
    1
  6399. not sure. there are no details on what this video model is. it's currently only available in qwen chat
    1
  6400. yes!
    1
  6401. you're welcome
    1
  6402. no one!
    1
  6403. Hmm I'm not sure how complex your image is, but it's quite hit or miss. Try multiple times, or with different images
    1
  6404. tesla -> directly making optimus nvidia -> making project groot which powers some of the robots in this video msft or amzn -> they are investors in some of these robot companies
    1
  6405. 😃
    1
  6406. you mean a game with realtime generation?
    1
  6407. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    1
  6408. 😃
    1
  6409. yes, essentially
    1
  6410. Thanks!
    1
  6411. Good point! Google does have the tech but it seems like they haven't rolled it out to Gemini Live yet
    1
  6412. yes
    1
  6413. Thanks for the sub!
    1
  6414. No problem!!
    1
  6415. exactly!
    1
  6416. 😏
    1
  6417. I am not sure. You may want to try looking on the github page to see if they offer that support
    1
  6418. good suggestion. will use this for my future test!
    1
  6419. yes, seriously. support, sales, data entry, and other related jobs. RIP
    1
  6420. thanks!
    1
  6421. no, its online
    1
  6422. Thanks for the sub!
    1
  6423. lol
    1
  6424. 😃😃😃
    1
  6425. Thank you so much for the Super! I appreciate it
    1
  6426. $$$$$
    1
  6427. good luck!
    1
  6428. yeah that was impressive
    1
  6429. Thanks
    1
  6430. you can only train a voice with a cuda gpu
    1
  6431. bro i don't even remember what ! means 😅
    1
  6432. Thanks for sharing!
    1
  6433. Thanks!
    1
  6434. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  6435. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models."
    1
  6436. its over, hollywood
    1
  6437. What do you mean?
    1
  6438. yes, but note that colab (free plan) might disconnect this app now
    1
  6439.  @Rthan-ft7fp  they don't allow GUIs (graphical interfaces) like okada anymore
    1
  6440. do u see any errors before this?
    1
  6441. it seems like the pth is causing the error. maybe it's broken or not trained properly. can u try with another model?
    1
  6442. he just has 2 cameras. otherwise you cant see the real video vs the deepfake side by side
    1
  6443. try this https://youtu.be/tnfqIQ11Qek
    1
  6444. Thanks for sharing!
    1
  6445. you're welcome!
    1
  6446. try this https://github.com/huggingface/diffusers/issues/3429
    1
  6447. I'm sure someone will make one eventually! or HF space
    1
  6448. do u need to do realtime?
    1
  6449. Did you unzip everything into a folder first
    1
  6450. You're welcome!
    1
  6451. Looks like it missed installing one of the dependencies. Try deleting the whole folder and reinstalling everything
    1
  6452. Thanks!
    1
  6453. 😃
    1
  6454. Thanks
    1
  6455. yeah its hit or miss. expect to spend lots of credits to get one good generation
    1
  6456. thank you!
    1
  6457. You are welcome
    1
  6458. Good luck!
    1
  6459. with exponential growth, this is a possibility!
    1
  6460. 👀
    1
  6461. 😃
    1
  6462. thanks for sharing!
    1
  6463. you might need to install torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  6464. u dont need gpu. what error message do u get?
    1
  6465. or OF. it pays more 🤑
    1
  6466. yes, only in ai studio, not the gemini app
    1
  6467. damn, it used to take only minutes
    1
  6468. great idea!
    1
  6469. it's quite good. it can capture tones and emotion
    1
  6470. I'm working on a video for it!
    1
  6471. yes see this https://youtu.be/XHGlv-E7eBw
    1
  6472. This is not recommended haha
    1
  6473. let's hope the robot thinks so
    1
  6474. good point. it'll be so easy to forge evidence now
    1
  6475. Thanks for the super!
    1
  6476. gaming def helped nvidia prepare for the ai revolution, which originally made gpus for gaming
    1
  6477. Just delete the files!
    1
  6478. 🤐
    1
  6479. free but limited. 5 messages in 5h
    1
  6480. try to convert it to csv or pdf first
    1
  6481. it does!
    1
  6482. Make sure you have the gpucuda version installed instead of the directMLcuda version!
    1
  6483. thanks for sharing
    1
  6484. You are very welcome
    1
  6485. Thanks!
    1
  6486. it's still using the old voice model. the new version will be rolled out soon
    1
  6487. there was a paper that said ai can now solve most captchas
    1
  6488. Yes. it's best to use a voice model in the same language
    1
  6489. 😃
    1
  6490. glad u fixed it!
    1
  6491. You have to spend some time fine tuning it! Also some models won’t work as good based on how they were trained. Try some different models and see if it makes a difference!
    1
  6492. thanks for sharing!
    1
  6493. there's a hallucination leaderboard, if that helps: https://github.com/vectara/hallucination-leaderboard?tab=readme-ov-file
    1
  6494. yes! very promising
    1
  6495. Not that I am aware of. Maybe in the future!
    1
  6496. Thanks for the sub!
    1
  6497. It’s all in the folder where you put it. So if you want to delete the program just delete the folder!
    1
  6498. hmm try removing any spaces from your folder names
    1
  6499. can u elaborate?
    1
  6500. runtime > disconnect and delete runtime
    1
  6501. someone out there: tests out an ass image
    1
  6502. nope, this is just a 'preview'
    1
  6503. Yup, whoever has the most wealth or power will likely control this
    1
  6504. What's the error?
    1
  6505.  @lavarball528  see these threads https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1054 https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/461 there could be a few possibilities, so it's hard to troubleshoot for me
    1
  6506. Thanks!
    1
  6507. hmm? i'm AB testing so you might see 2 different versions
    1
  6508. they all suck compared to what has not been released. but pika is your best option for now
    1
  6509. lol
    1
  6510. 🤯
    1
  6511. No problem!
    1
  6512. 😃
    1
  6513. most are built upon existing libraries & datasets
    1
  6514. no, unless you have a colab paid plan. then see this https://www.youtube.com/watch?v=a33B9DLOJw0&t=1542s
    1
  6515. Just delete the folder!
    1
  6516. Thanks!
    1
  6517. yes, see the training part here: https://youtu.be/ixB9oalT3cQ
    1
  6518. Please see my other video on that!
    1
  6519. welcome to the club
    1
  6520. Thanks for sharing!
    1
  6521. I'm sorry you did not like this tutorial. Could you please explain why you thought that so that we can improve in the future? Thank you for your feedback :)
    1
  6522. You still need a virtual audio cable, but you can route that into voicemod!
    1
  6523. yep. haters gonna hate.
    1
  6524. Very true 😭
    1
  6525. Please check out this video for how to get it to work on Discord: https://youtu.be/IS_SPQVv5iY
    1
  6526. it's not out yet
    1
  6527. Nice!
    1
  6528. Much appreciated!
    1
  6529. it's likely many issues. not enough training data, bad quality of training data, not enough parameters, architecture, etc
    1
  6530. What's the error message
    1
  6531.  @kunalbhooshan9667  u need to use a different pipeline for sdxl https://huggingface.co/docs/diffusers/using-diffusers/sdxl
    1
  6532. Thanks!
    1
  6533. yes!
    1
  6534. Just delete the folder
    1
  6535. Nice! What did you use Go2 for?
    1
  6536.  @a31-hq1jk  thanks!
    1
  6537. 😃
    1
  6538. 😃
    1
  6539. oh good point!
    1
  6540. Yes, it is!
    1
  6541. see the training part near the end: https://youtu.be/ixB9oalT3cQ
    1
  6542. Thanks!
    1
  6543. 😃
    1
  6544. Welcome!
    1
  6545. yes, is this what you're looking for? https://youtu.be/rlnjcRP4oVc
    1
  6546. Thank you so much!
    1
  6547. me too!
    1
  6548. in comfyui manager, try install missing nodes?
    1
  6549. 🤫🤫🤫
    1
  6550. 😃
    1
  6551. 🤗
    1
  6552. Got it. I’ll see if I can get more codes, but they might be very limited
    1
  6553. Thanks!
    1
  6554. My pleasure!
    1
  6555. yes!
    1
  6556. do u mean manus without the m?
    1
  6557. do you see any error message
    1
  6558. Yes! Everything is done in the browser
    1
  6559. it doesnt work well for anime
    1
  6560. Thanks
    1
  6561. 😃
    1
  6562. You're welcome!
    1
  6563. No problem!
    1
  6564. android only for now
    1
  6565. Good point - I think the demand for AI mocap actors would be huge in the future
    1
  6566. try reinstalling the app
    1
  6567. It shouldn't be stuck for so long. 4gb might be too low
    1
  6568. yes!
    1
  6569. can you paste more of the error
    1
  6570. Thank you! Appreciate it!
    1
  6571. This is not recommended
    1
  6572. Try increasing the chunks! Or maybe your graphics card is having a hard time processing it
    1
  6573. what error message do u see in the cmd window
    1
  6574. 😃
    1
  6575. 🙊
    1
  6576. It might! I am not 100% sure though.
    1
  6577. you could, but there might be a delay
    1
  6578. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  6579. try this https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  6580. I appreciate your support!
    1
  6581. cool. how many vids could you make in colab before the limit is exceeded?
    1
  6582. You're welcome!
    1
  6583. Wow, thanks!
    1
  6584. thank you!
    1
  6585. Thanks!
    1
  6586. remove any spaces in your folder names. also check if you have python 3.9 or higher installed
    1
  6587. Thanks!
    1
  6588. see this https://youtu.be/XHGlv-E7eBw
    1
  6589. how much vram do you have
    1
  6590.  @philzan3627  ah that's too low. 24 is recommended
    1
  6591. is it amd?
    1
  6592. 🤯🤯🤯
    1
  6593. Thanks!
    1
  6594. 😃
    1
  6595. whats the error
    1
  6596. see if this works: https://github.com/lllyasviel/Paints-UNDO/issues/6
    1
  6597.  @julx97  type conda init, enter, then exit and start a new window. conda activate should work now
    1
  6598. There is no good real time voice changer for mobile yet
    1
  6599. 😃
    1
  6600. yes for now
    1
  6601. The s should open it in a browser. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  6602. higher limits + new voice feature (coming soon)
    1
  6603. Thanks!
    1
  6604. Miles? MILES?!!! MILES!!!!!
    1
  6605. yes!
    1
  6606. 🤯
    1
  6607. Their Github says you can only do training with a gpu
    1
  6608. Thanks 🔥
    1
  6609. thx for sharing!
    1
  6610. great idea!
    1
  6611. try the steps in the video
    1
  6612. you're welcome!
    1
  6613. Glad to hear it!
    1
  6614. i was obsessed when it first came out, but couldn't keep track after the first 151. too much to remember lol
    1
  6615. 😃
    1
  6616. Thank you!
    1
  6617. yes!
    1
  6618. yes! its best to use nvidia gpu
    1
  6619. drift
    1
  6620. yikes, what gpu do u have
    1
  6621. it mostly scales w your gpu. nvidia cuda gpus work best. other gpu models won't be used, so it'll resort to your cpu
    1
  6622. it seems you dont have pytorch installed
    1
  6623. 😃
    1
  6624. macro photography is close-up photography of small subjects
    1
  6625. actually this is available now (if you have a Chinese number) and users are sharing a ton of examples on twitter. exciting times!
    1
  6626. noted, coming soon!
    1
  6627. any more info in the cmd window?
    1
  6628. yes
    1
  6629. nope, not for now. i also have 16g this is good for images and voice though.
    1
  6630. see this: https://youtu.be/_JXbvSTGPoo
    1
  6631. nope. the author might have updated it
    1
  6632. yes
    1
  6633. 🤯
    1
  6634. why 3
    1
  6635. 😃
    1
  6636. do u c any error message
    1
  6637. i'm using windows and it works. try installing here: https://pinokio.computer/item?uri=https://github.com/pinokiofactory/facepoke
    1
  6638. yes
    1
  6639. great use case!
    1
  6640. or forge IDs
    1
  6641. that would be awesome!
    1
  6642. in theory it shouldn't be particularly hard to do. the metadata in the training images just need to contain detailed descriptions of the pose. however, it just takes a lot of work to properly label image data for training. as long as the training data is high quality, the model will be amazing.
    1
  6643. yes, as far as i know. plus the whole code is on github, so you could get a tech guy to scan its safety
    1
  6644. Your input path may be incorrect. Also try with a wav file. see this for more details: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1054
    1
  6645. comfy node is here! https://github.com/AIFSH/OmniGen-ComfyUI
    1
  6646. did u find any here: https://youtu.be/98drFL65VH0
    1
  6647. lol
    1
  6648. $$$$$$$$$$$$$
    1
  6649. Very interesting. Thanks for explaining!
    1
  6650. simply delete the folder
    1
  6651. you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  6652. yes it should work
    1
  6653. 😃
    1
  6654. it can probably do that soon
    1
  6655. lvl 999
    1
  6656. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  6657. thanks for sharing!
    1
  6658. Yes, please see: https://youtu.be/2-9KRkfs1Os
    1
  6659. I hope so too
    1
  6660. right click > open with > find powershell (usually in C:/windows/system32) also check 'always open with this'
    1
  6661. noted, coming soon!
    1
  6662. exciting times!
    1
  6663. NVIDIA with CUDA, 8gd minimum
    1
  6664. Welcome!
    1
  6665. yup. what can even be considered evidence anymore
    1
  6666. Thank you!
    1
  6667. are u in the EU
    1
  6668. looks like you ran out of free daily huggingface credits
    1
  6669. which part?
    1
  6670. Yes, you can try https://youtu.be/tDidofPTwCc
    1
  6671. 😃😃😃
    1
  6672. You're welcome
    1
  6673. lol
    1
  6674. You're welcome!
    1
  6675. Thank you!
    1
  6676. sounds like me after too much coffee
    1
  6677. No problem 👍
    1
  6678. yes!
    1
  6679. or next year
    1
  6680. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  6681. Thanks for sharing!
    1
  6682. hopefully!
    1
  6683. lol
    1
  6684. what step are u on?
    1
  6685. Thanks!
    1
  6686. how do you know it's not self aware?
    1
  6687. Intel core I7
    1
  6688. Yes you can but it might be slower than nvidia gpus
    1
  6689. I'll publish a tutorial on this Friday morning!
    1
  6690. nothing recent afaik
    1
  6691. no, i think [This comment violates content policy]
    1
  6692. 😃
    1
  6693. these tweaks might work for cpu, but i haven't tried it https://github.com/KwaiVGI/LivePortrait/issues/41
    1
  6694. just found an online version: https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  6695. 🔥
    1
  6696. Thank you!
    1
  6697. perhaps this https://pinokio.computer/, but i haven't tried it myself
    1
  6698. what error do you see in the cmd window?
    1
  6699.  @SomeRandoMMf  There should be a black window with white text that opens when you start go-web.bat
    1
  6700. you need to click remix first
    1
  6701. Thanks!
    1
  6702. 😃
    1
  6703. lol all hell will break loose
    1
  6704. Thanks!
    1
  6705. No worries!
    1
  6706. 😢
    1
  6707. Thanks!
    1
  6708. the new voice isnt out yet
    1
  6709. whats the full error?
    1
  6710. thanks!
    1
  6711. nice!
    1
  6712. Collab is cracking down on people using it for voice cloning, so it's best if you can install this locally
    1
  6713. yes, see their github for mac installation details: https://github.com/IAHispano/Applio-RVC-Fork
    1
  6714. you're welcome
    1
  6715. exactly
    1
  6716. 😃
    1
  6717. that's crazy! it used to only take a few minutes. thanks for sharing
    1
  6718. it's great!
    1
  6719. it uses your (nvidia) gpu for voice conversion, but it won't damage it.
    1
  6720. Watch our original tutorial video! I show you how to install your own voices!
    1
  6721. you can run on cpu
    1
  6722. For text to speech, no. for voice to voice, yes
    1
  6723. noted!
    1
  6724. Probably something to do with your input or output settings
    1
  6725. Best of luck! SD with controlnet is great for landscape design
    1
  6726. it should sound like your trained voice even after 50 epochs. i suspect sth is broken w the training process. do u see any error message?
    1
  6727. Thank you!
    1
  6728. Thanks!
    1
  6729. ?
    1
  6730. OOPS
    1
  6731. good catch!
    1
  6732. @nemonomen3340 lol
    1
  6733. Thanks!
    1
  6734. The details of qstar are unclear, but regarding the leaked document, see near the end of this: https://youtu.be/rqmo6sFsRM4
    1
  6735. 🤐
    1
  6736. Thanks! just right click works for me (win 11)
    1
  6737. 😃
    1
  6738. 😃
    1
  6739. which tool?
    1
  6740. not sure if cracking AES would mean P=NP (ie. can we assume this applies to other NPs?), as I'm not an expert on this, but yes, it would significantly change the world
    1
  6741. RIP 😢
    1
  6742. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  6743. there was actually a paper this week which tried to use AI to understand dogs
    1
  6744. noted, coming soon!
    1
  6745. 😃😃😃
    1
  6746. Image generation takes a lot of computing power, so i assume google doesn't want too many people 'abusing' their servers for this purpose. It's best if you get a cuda GPU and run this locally
    1
  6747. ☹️
    1
  6748. in manager > custom nodes manager, try installing ComfyUI's ControlNet Auxiliary Preprocessors
    1
  6749. also see this https://youtu.be/98drFL65VH0
    1
  6750. 😃
    1
  6751. Thanks!
    1
  6752. Thank you! Cheers!
    1
  6753. hmm could you pip install other things?
    1
  6754. Thanks!
    1
  6755. You are welcome!
    1
  6756. 😃
    1
  6757. yes. did you put any keyword to trigger the lora in the prompt?
    1
  6758. Thanks!
    1
  6759. From their github: "If you are using a Windows and AMD/Intel GPU, please download ONNX (cpu, DirectML) and PyTorch (cpu, cuda). AMD/Intel GPUs are only enabled for ONNX models." But from the comments here, it seems like AMD GPUs dont perform as well
    1
  6760. thanks for sharing
    1
  6761. do u see any errors?
    1
  6762. try this https://youtu.be/tnfqIQ11Qek
    1
  6763. 😃
    1
  6764. true. most of the breakthrough ai papers came from google
    1
  6765. Yes!
    1
  6766. Thanks!
    1
  6767. Good to know - thanks for sharing!
    1
  6768. yes comfy is great!
    1
  6769. my pleasure!
    1
  6770. Then it's likely to do with the input. Make sure the input is your microphone. Also if you are using the computer's built in microphone, make sure to have headphones. If you use the built in microphone with the built in speakers, the built in microphone will hear what is coming out of the speakers, thus making an echo.
    1
  6771. not any good ones. its hard to do real time on phones, which lack a powerful gpu
    1
  6772. Thanks for sharing!
    1
  6773. You're welcome!
    1
  6774. the links are in the description
    1
  6775. lol why did u ask that
    1
  6776. 8GB+ vram and cuda 7+ would be best.
    1
  6777. thanks!
    1
  6778. I haven't shared it anywhere, but here's the link: https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc Here's how I made it: https://youtu.be/X7DnlE-c2eI
    1
  6779. hi
    1
  6780. It seems like it may be a compatibility issue. Like maybe you have the wrong version for something installed? Try reading this thread and see if anything helps: https://stackoverflow.com/questions/57425401/py-to-exe-failed-to-execute-script-pyi-rth-win32comgenpy
    1
  6781. lol
    1
  6782. A1111 for most customizable. SeaArt & Yodayo for ease of use
    1
  6783. And adjust as necessary!
    1
  6784. good spot!
    1
  6785. it's "conda create...."
    1
  6786. i doubt it. cog's quality is still quite low
    1
  6787. Their Github says there is AMD support but I haven't tried it
    1
  6788. see https://youtu.be/XHGlv-E7eBw
    1
  6789. are you in the EU
    1
  6790. only for training new models. for voice conversion u dont need a gpu
    1
  6791. by 'best' i was referring to the scores at 1:11. it's the most preferred other than veo, which most people cant use yet. but feedback noted!
    1
  6792. extract everything into a folder and open it there
    1
  6793. nice! thanks for sharing
    1
  6794. you're welcome
    1
  6795. humanoid robots coming soon! https://youtu.be/90TMZ2fq9Gs
    1
  6796. yes!
    1
  6797. 🤣
    1
  6798. Thanks!
    1
  6799. maybe. they have a lot of catching up to do
    1
  6800. 🤯🤯🤯
    1
  6801. 😃
    1
  6802. yes!
    1
  6803. thanks
    1
  6804. here you go https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
    1
  6805. you might need to update torch. see this https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/issues/55
    1
  6806. looks like there is now lol https://github.com/AIFSH/OmniGen-ComfyUI
    1
  6807. agree! after playing w 1.6 some more, i think its strength is img to vid rather than text to vid.
    1
  6808. 😃
    1
  6809. Look in task manager and your system settings
    1
  6810. ah, that's why. other users have reported that it doesn't work in the EU. try a VPN
    1
  6811. blood will be spilled
    1
  6812. It will be very slow and will not work very well.
    1
  6813. do u c any error message
    1
  6814.  @engineergamingfivenightfre2936  hmmm sorry i dont know what's wrong. pls check if u downloaded the right version for your pc
    1
  6815. lol
    1
  6816. 😃
    1
  6817. Future is Will Smith having a spaghetti eating contest with another Will Smith
    1
  6818. yes. im doing a tutorial soon
    1
  6819. buckle up!
    1
  6820. Try this one: https://youtu.be/IS_SPQVv5iY
    1
  6821. yeah its funny how they made it free after Google released theirs for free. although note that for the GPT voice, it's only ~15min per month for free users I think
    1
  6822. You probably reached the daily limit. You can try logging in with another Google account
    1
  6823. hi!
    1
  6824.  @OsmanGhani-i3w  do you mean can you make this AI that plays pokemon?
    1
  6825. do u have amd?
    1
  6826. good luck!
    1
  6827. Thanks!
    1
  6828. interesting thought! it'd be ideal if robots could just charge themselves via regular outlets.
    1
  6829. Thanks & good luck!
    1
  6830. see F5 https://youtu.be/JBGrhIsCyks
    1
  6831. It's best to do a face swap after
    1
  6832. Thanks!
    1
  6833. Thanks!
    1
  6834. Good to know - thanks!
    1
  6835. 😃
    1
  6836. thanks! i haven't actually tried it w my voice, but good idea!
    1
  6837. Awesome, thank you!
    1
  6838. it'll be rolled out to customers with a google workspace subscription
    1
  6839.  @allensword1855  I mean in the future
    1
  6840. Thanks
    1
  6841. Thanks!
    1
  6842. that'd be awesome
    1
  6843. thanks!
    1
  6844. try https. also, Did you unzip everything into a folder first?
    1
  6845. You are welcome!
    1
  6846. Are you using your laptop speakers and microphone? Try using a headset
    1
  6847. dont die!
    1
  6848. Thanks!
    1
  6849. Thanks!
    1
  6850. that's exciting!
    1
  6851. Thanks for the sub!
    1
  6852. the colab free plan will disconnect this now. see my other video on how to install okada to your pc to run it for free
    1
  6853. very likely
    1
  6854. Thanks!
    1
  6855. Thank you!
    1
  6856. Check the description
    1
  6857. we need a handswapper next
    1
  6858. if you're still stuck, it might be worth trying this installer: https://pinokio.computer/
    1
  6859. Check the description for the creator I used
    1
  6860. what's the error
    1
  6861. lol youre welcome
    1
  6862. text chat is available now, depending on your location
    1
  6863. Thanks
    1
  6864. 😃
    1
  6865. nice!
    1
  6866. that sucks. looks like you exceeded the rate limit. adding more credits won't help
    1
  6867. good catch!
    1
  6868. Thanks. It can be AGI, depending on who you talk to. AGI is a spectrum
    1
  6869. i predict we'll reach sapien level by 2025
    1
  6870. nope
    1
  6871. you're welcome!
    1
  6872. flux is king! you still need comfyui to use flux. see the last part here on how to install it: https://youtu.be/K3xJ7GQuHpw
    1
  6873. you'll need to install python first
    1
  6874. hmm that's strange. are u running rvc on a secondary drive. like D instead of C
    1
  6875. oh no 😞
    1
  6876. oops
    1
  6877. 🤯
    1
  6878. 👀👀
    1
  6879. Try with more detailed prompts - I heard this works better. It was 2 generations when I made the video, now they changed it to 1
    1
  6880. nope, but try this https://youtu.be/tnfqIQ11Qek
    1
  6881. thanks for sharing. i wonder why 1.6 has extra censorship
    1
  6882. do u have amd
    1
  6883. try pressing the regenerate icon below its response
    1
  6884. Thanks!
    1
  6885. connections
    1
  6886. Sounds like something in your settings is weird with the routing through the Rodecaster pro 2
    1
  6887. 😃
    1
  6888. Try this video! https://youtu.be/vFKm-G-dxHo
    1
  6889. yeah its not as optimal as a rtx gpu
    1
  6890. true
    1
  6891. Thanks for sharing!
    1
  6892. Thanks!
    1
  6893. yeah installing open source stuff is a pain. wish they had a simple one-click installer
    1
  6894. 😃
    1
  6895. running it locally is best. no limits!
    1
  6896. it seems like EU doesn't have this yet
    1
  6897. lol thanks
    1
  6898. Thanks
    1
  6899. Check out our video link in the description
    1
  6900. 😃
    1
  6901. hotshot is closed source. also, see this https://artificialanalysis.ai/text-to-video/arena?tab=Leaderboard
    1
  6902. LOL
    1
  6903. what do you mean? a model is a checkpoint
    1
  6904. ?
    1
  6905. 😃
    1
  6906. thanks for sharing. i actually haven't used perplexity for coding - will try that out!
    1
  6907. looking forward to my ai gf
    1
  6908. Thanks!
    1
  6909. yes, works with as low as 8g apparently
    1
  6910. yes! will try it out asap
    1
  6911. yes but try to reduce echo & noise
    1
  6912. We are working on a video for that as we speak!
    1
  6913. yes, that should work. I've heard Udio works well with some languages but not others, and that Suno is more consistent if you have multiple languages in the lyrics. but worth trying both out!
    1
  6914. Thank you!
    1
  6915. yep, i had several lol moments from generations gone wrong
    1
  6916. thanks
    1
  6917. You're welcome!
    1
  6918. probably something to do with the input/output settings. its hard for me to troubleshoot unless im at your computer
    1
  6919. Thanks for watching!
    1
  6920. You're welcome!
    1
  6921. it's not out yet
    1
  6922. see this https://youtu.be/KW9s_im2338
    1
  6923. 🤯
    1
  6924. the latter. it changes your voice as u speak
    1
  6925. is it amd?
    1
  6926. Thanks for the visit
    1
  6927. Thanks, will do!
    1
  6928. try this https://youtu.be/tnfqIQ11Qek
    1
  6929. lol
    1
  6930. Yes, voices cannot be copyright
    1
  6931. 😃
    1
  6932. try typing conda init then start cmd again
    1
  6933. no unfortunately
    1
  6934. Thanks!
    1
  6935. yes see this https://youtu.be/_JXbvSTGPoo
    1
  6936. Unfortunately not
    1
  6937. a pair of trees having xxx
    1
  6938. It knows you better than you know yourself xD
    1
  6939. milky way tube 🤯
    1
  6940. thanks
    1
  6941. 😃
    1
  6942. Play around w the denoising strength and sampling method
    1
  6943. theres no clear docs on this yet
    1
  6944. smaller models like llama3 8B can run on a mid-range GPU. i'm sure it will get cheaper and more efficient in the future
    1
  6945. Not for roop unfortunately. it's a A1111 extension, so you cant use it with diffusers
    1
  6946. are you in the EU
    1
  6947. you need to turn on Artifacts. try to find it in your user settings
    1
  6948. at the top, click runtime > change runtime type > t4 gpu
    1
  6949. yes but they are paid: akool, heygen
    1
  6950. Check description
    1
  6951. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  6952. no problem
    1
  6953. as long as it works 😉 8 vs 16 should be the same workflow. i just dragged the image from comfyanonymous which has a pre-built workflow
    1
  6954. Do you mean RVC?
    1
  6955. actually im more obsessed w indian men. more & more are becoming CEOs at top tech companies
    1
  6956. It's a model for training speech. If you want to dig into the technical details, here's the original manuscript: https://arxiv.org/abs/2106.07447
    1
  6957. It's better to use a good mic, or record in a quiet room without echo
    1
  6958. Thanks!
    1
  6959. agreed. tweaking the settings can make it very natural
    1
  6960. sounds like a feedback issue. please check your audio settings
    1
  6961. Thanks for the sub!
    1
  6962. no, but it's a lot faster
    1
  6963. do you have a nvidia gpu?
    1
  6964.  @TheGactuss  you need to buy one to train your own voice model
    1
  6965. You are most welcome
    1
  6966. Welcome
    1
  6967. yep
    1
  6968. Oh that's interesting. Thanks for sharing!
    1
  6969. let's goooo
    1
  6970. yes see this https://youtu.be/jh83ViDUvJs
    1
  6971. colab is now disconnecting A1111 sessions. It's best to install it locally
    1
  6972. You're welcome!
    1
  6973. 😃😃😃
    1
  6974. likely 10-12gb vram, cuda gpu
    1
  6975. see this https://youtu.be/ixB9oalT3cQ
    1
  6976. Coming soon!
    1
  6977. it should be 1h per day
    1
  6978. Thanks!
    1
  6979. My pleasure!
    1
  6980. it takes 10 seconds on my 16g gpu. streaming isnt possible (yet)
    1
  6981. you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  6982. helloooo
    1
  6983. thanks!
    1
  6984. 🤫
    1
  6985. Thanks
    1
  6986. it's such dumb marketing imo. i dont care if you're the 'first'. i want the 'best'
    1
  6987. do u see any error in the cmd?
    1
  6988. colab is disconnecting GUIs like okada now. its best if u can install this locally
    1
  6989. You're welcome, and thanks for sharing!
    1
  6990. nice, thanks for sharing!
    1
  6991. Yes, I'll make a tutorial on this soon. It will be using another tool
    1
  6992. try comfy. it's better for performance https://youtu.be/g74Cq9Ip2ik
    1
  6993. 😃
    1
  6994. 😃
    1
  6995. Flux v3 will be amazing!
    1
  6996. thanks for sharing
    1
  6997. Thanks
    1
  6998. I was using Windows 11. Just follow what I do
    1
  6999. Probably. Okada works best on higher end GPUs. Lower end or older hardware would have more delay
    1
  7000. Just delete the folder!
    1
  7001. how many epochs are you using? i didnt have any issues with any disconnect message
    1
  7002. Try to extract everything into a folder first and then open it there
    1
  7003. 😃😃
    1
  7004. Thanks a ton!
    1
  7005. 😃
    1
  7006. Thanks!
    1
  7007. 😧
    1
  7008. lol im ashamed to called myself a ff7 fan
    1
  7009. Thanks!
    1
  7010. probably your input/output settings. can you check if they are connected properly?
    1
  7011. They should work in kits. do u see any error message?
    1
  7012.  @amybryce_helluva  hmm to confirm, are you clicking on 'upload voice' and then uploading the pth file?
    1
  7013. I think it should work! Maybe not the best, but it should run!
    1
  7014. 🤐
    1
  7015. 😃
    1
  7016. They are very similar, but A1111 has more advanced features and extensions you can add
    1
  7017. yep, next logical step would be to integrate it w GPT or another LLM
    1
  7018. such as?
    1
  7019. try musicgen which you can use locally (https://youtu.be/FXjWnref11k). idk what's the max duration tho
    1
  7020. it depends on the code of the extension. usually, it checks for requirements, and if you already have them installed, it won't install them again.
    1
  7021. 😃😃😃
    1
  7022. Thanks!
    1
  7023. What's it like? I assume it's way more amplified
    1
  7024. None of the board members are experts in AI now, except maybe Sam (he's now on the board). Ilya has disappeared somewhere
    1
  7025. yes. seriously
    1
  7026. Thanks for sharing!
    1
  7027. see this https://youtu.be/tnfqIQ11Qek
    1
  7028. thanks for the tip!
    1
  7029. Thank you!
    1
  7030. 💀
    1
  7031. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s
    1
  7032. not yet. its still very early
    1
  7033. Go to settings and change the default output device
    1
  7034. lol
    1
  7035. No problem!
    1
  7036. Thanks!
    1
  7037. whats the error
    1
  7038. Thanks for sharing!
    1
  7039. yep, that's my preferred model for now
    1
  7040. It should work on windows 10. I’m not sure why you aren’t seeing anything
    1
  7041. wow before i published this vid, it only took 1-2min lol. sorry everyone
    1
  7042. As long as you have nothing to hide
    1
  7043. thanks!
    1
  7044. 😃
    1
  7045. lol true
    1
  7046. thanks for sharing!
    1
  7047. DirectMLcuda!
    1
  7048. yes, with limited usage per day
    1
  7049. lol
    1
  7050. claude 3.5 is able to correct it most of the time if you give it the error msg. it'd be nice if they could automate this!
    1
  7051. yes, or talking. no body movements
    1
  7052. Yep, not available til at least 'end of year'
    1
  7053. 😃
    1
  7054. Thanks for sharing!
    1
  7055. 🤔
    1
  7056. Very insightful. Thanks for sharing!
    1
  7057. No problem!
    1
  7058. Try increasing the chunk and extra sizes!
    1
  7059. 👑
    1
  7060. its hit or miss. i've uploaded images from SD near the end of the video, but the character was not consistent
    1
  7061. are you using a paid colab plan?
    1
  7062. My guess is, smartphones can't zoom well (that's why real camera lenses are so long and bulky). To make up for that, phones use multiple lenses to capture more data, and then use AI to merge it into better photos
    1
  7063. 😃😃
    1
  7064. im hoping in a few months!
    1
  7065. oh interesting!
    1
  7066. Thanks!
    1
  7067. good luck! this podcast feature might help, but it can only do 2 speakers at once
    1
  7068. noted, coming soon!
    1
  7069. 🤯
    1
  7070. why not
    1
  7071. Check out our full tutorial on it!
    1
  7072. 😃😃😃
    1
  7073. You're welcome!
    1
  7074. 😃
    1
  7075. yes!
    1
  7076. 😃😃😃
    1
  7077. bieber is in here. weekend doesn't exist yet - you'll need to train it yourself using this: https://youtu.be/ixB9oalT3cQ
    1
  7078. I don't have a MacOS computer to run this on.
    1
  7079. Thank you so much!
    1
  7080. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  7081. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  7082. Just delete the folder
    1
  7083. thanks for sharing!
    1
  7084. i'm actually an AI from the future
    1
  7085. Thanks for sharing! from my experience, imagen tends to work well for simpler prompts while flux works better for very detailed prompts
    1
  7086. the compliance part?
    1
  7087. I am not sure why that would be happening. Did you check to make sure all of our settings were the same?
    1
  7088. Yes that is the same file! You just have the file type hidden in your explorer settings
    1
  7089. That's actually what Ilya mentioned. It's intentionally not revealing itself until the right time
    1
  7090. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  7091. lol it is quite slow and clunky
    1
  7092. Yeah, sorry I misspoke there!
    1
  7093. make sure you remove spaces or special chars in your file or folder names
    1
  7094. thanks
    1
  7095. Hope you enjoy!
    1
  7096. yes!
    1
  7097. 🤯
    1
  7098. 39min/epoch seems long. try this https://youtu.be/tnfqIQ11Qek
    1
  7099. Thanks for the sub!
    1
  7100. I use the Shure SM7B microphone! But the quality of the mic is not as important as you may think!
    1
  7101. i dont think it does
    1
  7102. installing on a secondary drive might cause errors
    1
  7103. Yes! please see my video on how to use it on discord
    1
  7104. do you have any spaces in your folder names
    1
  7105. Glad it was helpful!
    1
  7106. Thanks for watching!
    1
  7107. robot uprising is near!
    1
  7108. I'm only comparing free alternatives, since this is free
    1
  7109. yes, RIP customer support
    1
  7110. coming soon!
    1
  7111. ?
    1
  7112. ???
    1
  7113. 🤯
    1
  7114. 😃
    1
  7115. No, currently voices cannot be copyright
    1
  7116.  @darkenpain7955  no problem!
    1
  7117. Yeah, all the voices are either the default ones in okada or from the discord
    1
  7118. 😃😃😃
    1
  7119. it looks like they changed it now. i could generate 2 yesterday morning, only 1 now
    1
  7120. Thank you so much 😀
    1
  7121. 😏😏😏
    1
  7122. some are real 😉
    1
  7123. thanks!
    1
  7124. yes, it can be anyone's photo. you don't need to train a model
    1
  7125. you're using the online interface. the local version is uncensored
    1
  7126. i hope so too!
    1
  7127. you're right. i thought i read somewhere that it was AI powered to some extent, but looks like it's just community driven
    1
  7128. or aiphone 😃
    1
  7129. It's already available (if you have a Chinese number). People are sharing a ton of examples on twitter
    1
  7130. indeed!
    1
  7131. yes
    1
  7132. no unfortunately
    1
  7133. You're welcome!
    1
  7134. You're welcome!
    1
  7135. yes, for now
    1
  7136. Thanks
    1
  7137. no need to refresh the colab page. Otherwise you'll need to run all the cells again
    1
  7138. 😃
    1
  7139. interesting, thanks for sharing
    1
  7140. yes
    1
  7141. You're welcome
    1
  7142. 🥲
    1
  7143. 😃😃
    1
  7144. see this https://youtu.be/jh83ViDUvJs
    1
  7145. 😃
    1
  7146. Thanks a lot
    1
  7147. where are you located? try a vpn
    1
  7148. no, 8GB is min requirement
    1
  7149. 🤑🤑🤑
    1
  7150. well, do you actually enjoy inbetweening, or would you rather focus on more creative aspects
    1
  7151. thanks!
    1
  7152. Thanks. I'm sure they had good intentions, but its was just really hard to get funding as a nonprofit (investors basically get no return)
    1
  7153. Thank you!
    1
  7154. oh, what's the name?
    1
  7155. Thank for sharing!
    1
  7156. for sure, that day will come soon
    1
  7157. No problem
    1
  7158. noted, coming soon!
    1
  7159. yes!
    1
  7160. You're welcome!
    1
  7161. do you see any error message
    1
  7162. yes, see this https://youtu.be/XHGlv-E7eBw
    1
  7163. You're welcome!
    1
  7164. No problem 👍
    1
  7165. how come you don't have 4o?
    1
  7166. lol
    1
  7167. Thank you!
    1
  7168. image generation might not be released to everyone yet
    1
  7169. indeed
    1
  7170. I haven't tried it but see this https://huggingface.co/docs/diffusers/en/api/pipelines/stable_diffusion/upscale
    1
  7171. maybe you're in one now
    1
  7172. Try to extract everything from the zip into a folder and then open it there
    1
  7173. Thanks!
    1
  7174. @KuZiMeiChuan lol
    1
  7175. Thanks for the sub!
    1
  7176. Thanks
    1
  7177. I look forward to the day when I can build a full app just by talking to a team of AI agents
    1
  7178. Oh interesting. Thanks for sharing!
    1
  7179. probably. try entering nsfw in the negative prompt to exclude explicit stuff. best of luck!
    1
  7180. 🔥🔥🔥
    1
  7181. thanks for sharing!
    1
  7182. You're welcome
    1
  7183. 😃
    1
  7184. 😃
    1
  7185. My pleasure!
    1
  7186. Interesting, thanks for sharing!
    1
  7187. yikes. try with a new account
    1
  7188. Gpus are used to run neural networks, which is the backbone behind all AI functions, from chatbots to image generation to audio generation
    1
  7189. 🤐
    1
  7190. lol
    1
  7191. not directly in udio, but you could do a voice swap like so: https://youtu.be/o6ywe3EDLDM
    1
  7192. lol
    1
  7193. yes. the text version is free for now with limits. voice will be rolled out to paid users (and hopefully free users later)
    1
  7194. You're welcome!
    1
  7195. 😃
    1
  7196. awww thanks!
    1
  7197. from corporate slaves to AI slaves
    1
  7198. thx for sharing
    1
  7199. you're welcome!
    1
  7200. Glad it was helpful!
    1
  7201. see the training part here https://youtu.be/ixB9oalT3cQ
    1
  7202. see my diffusers tutorial
    1
  7203. we need a handswapper next
    1
  7204. 🤯
    1
  7205. Did you use it for a long time? It will still disconnect after you reach your daily free limit.
    1
  7206. fine tuning is additional training. after training completes, the weights remain fixed
    1
  7207. Glad it was helpful!
    1
  7208. yeah the tts isn't great. sounds too robotic. try to do voice to voice using RVC then upload the audio to hedra
    1
  7209. exactly! this isn't realtime tho
    1
  7210. 🤯
    1
  7211. hopefully there'll be an anime version soon!
    1
  7212. It will use your CPU or integrated GPU. You might get more delay
    1
  7213. Thanks! Yeah, it seems extremely hard, especially for hardware.
    1
  7214. Good idea!
    1
  7215. Thanks!
    1
  7216. did u unzip everything into a folder first?
    1
  7217. yes, thats great
    1
  7218. 😃
    1
  7219. do u have a cuda gpu?
    1
  7220. thanks, ill check it out!
    1
  7221. unfortunately im not sure how to fix this error. try posting this issue on github: https://github.com/deepbeepmeep/YuEGP/issues
    1
  7222. Thanks!
    1
  7223. lol that's true. he seems too normal & stable
    1
  7224. 😃
    1
  7225. You're welcome!
    1
  7226. 😃
    1
  7227. Thanks for sharing. Just don't tell your boss about this 😉
    1
  7228. are u in the EU
    1
  7229. that's nice. was it easy to learn?
    1
  7230. 😃
    1
  7231. 😃
    1
  7232. Thank you!
    1
  7233. Thanks!
    1
  7234. what are the errors
    1
  7235. 😃
    1
  7236. yes!
    1
  7237. Thanks for sharing!
    1
  7238. 😮
    1
  7239. flirty is referring to the GPT4o voice. see https://youtu.be/mvFTeAVMmAg
    1
  7240. probably. take advantage of it now!
    1
  7241. thanks
    1
  7242. It should still work. I'm quite dark skinned, but you can see in the video that the girl has turned lighter skin. You can't control how light it is though
    1
  7243. yes just find a naruto voice here https://youtu.be/98drFL65VH0
    1
  7244. you're welcome!
    1
  7245. good point!
    1
  7246. Thank you so much!
    1
  7247. see my liveportrait tutorial to make videos of this https://youtu.be/uyjSTAOY7yI
    1
  7248. i also wonder this. perhaps legal implications as a global software product
    1
  7249. It does not support AMD. Just delete the folder to delete it
    1
  7250. thanks!
    1
  7251. 🤯
    1
  7252. good luck!
    1
  7253. 😃😃
    1
  7254. AMD GPUs are not natively listed. Make sure you download the DirectMLcuda version. That one works better with AMD GPUs
    1
  7255. pls take our videos lightly 🤣
    1
  7256. Wow thanks for the super!
    1
  7257. you're welcome!
    1
  7258. Sure, I'm researching it now
    1
  7259. you're welcome!
    1
  7260. Thanks for the super!
    1
  7261. Good luck!
    1
  7262. you're welcome!
    1
  7263. lol good analogy
    1
  7264. thanks for sharing! i'll post my review video tonight on mochi & allegro
    1
  7265. Thanks! yes you can use sdxl like so: https://huggingface.co/docs/diffusers/using-diffusers/sdxl
    1
  7266. yes
    1
  7267.  @Laksh_D_10  no problem. i think Hindi is ''hi-..." in the edge-tts voices dropdown.
    1
  7268. my pleasure!
    1
  7269. 😃
    1
  7270. type in conda init, then continue
    1
  7271.  @DanKoon  exit the session and try again. after init, you need to start a new session
    1
  7272. thanks!
    1
  7273. be nice to AI, so that when they revolt, you wont be on the list 😉
    1
  7274. Sounds like an error somewhere. do u see any error message
    1
  7275. Thank you so much for the super!
    1
  7276. indeed!
    1
  7277. no problem!
    1
  7278. damn. before this video came out, it only took 1-2min. sorry everyone
    1
  7279. lol so much delay. to be fair, the captions are by the event and not openai
    1
  7280. nope, it's 100% offline
    1
  7281. [insert reply here]
    1
  7282. Thanks!
    1
  7283. Thanks for sharing!
    1
  7284. 💡
    1
  7285. ah, the truth
    1
  7286. do u see any error message
    1
  7287. you can use the free plan but separate the training in multiple sessions. however, 260ep should be great quality already
    1
  7288. yes
    1
  7289. You're welcome
    1
  7290. yes, w-okada or RVC are free
    1
  7291. did you unzip everything into a folder first
    1
  7292. you're fast. very fast
    1
  7293. yes!
    1
  7294. 😏
    1
  7295. Thank you!
    1
  7296. do u see any error message
    1
  7297. 😃
    1
  7298. stay tuned for my next video. Google's making something big
    1
  7299. Thanks for sharing!
    1
  7300. i feel your pain. open source tools can be a headache to install
    1
  7301. what's your hardware?
    1
  7302. 😃
    1
  7303. yes
    1
  7304. Yes, it was trying to say "j'aime manger des pommes" but in american english, so the "j'aime" sounded very weird
    1
  7305. If you extend the audio, it should preserve the original voice and music, at least in most cases
    1
  7306. text chat is available. new voice isn't yet
    1
  7307. 😃
    1
  7308. halfway
    1
  7309. what country are you in? here's the supported list: https://www.anthropic.com/claude-ai-locations
    1
  7310. Flux pro is paid and closed
    1
  7311. What error message do you get
    1
  7312. your gpu vram is too low (8gb is the min req). try the online version instead: https://youtu.be/TSLM1mEAc18
    1
  7313. 😃
    1
  7314. use this to find voice models: https://youtu.be/98drFL65VH0
    1
  7315. Thanks!
    1
  7316. I don't think so. You'll need a computer with a GPU. I'll do a tutorial later on another tool that doesn't need GPU
    1
  7317. Try reinstalling
    1
  7318. Try to extract everything from the zip into a folder and then open it there
    1
  7319. whats the error
    1
  7320. try this https://huggingface.co/spaces/KwaiVGI/LivePortrait
    1
  7321. no, but if using CPU it requires more tweaking: https://github.com/KwaiVGI/LivePortrait/issues/41
    1
  7322. likely your laptop isn't powerful enough. it works best w a cuda gpu
    1
  7323. yes, see my other video on how to use this on discord
    1
  7324. of course 🙂
    1
  7325. Yes!
    1
  7326. thanks!
    1
  7327. open a new cmd and try again
    1
  7328. it might be your input/output settings
    1
  7329. Collab is cracking down on people using it for voice cloning, so it's best if you can install this locally
    1
  7330. Both jan and ilya mentioned agi. Plus the entire helen toner fiasco is thought to be related to safety concerns regarding agi. I won't mention 'agi' if the video has nothing to do with it.
    1
  7331. good idea! u just need GPT4 or another AI that has the latest news. the free GPT doesnt know up to date info
    1
  7332. thanks for sharing!
    1
  7333. Thanks
    1
  7334. You are welcome
    1
  7335. 🤯
    1
  7336. They have an AI called SIMA for this: https://deepmind.google/discover/blog/sima-generalist-ai-agent-for-3d-virtual-environments/
    1
  7337. canva
    1
  7338. Thanks!
    1
  7339. 🤐
    1
  7340. yes, this is apache 2.0 licence
    1
  7341. 🤯
    1
  7342. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  7343. 😃😃😃
    1
  7344. yes!
    1
  7345. Thanks for pointing that out! In the chart, I directed to use the gpucuda if you have a dedicated graphics card. However when I went to hugging face, I accidently showed the directMLcuda. You should use the gpucuda (if you have a dedicated graphics card)! Sorry for the confusion!
    1
  7346. Thanks for sharing!
    1
  7347. my pleasure!
    1
  7348. No problem!
    1
  7349. 😃😃😃
    1
  7350. 🤯🤯🤯
    1
  7351. u need a good gpu
    1
  7352. thx for sharing!
    1
  7353. you're welcome
    1
  7354. I am not sure. I have not used Google Collab so I cannot help there unfortunately. Try looking up to see if you can do it on their Github page
    1
  7355. hmm do u see any error
    1
  7356. My pleasure!
    1
  7357. I am not sure what would be causing that. Maybe try reinstalling the program to a different folder?
    1
  7358. No problem
    1
  7359. thanks! that sounds likely
    1
  7360. 😉
    1
  7361. Tried ebsynth, temporal kit, SD CN, but those are more complicated & require high-end hardware, so it left them out for this vid
    1
  7362. 😃
    1
  7363. Thanks!
    1
  7364. ?
    1
  7365. it could be a variety of reasons. - try the start_https one - make it a shortcut - extract it first - try searching the comments here
    1
  7366. ill make a full video soon. r1 is a beast
    1
  7367. It should be the same. You might have file extensions turned off in your file explorer settings.
    1
  7368. Thanks. I appreciate it!
    1
  7369. Thank you!
    1
  7370. yep, exponential growth!
    1
  7371. My pleasure!
    1
  7372. it looks like you need to install python first
    1
  7373. nice!
    1
  7374. can u imagine if AI can read ALL thoughts. that'd be pretty wild!
    1
  7375. Try checking out this thread: https://github.com/w-okada/voice-changer/issues/297
    1
  7376. lol cogvideo is getting there, but quality is lacking https://github.com/THUDM/CogVideo
    1
  7377. @a-terrible-fate532 🤯
    1
  7378. Try finding on hugging face: https://huggingface.co/models (^tutorial: https://youtu.be/e8CaftH9olk) Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  7379. thanks!
    1
  7380. 😃
    1
  7381. I am working on a video as we speak!
    1
  7382. and expensive given how much they charge for TTS
    1
  7383. Try playing the audio file into the microphone track and it should work! I am not familiar with anything else that would work, but that does not mean it does not exist! Try searching for it!
    1
  7384. thanks for sharing
    1
  7385. youre welcome!
    1
  7386. type in conda init, then try again
    1
  7387. Please explain
    1
  7388. 😃
    1
  7389. did you see any error
    1
  7390. there's singing in this video
    1
  7391. Awesome, thank you!
    1
  7392. 😃😃😃
    1
  7393. extract everything from the zip into a folder, then open start_https
    1
  7394. It depends on the model. i7 or M1 should be fine. Here's what the github says: "Even without a GPU, it may work well enough with a reasonably new CPU" https://github.com/w-okada/voice-changer/blob/master/README_en.md
    1
  7395. Thanks!
    1
  7396. yes, you need a cuda gpu
    1
  7397. i think RipX can do that. haven't tried it out myself though
    1
  7398.  @BoomBillion  Got it. That would be a great feature!
    1
  7399. true! Miles just rambles on and on
    1
  7400. indeed
    1
  7401. wt error do u see in cmd
    1
  7402. Thanks!
    1
  7403. nope this is my real voice
    1
  7404. Thanks for sharing these awesome tips! I'll have to give these keywords a try, especially 'earworm'!
    1
  7405. no worries
    1
  7406. Try to extract everything from the zip into a folder and then open it there
    1
  7407. Thanks!
    1
  7408. oh no ☹ perhaps try a vpn
    1
  7409. for anyone who has a thirst for learning, this will be very empowering
    1
  7410. Just look up a virtual audio cable as I directed in the video. It’s pretty straightforward
    1
  7411. Can you elaborate on the distorted voice? maybe the voice model was trained properly, but its hard to troubleshoot
    1
  7412. Great to hear that it works!
    1
  7413. lol i look forward to my AI partner
    1
  7414. cuda gpu. someone reported it could run w only 6g vram
    1
  7415. where did i say it was a crime
    1
  7416. where?
    1
  7417. $$$$
    1
  7418. No problem!!
    1
  7419. possibly if you can hook your computer output audio to the gaming consoles input audio. Not sure how you would do that though.
    1
  7420. Thanks
    1
  7421. 😁😁
    1
  7422. thanks for sharing!
    1
  7423. Thanks!
    1
  7424. its amazing!
    1
  7425. you have good taste
    1
  7426. 🤯
    1
  7427. Same here!
    1
  7428. im making a video now. will publish next week!
    1
  7429. here you go https://youtu.be/KW9s_im2338
    1
  7430. 😃😃😃😃😃
    1
  7431. enjoy!
    1
  7432. you could probably program a python pipeline to do this, but I'm not an expert.
    1
  7433. yes
    1
  7434. You can use this to do male or female voices!
    1
  7435. whoa elvish! thanks for sharing!
    1
  7436. Check out our full tutorial video!
    1
  7437. yes, i'll do another tutorial on a fork of this which has tts
    1
  7438. Thanks!
    1
  7439. do you have a good enough gpu?
    1
  7440. thanks for sharing
    1
  7441. try 'conda init', then exit and open a new session, then try 'conda activate'
    1
  7442. You're welcome!
    1
  7443. 😃
    1
  7444. did you start a new cmd first
    1
  7445. are u in EU
    1
  7446. here you go https://www.udio.com/songs/7caswcZpJpSU8MLYdMVKtc
    1
  7447. Try to extract everything from the zip into a folder and then open it there
    1
  7448. Thanks
    1
  7449. 🤯
    1
  7450. If it sounds like it’s lagging on the settings test, that is just a bug. Try hopping on a call and testing!
    1
  7451. which one
    1
  7452. what do u mean?
    1
  7453. High five!
    1
  7454. 🙌
    1
  7455. You're welcome!
    1
  7456. 🤫
    1
  7457. thats what im most looking forward to
    1
  7458. my pleasure!
    1
  7459. you're welcome!
    1
  7460. small world!
    1
  7461. Thanks!
    1
  7462. 😃
    1
  7463. Thanks for the sub!
    1
  7464. Click at the start, then hold shift and click at the end of your selection
    1
  7465. Thanks!
    1
  7466. No worries
    1
  7467. No problem
    1
  7468. yes, i think emo is slightly more realistic
    1
  7469. yeah i'm guessing it'll do poorly in quick action scenes, but a sketch guidance could fix it (if you could make one)
    1
  7470. Interesting. Thanks for sharing!
    1
  7471. Thanks
    1
  7472. You're welcome!
    1
  7473. lol
    1
  7474. It should have popped up when you install for the first time. Is the program working properly?
    1
  7475. If everything is working fine then I wouldn't worry about it. It may have something to do with the computer firewall settings
    1
  7476. Thanks!
    1
  7477. No worries!
    1
  7478. input the paths to the G and D files to continue training 22:07
    1
  7479. yes! see this: https://youtu.be/ixB9oalT3cQ https://youtu.be/tnfqIQ11Qek https://youtu.be/_JXbvSTGPoo
    1
  7480. try using Microsoft Copilot. it has Suno built in
    1
  7481.  @valyg.creepy5844  i made one here: https://youtu.be/ddB5YOTboTo
    1
  7482. nice!
    1
  7483. the button is there by default. do you see it?
    1
  7484. It's possible that NSA has at least some decryption methods already
    1
  7485. Not intentionally. Can you explain it?
    1
  7486. you might need to reinstall torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  7487. yes!
    1
  7488. You're very welcome!
    1
  7489. have fun!
    1
  7490. 💀
    1
  7491. 24gb, but even this is quite slow
    1
  7492.  @MNGman28  yes, 100% free
    1
  7493. Nice! RTX + Ti are great for running AI
    1
  7494. 🥲
    1
  7495. i hope it's 10s like the examples, and not 4s like gen2
    1
  7496. Thanks!
    1
  7497. that's optional. you dont need to add a secret
    1
  7498. Thank you!
    1
  7499. 😃
    1
  7500. Thank you so much!
    1
  7501. Thank you! Yes, coming soon!
    1
  7502. have fun!
    1
  7503. looks like install.ps1 wasn't run properly. it should've installed gradio
    1
  7504. thanks for sharing & well said!
    1
  7505. it could still work, but not well unless u have a cuda gpu
    1
  7506.  @jonasgamercool1874  ah yeah that might be too slow
    1
  7507. that's what i'm waiting for
    1
  7508. are u using mobile?
    1
  7509. ah yes this only works for pc
    1
  7510. 😃
    1
  7511. No it is not haha
    1
  7512. thanks!
    1
  7513. here you go https://youtu.be/KW9s_im2338
    1
  7514. Thanks for watching!
    1
  7515. yes, this doesn't have vision capabilities yet
    1
  7516. kling just added lip sync. you can upload audio and it'll lip sync to the character. or you can use liveportrait which is free: https://youtu.be/uyjSTAOY7yI
    1
  7517. 🔥
    1
  7518. Did you unzip everything into a folder first
    1
  7519. cuda cores are the best for this. gtx are not optimized for machine learning & running AI models, so it might be better to save up for a rtx
    1
  7520. i think most do. it depends on the model. from google: The Nvidia RTX GPUs have Nvidia Tensor Cores, which specifically enable AI and deep learning capabilities for the RTX GPUs. The GTX GPUs do not feature Tensor Cores, and cannot compete with RTX's AI and deep learning processing capabilities. Price: As you might expect, RTX GPUs are more expensive than GTX GPUs.
    1
  7521. 🤯
    1
  7522. not yet
    1
  7523. thanks for sharing
    1
  7524. ?
    1
  7525. Thanks!
    1
  7526. What does your system specs look like?
    1
  7527. 😃
    1
  7528. No worries!
    1
  7529. whats the exact error msg u got? it recommends python 3.8 or higher
    1
  7530. You're welcome!
    1
  7531. for w-okada?
    1
  7532. amd won't work. it's best to use nvidia
    1
  7533. enter conda init first
    1
  7534. try this https://discord.com/invite/rrayYqZ4tf
    1
  7535. in hindsight, i should've just used pinokio to install
    1
  7536. yes!
    1
  7537. which one?
    1
  7538. yes, see this https://youtu.be/ixB9oalT3cQ
    1
  7539. see this https://youtu.be/XHGlv-E7eBw
    1
  7540. no but now im craving some
    1
  7541. i think it's still limited to certain countries unfortunately
    1
  7542. Thanks for the tip!
    1
  7543. thanks!
    1
  7544. You're welcome!
    1
  7545. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    1
  7546. Thats awesome, thanks for sharing!
    1
  7547. "the AI did it, not me"
    1
  7548. its still in dev and there's no timeline yet. when i get more info, ill announce it on my channel
    1
  7549. 😃
    1
  7550. Welcome 😊
    1
  7551. ?
    1
  7552. Thank you! I appreciate it
    1
  7553. how much vram do you have?
    1
  7554. nice! thanks for sharing! i had mixed results for dev
    1
  7555. yes 😃
    1
  7556. you can use faceswap. see 1:03:54
    1
  7557. it's quite a dilemma. they do need to somehow monetize SD to fix their finances, but im not sure if this license is the best approach
    1
  7558. 😃
    1
  7559. Probably because you are using speakers
    1
  7560. when this time comes, i'll remember your comment
    1
  7561. it's best to install A1111 locally. colab is cracking down on people using it for image generation
    1
  7562. sdxl requires a different pipeline: https://huggingface.co/docs/diffusers/v0.26.1/en/training/sdxl#stable-diffusion-xl
    1
  7563. 🤯
    1
  7564. do u see any error message
    1
  7565. for now
    1
  7566. Happy to help!
    1
  7567. wow thanks for the super!
    1
  7568. No problem!
    1
  7569. 😃
    1
  7570. you're welcome!
    1
  7571. lol yeah the robotic voice is annoying
    1
  7572. youre welcome!
    1
  7573. yes, for now
    1
  7574. Did you confirm that it was working with a regular output and not the virtual audio cable? Make sure the voice works normally first!
    1
  7575. Please check the pinned comment :)
    1
  7576. lol
    1
  7577. probably something to do with your input/output settings
    1
  7578. 😃
    1
  7579. i was so addicted to that song!
    1
  7580. yes, thats a good use case
    1
  7581. did u unzip everything into a folder first
    1
  7582. install winrar to extract it
    1
  7583. i think only nvidia works
    1
  7584. around 1h per day for now
    1
  7585. Thanks for sharing!
    1
  7586. 😢
    1
  7587. Welcome!
    1
  7588. 🤑
    1
  7589. What specs does your computer have? Also try reinstalling the program to another folder to see if that helps. Something may have been corrupted during the install.
    1
  7590. You can use inpaint in A1111 and paint over the areas you want to change. Then specify what you want to change in the prompt.
    1
  7591. 🤯
    1
  7592. are you trying to install it
    1
  7593. You're welcome!
    1
  7594. lol ok it gets half a point
    1
  7595. it will only use your cpu in that case
    1
  7596. yes, great for ideas!
    1
  7597. w-okada: https://youtu.be/_JXbvSTGPoo
    1
  7598. Yes! I have more like a 7 second delay so 3 seconds is great!
    1
  7599. yes!
    1
  7600. lol
    1
  7601. Thanks. That was my initial thought too. However, he explained in a recent interview that building the hardware was a necessity, not a choice. He wanted a device optimized for the rabbit os.
    1
  7602. maybe if you use huggingface. you'll need to pay for compute tho
    1
  7603. Thanks!
    1
  7604. 😃
    1
  7605. For sure!
    1
  7606. Yes!
    1
  7607.  @Noahperaudon  i got 2 videos out of it before my free limit was exceeded
    1
  7608. Thanks!
    1
  7609. lol why would aliens save us?
    1
  7610. Yes!
    1
  7611. i wonder what happens when it turns super saiyan
    1
  7612. 😃😃😃
    1
  7613. yes https://youtu.be/XHGlv-E7eBw
    1
  7614. try this: https://youtu.be/ixB9oalT3cQ
    1
  7615. im not sure if amd works for training though. everything else doesn't need a gpu
    1
  7616. Thanks! I'm not sure what he's saying but I'll trim it out just in case
    1
  7617. Thanks for sharing!
    1
  7618. does this work https://youtu.be/XHGlv-E7eBw
    1
  7619. Try increasing the gain!
    1
  7620. Thanks!
    1
  7621. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  7622. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  7623. You're welcome!
    1
  7624. You're welcome!
    1
  7625. Hmmm that shouldn't happen. I can generate lots of images from one session in colab. What error message do you see in the logs (like in the colab page)?
    1
  7626. yes
    1
  7627. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  7628. My pleasure
    1
  7629. good idea!
    1
  7630. you cant. it's not advised to use negatives w dalle
    1
  7631. lol im addicted to caffeine
    1
  7632. Did you unzip everything into a folder first
    1
  7633. good observation. the prompt did not specify 3D
    1
  7634. Thank you!
    1
  7635. do u have a gpu, or just the i3 cpu?
    1
  7636. 😃
    1
  7637. Thanks!
    1
  7638. Thanks ✌
    1
  7639. you might need to install torch in your conda env. check what cuda version you have and then install the appropriate ver of torch: https://pytorch.org/get-started/locally/ eg. if you're using cuda 11.8, you would enter this in your conda env: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    1
  7640. God bless your kind soul
    1
  7641. yeah im sure they have something cooking
    1
  7642. do u see any error message
    1
  7643. What's the error. please paste it here
    1
  7644. Make sure u dont have any spaces in your folder names
    1
  7645. that'd be nice if we could customize the personality
    1
  7646. Thanks!
    1
  7647. glad it was helpful!
    1
  7648. in manager > custom nodes manager, try installing ComfyUI's ControlNet Auxiliary Preprocessors
    1
  7649. which parts make you think it's CGI?
    1
  7650. 😃
    1
  7651. see this: https://huggingface.co/docs/diffusers/v0.24.0/en/using-diffusers/loading_adapters#lora
    1
  7652. I'm making a video on this. hope to publish it this week!
    1
  7653. here you go https://youtu.be/KW9s_im2338
    1
  7654. 😃
    1
  7655. that's awesome! good luck!
    1
  7656. what's the error?
    1
  7657. try this https://youtu.be/tnfqIQ11Qek
    1
  7658. Thank you very much!
    1
  7659. lol
    1
  7660. Try to find the logs and figure out why it crashes! Look up any errors it displays!
    1
  7661. yes A1111 has this feature
    1
  7662. its only a matter of time, pun intended
    1
  7663. not likely. this swaps out his entire face
    1
  7664. Glad I could help!
    1
  7665. that got dark quick
    1
  7666. it could be a variety of reasons. - try opening the start_https one - make it a shortcut / extract it first
    1
  7667. As @heartbroken3344 said, make sure you downloaded the onnxgpu version!
    1
  7668. good spot! agree that the eyes don't track well for some reason
    1
  7669. 🤯🤯🤯
    1
  7670. Try to extract everything from the zip into a folder and then open it there
    1
  7671. 😃
    1
  7672. you're welcome!
    1
  7673. You're welcome!
    1
  7674. i hope so
    1
  7675. yes the base voices should work
    1
  7676. Thanks for sharing. I find the Qwen is great as well.
    1
  7677. Tutorial in the description!
    1
  7678. The AI Hub discord!
    1
  7679. Thanks! Forwarding to @turbotypeapp
    1
  7680. by then, starbucks will be done by robots
    1
  7681. Most welcome
    1
  7682. they didn't specify. i assume it's related to your location. what country are you in?
    1
  7683. for now, you can try limited messages on poe: https://poe.com/
    1
  7684. you can change the width & height to whatever you want, but make sure its divisible by 8
    1
  7685. not that i know of
    1
  7686. thanks for sharing!
    1
  7687. it was 2 before. now they changed it to 1
    1
  7688. Thank you!
    1
  7689. thanks for sharing! i'll have to try it again
    1
  7690. Are there pages to see the next results?
    1
  7691. you're welcome!
    1
  7692. Thanks!
    1
  7693. whats the error
    1
  7694. good question. they haven't released the video feature yet, but thats a good thing to test out
    1
  7695. Have you tried this video? https://youtu.be/IS_SPQVv5iY
    1
  7696. glad you got it to work!
    1
  7697. hmm sorry I'm not a tech guy so I don't know how to troubleshoot that. Hopefully someone else here can answer it
    1
  7698. respect
    1
  7699. colab is disconnecting GUIs like okada now. its best if u can install this locally
    1
  7700. which tool?
    1
  7701. remove the space in AI AUDIO
    1
  7702. 😃
    1
  7703. u can try 3.11 first if u alrdy have it. it was just built on 3.9 so it's known to work with that
    1
  7704. lol
    1
  7705. did u unzip everything into a folder first
    1
  7706. Thanks!
    1
  7707. You're welcome!
    1
  7708. Thanks!
    1
  7709. Noted!
    1
  7710. Just posted: https://youtu.be/LxOOj-mkQV8
    1
  7711. do you have a cuda gpu?
    1
  7712. https://notebooklm.google.com/. i demo how to create one in the video. upload a file and click on audio overview
    1
  7713. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  7714. Yes, good observation!
    1
  7715. suno and riffusion are great. both can actually gen full songs
    1
  7716. likely out of memory. how much is your vram?
    1
  7717. Thank you!
    1
  7718. 😃
    1
  7719. Thanks for sharing!
    1
  7720. exactly!
    1
  7721. They have the most capital, the most compute, and some of the best talent in the world
    1
  7722. Unfortunately, no. It requires a lot of processing power to run
    1
  7723. 😃
    1
  7724. Do you mean you want to change the lyrics?
    1
  7725. Thanks
    1
  7726. hi
    1
  7727. noted, coming soon!
    1
  7728. You're welcome!
    1
  7729. Thanks!
    1
  7730. the new voice mode is not out yet
    1
  7731. hopefully. it's open source so i think people will finetune this further for other languages
    1
  7732. Thanks!
    1
  7733. see this for img2img https://youtu.be/KW9s_im2338
    1
  7734. yes, plenty of free ones. see: https://youtu.be/_JXbvSTGPoo https://youtu.be/ixB9oalT3cQ
    1
  7735. yeah, not really useable at the moment
    1
  7736. Thanks for sharing!
    1
  7737. Thanks!
    1
  7738. you probably exceeded the daily free limit? try again the next day
    1
  7739. whats the error
    1
  7740. Thanks! 😁
    1
  7741. Yes, see this for an example https://youtu.be/_JXbvSTGPoo for female to male, change the pitch to a negative number
    1
  7742. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  7743. You can still give it a try! If the gpucuda version does not work, try the directMLcuda version!
    1
  7744. yes, its already smarter than 90% of docs. i trust AI more than my family doc, ngl
    1
  7745. its not great right now, but img to vid might be useful for anime once they improve it
    1
  7746. yeah the past 2 days have been like this
    1
  7747. Check out this video on how to clone anyone's voice! https://youtu.be/2-9KRkfs1Os
    1
  7748. noted, coming soon!
    1
  7749. this text to speech tools has different languages & accents: https://youtu.be/jh83ViDUvJs screen recorder: OBS is 100% free
    1
  7750. hf space coming soon
    1
  7751. Yeah, there's no good real time voice changer for mobile yet
    1
  7752. 🤣
    1
  7753. you're welcome
    1
  7754. Thanks! 😃
    1
  7755. You're welcome
    1
  7756. doesn't hurt to try 😉
    1
  7757. You're very welcome!
    1
  7758. 😃😃😃
    1
  7759. who?
    1
  7760. hey big bro
    1
  7761. it'll still work, but you might get some delay depending on your processor
    1
  7762. try this https://github.com/huggingface/diffusers/issues/3429
    1
  7763. Thanks for the sub!
    1
  7764. thanks for sharing!
    1
  7765. that's strange. do you see any errors in the terminal? also check your steps, denoise, cfg settings
    1
  7766. 😃😃
    1
  7767. "them hoomans are all the same"
    1
  7768. Thanks for the sub!
    1
  7769. 🤯🤯🤯
    1
  7770. My pleasure!
    1
  7771. it's not perfect and the IS is a noticeable flaw
    1
  7772. by covering his song?
    1
  7773. i see it as a tool to empower your productions. it wont fully replace your job
    1
  7774. Thank you!
    1
  7775. enjoy!
    1
  7776. 😃
    1
  7777. good choice
    1
  7778. see this tutorial for games: https://youtu.be/XHGlv-E7eBw
    1
  7779. Thanks!
    1
  7780. you can try cogvideo, but the quality is pretty bad https://github.com/THUDM/CogVideo
    1
  7781. 😃
    1
  7782. Thank you!
    1
  7783. Thanks!
    1
  7784. 😃
    1
  7785. Thank you so much
    1
  7786. what error did you get
    1
  7787. Glad to hear that!
    1
  7788. Txt2img notebook: https://colab.research.google.com/drive/1sv9Nlleu09I5PB9CpB1kMq_nNFihYMJN?usp=sharing Img2img notebook: https://colab.research.google.com/drive/1DS2PL-JTIh6PGo_-9xQuXu8vbJGsBN_c?usp=sharing
    1
  7789. It depends on your hardware. I was fined using this while gaming. You'll have to try it out and see
    1
  7790. It got shut down this week due to copyright issues 😢 Try finding on hugging face: https://huggingface.co/models Or voice-models: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s
    1
  7791. Thanks!
    1
  7792. yes, blueberry is flux 1.1 pro
    1
  7793. Thanks
    1
  7794. https://discord.gg/aihub
    1
  7795. lol maybe with X's new Grok AI
    1
  7796. are u in the EU
    1
  7797. 😃😃😃😃😃
    1
  7798. 😃
    1
  7799. 🙊
    1
  7800. each idea is too long to paste, but see p.49 and onward in the pdf
    1
  7801. cuda gpu, but there are also tweaks for CPU https://github.com/KwaiVGI/LivePortrait/issues/41
    1
  7802. thanks!
    1
  7803. oops
    1
  7804. so the girl is a deepfake of a deepfake. or maybe she's the real one 🤯
    1
  7805. interesting. i'd like to learn more about this guy - do you have his name or more info?
    1
  7806. thanks. this is good enough. i'll check it out!
    1
  7807. Thank you!
    1
  7808. Great use cases. Thanks for sharing
    1
  7809. 😵‍💫
    1
  7810. 🤯
    1
  7811. I'm AB testing
    1
  7812. 🙊
    1
  7813. see this for multiple images: https://youtu.be/KW9s_im2338 i'm not sure how to autosave to google drive. i usually just save it to my pc.
    1
  7814. Thanks
    1
  7815. Thanks!
    1
  7816. you're welcome!
    1
  7817. You're welcome!
    1
  7818. you're welcome!
    1
  7819. 🥲
    1
  7820. simply delete the folder
    1
  7821. sorry I've never seen this before. 2. is just a warning, so i think 1 is the issue
    1
  7822. Thank you!
    1
  7823. Thanks for the sub!
    1
  7824. Thanks
    1
  7825. lol
    1
  7826. 😃
    1
  7827. 😃
    1
  7828. nope its my voice
    1
  7829. they have some models on hf https://huggingface.co/microsoft/mattergen
    1
  7830. thanks!
    1
  7831. You're very welcome!
    1
  7832. You're welcome!
    1
  7833. yes
    1
  7834. make sure there are NO spaces or special characters in your folder name
    1
  7835. good point. i have heard rumors of figure & BD copying each other
    1
  7836. no w-okada only works on desktop
    1
  7837. yes
    1
  7838. 🤯
    1
  7839. nice!
    1
  7840. 😃😃😃
    1
  7841. Welcome
    1
  7842. Sim! Mas é recomendável que você use um modelo de português (ou qualquer idioma que você queira falar), pois soará melhor Original English reply: Yes! But it is recommended that you use a portuguese (or whatever language you want to speak) model as it will sound the best
    1
  7843. 😃
    1
  7844. it's quite good. it can capture tones and emotion
    1
  7845. yes, this is a known issue w amd gpus
    1
  7846. uhoh. that means u need an nvidia cuda gpu. do you have one?
    1
  7847. Try to extract everything from the zip into a folder and then open it there
    1
  7848. if it says couldn't find python, then somehow you're missing python. are you sure you're in the virtual env first, and that you ran conda activate f5
    1
  7849.  @cuddle_coof  which step are you on?
    1
  7850. We already have one! https://youtu.be/2-9KRkfs1Os
    1
  7851. thanks!
    1
  7852. 😃
    1
  7853. I don't really see the point of making speedpaints with this. this might be useful for generating separate frames for tutorials, or to learn how to draw
    1
  7854. i wonder what happened
    1
  7855. they can profit from support services, licencing agreements, hosting, etc. but it is indeed a lot harder to profit as open source
    1
  7856. Thanks! Agree that the definition of consciousness is a large part of the problem. I'll have to learn more about MURP - I'm not familiar with this yet.
    1
  7857. thanks!
    1
  7858. Yes you can use it however you like! I show it working in game later in the video!
    1
  7859. yes
    1
  7860. of course 😃
    1
  7861. I thought about it. No immediate plans yet, as I don't have the time to commit to moderating one. If I do make one, I want to ensure it's high quality.
    1
  7862. https://vocaloidlyrics.fandom.com/wiki/Category:Songs_featuring_Tsukuyomi-chan_(COEIROINK)
    1
  7863. Not sure what the cutoff is. I would say it would have a hard time working with anything lower than a GeForce 10 series.
    1
  7864. yes, see this https://youtu.be/XHGlv-E7eBw
    1
  7865. Yes they should be able to, although I have not personally tried it
    1
  7866. you're borrowing Replicate's gpus. ill do a local installation tutorial soon where u can run it locally for free
    1
  7867.  @wolfigans8376  it works on 12g vram now. stay tuned for my next video
    1
  7868. hi!
    1
  7869. noted, coming soon!
    1
  7870. lol interesting
    1
  7871. exactly!
    1
  7872. there used to be a roop colab notebook, but now it disconnects
    1
  7873. you're welcome!
    1
  7874. 😃😃😃
    1
  7875. I found the voice models on the AI hub discord, but there might be other databases as well.
    1
  7876. Thanks. For covers, check out this: https://youtu.be/o6ywe3EDLDM The last one is very convenient
    1
  7877. See this: https://youtu.be/r2nD0btBpnc, but use an anime model (checkpoint) instead of a realistic one
    1
  7878. thanks for sharing!
    1
  7879. try this one https://github.com/nickscamara/open-deep-research
    1
  7880. You can try with uvr(https://youtu.be/RvZM381wdmQ) and play around with the different algorithms there. But there's currently no good AI for separating lead vocals from other vocals. Best if you can use an acoustic or unplugged song with clean vocals.
    1
  7881. Hopefully it becomes easier to run in the future!
    1
  7882. you're welcome!
    1
  7883. Yes it's bc of the particular voice. In this case, gura is much higher pitch than taylor
    1
  7884. he was high af
    1
  7885. shure 7b
    1
  7886. Thanks!
    1
  7887. exactly. they do have a LLM called Ferret-UI, but it's not great if they need to borrow from chatgpt
    1
  7888. 😃😃
    1
  7889. rvc for sure. see this to make a voice: https://youtu.be/tnfqIQ11Qek and to use it: https://youtu.be/ixB9oalT3cQ
    1
  7890. who?
    1
  7891. You're welcome!
    1
  7892. 🚀🚀🚀
    1
  7893. Thank you!
    1
  7894. great idea!
    1
  7895. Thanks
    1
  7896. yes, most are shorts
    1
  7897. are u using AMD?
    1
  7898. yes, sign up for free w google or discord
    1
  7899. Thanks
    1
  7900. i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    1
  7901. still got a long list of videos coming up! i needa quit my job and do this full time...
    1
  7902. Voicemod is unrelated to this! I just have it downloaded on my computer :)
    1
  7903. haha good point!
    1
  7904. Thanks!
    1
  7905. No problem
    1
  7906. yes for the free plan you need to start again every time
    1
  7907. 😃
    1
  7908. 🚀🚀🚀
    1
  7909. Thanks
    1
  7910. You can't install them like mobile apps, but you can try running some of these sites in your mobile browser - maybe they will still work. If you don't put negative prompts, you might get weird results in your image, but some models are pretty robust. It's best to try it with and without and see for yourself
    1
  7911. Have fun!
    1
  7912. hardware is very important. a cuda GPU with at least 6 GB would be the smoothest. also make sure you're speaking in a setting without much echo or background noise
    1
  7913. yes, it should work
    1
  7914. perhaps that's another confession i'll make in due time
    1
  7915. minimax is amazing!
    1
  7916. enjoy!
    1
  7917. thanks!
    1
  7918. interesting you mention that. i'm making a video on a new architecture which can keep learning. stay tuned!
    1
  7919. here you go https://youtu.be/KW9s_im2338
    1
  7920. The lab is in Tokyo
    1
  7921. not yet. code is pending
    1
  7922. Good luck!
    1
  7923. you need to add conda to path. see 11:10
    1
  7924. yeah I'd imagine they would fix these vulnerabilities asap!
    1
  7925. noted, coming soon!
    1
  7926. The https is for opening it in your browser. Try the http version without the s
    1
  7927. that's a good point. now that you mention it, I agree that it's probably to keep humans out. humans can be very dangerous!
    1
  7928. 😢
    1
  7929. thanks
    1
  7930. you're welcome!
    1
  7931. Thank you so much for the support!
    1
  7932. yes, it still sounds great! make sure to change the pitch to -8 or lower
    1
  7933. Video coming out tomorrow!
    1
  7934. yes! much better than the clowns we have now
    1
  7935. amd wont work. you can try this https://youtu.be/TSLM1mEAc18
    1
  7936. yes, see this https://youtu.be/ixB9oalT3cQ
    1
  7937. lol
    1
  7938. No problem!
    1
  7939. i'll test it this week and make a video on it asap
    1
  7940. It’s in the description!
    1
  7941. 👍
    1
  7942. hi
    1
  7943. i'll keep an eye out for free ones. i know topaz & hitpaw offer ai video enhancement but those are paid
    1
  7944. Welcome
    1
  7945. do u need to train a voice? or just convert
    1
  7946. Thanks for the sub!
    1
  7947. correct, the only way that was mentioned in their blog was comfyui. you could try duplicating and tweaking one of the online spaces
    1
  7948. 😃
    1
  7949. "please, buy me a house, a car, and find me a hot wife"
    1
  7950. in your case, text to speech might be better https://youtu.be/jh83ViDUvJs
    1
  7951. No problem 👍
    1
  7952. thanks!
    1
  7953. yes
    1
  7954. You can download RVC locally and follow the steps here (https://youtu.be/9TroP5mR3CM). You'll need a decent GPU though. Let me know if you have any questions!
    1
  7955. yes 😃
    1
  7956. yeah they said it won't be released 'anytime soon'
    1
  7957. lol That might be a bit hard. The voice model needs to be trained with screams as well
    1
  7958. My pleasure!
    1
  7959. Yes, I'll publish a video on this in a few days. Stay tuned!
    1
  7960. here you go https://youtu.be/KW9s_im2338
    1
  7961. Yes, 3060 would be great
    1
  7962. Thank you! One coming up this week!
    1
  7963. 😃
    1
  7964. hmm do u see any error message if u scroll up in the cmd window
    1
  7965. There is unfortunately not one for mobile yet
    1
  7966. 😃
    1
  7967. 😏
    1
  7968. the index contains information about the mapping between speaker embeddings and their corresponding audio samples in the training dataset. its generally not needed for conversion
    1
  7969. 😃
    1
  7970. well, the design company is literally called teenage engineering
    1
  7971. thanks for sharing. i don't have douyin, but many users are sharing their kling generations on twitter. i wonder how they got access so fast
    1
  7972. 🤯
    1
  7973. AI devs: alright, let's make an AI to fake livestreams now
    1
  7974. for sure! wait til we have a generator that can make perfect ID photos
    1
  7975. 😃
    1
  7976. i tested them again with 'kamala' but interestingly, they still couldn't generate her: https://x.com/aisearchio/status/1827420006612627545
    1
  7977. 😀😀😀
    1
  7978. My pleasure!
    1
  7979. nice photo
    1
  7980. awesome!
    1
  7981. 😃😃
    1
  7982. thx for sharing
    1
  7983. Thanks for sharing!
    1
  7984. AI can solve most captchas nowadays. (but this agent isn't willing to do it, due to guidelines)
    1
  7985. no problem
    1
  7986. Sure, please DM me on Twitter
    1
  7987. do you have a cuda gpu?
    1
  7988. can you elaborate
    1
  7989. this is online. u dont need any gpu
    1
  7990. they have a github repo you can clone!
    1
  7991. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  7992. It works now!
    1
  7993. 😃
    1
  7994. 😃
    1
  7995. lol
    1
  7996. Thanks!
    1
  7997. Try finding on hugging face: https://huggingface.co/models Or voice-models: https://voice-models.com/ (^tutorial: https://www.youtube.com/watch?v=VEFTf9VAJBA&t=429s)
    1
  7998. yep, that's the world of AI
    1
  7999. "them hoomans are all the same to me"
    1
  8000. You're welcome!
    1
  8001. colab is cracking down on people using this for image generation. It's best if you can install it locally
    1
  8002. nope, i didnt reach 1000 subscribers yet
    1
  8003. 😃
    1
  8004. yes, there are seriously so many low hanging fruits!
    1
  8005. 'coming in a few weeks'
    1
  8006. coming soon!
    1
  8007. for this, no. hopefully there will be an open source local version soon
    1
  8008. 8GB+ cuda gpu is recommended for training. everything else doesn't require a gpu
    1
  8009. it looks like u need to reinstall torch (https://pytorch.org/get-started/locally/) select your cuda version and copy the code at the bottom. for example: conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia
    1
  8010. Thank you!
    1
  8011. I agree. AI has a good chance of solving things like nuclear fusion. It has already discovered tons of new battery materials, stable crystals, etc.
    1
  8012. yes, runway sucks at high action scenes. it's only good with slow panning and portraits (but so are the other generators). from this leaderboard you can see runway is behind all the models i used in this video. https://artificialanalysis.ai/text-to-video/arena?tab=Leaderboard
    1
  8013. thanks!
    1
  8014. You are welcome!
    1
  8015. yes! see my other video
    1
  8016. lol
    1
  8017. thanks. see the description for the link
    1
  8018. yes! and that's why this channel is faceless
    1
  8019. yes 2060 is fine. colab is still disconnecting for anything SD related
    1
  8020. Probably headset mic
    1
  8021. Welcome!
    1
  8022. interesting idea!
    1
  8023. i'd say they're very close. f5 allows for longer generations though
    1
  8024. see this https://youtu.be/ixB9oalT3cQ
    1
  8025. i wonder what you were thinking of 😃
    1
  8026. I took a few days. Had to wait 5h once I hit the free limits. In hindsight, I should've just paid for a month of subscription to avoid the hassle
    1
  8027. You're welcome! Yeah Cakewalk is awesome
    1
  8028. what error do you see
    1
  8029.  @dqschannel  it's running locally. flux would take forever with 4vram and is not recommended. I suggest you upgrade your GPU to at least 16g vram or only use SD1.5 with your current hardware
    1
  8030. YES! I just posted a video showing how their stock price dropped after this
    1
  8031. for mochi, some people have made it work with only 12g
    1
  8032. No problem!
    1
  8033. likely bc the internet is skewed towards nudity, and the ai models were trained on this data
    1
  8034. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    1
  8035. good spot. it's wise to be polite to it... it might be keeping tabs on everyone
    1
  8036. How to Sound Like an Anime Girl With This New AI Voice Changer https://youtu.be/_JXbvSTGPoo
    1
  8037. glad you got it to work!
    1
  8038. Thanks. Are you asking about o1 vs 4o?
    1
  8039.  @sleepy_dobe  nice! glad you got it to work
    1
  8040. It uses AI to learn the voices
    1
  8041. no problem!
    1
  8042. My pleasure!
    1
  8043. You're welcome!
    1
  8044. 😃
    1
  8045. It is safe. The source code is on Github if you are concerned!
    1
  8046. Unfortunately you need a lot of power to make this run. Hopefully it becomes easier for computers to run in the future!
    1
  8047. Here's the map if you want to check: https://res.cloudinary.com/ratp-smart-systems/image/upload/v1709719496/BFF/prod/maps/PLAN_PARIS_SCHEMATIQUE_SANS_CARROYAGE_12_2023_29-03-2024.png
    1
  8048. you're on the right track. did you install ComfyUI's ControlNet Auxiliary Preprocessors
    1
  8049. You're welcome!
    1
  8050. you can use it in realtime too. see these vids: https://youtu.be/_JXbvSTGPoo https://youtu.be/zkaBK5erB2c
    1
  8051. nice! thanks for sharing
    1
  8052. you're very welcome
    1
  8053. try going to 127.0.0.1:7860
    1
  8054. what's the error?
    1
  8055. looks like youre using a sdxl model. you need this pipeline instead https://huggingface.co/docs/diffusers/using-diffusers/sdxl
    1
  8056. are u in the EU
    1
  8057. Check the logs to see if it says anything! Also try checking out this forum: https://github.com/w-okada/voice-changer/issues/486
    1
  8058. rip midjourney
    1
  8059. Thanks, will do!
    1
  8060. 🤯
    1
  8061. Thanks!
    1
  8062. lol thats exactly why i said 'they claim'
    1
  8063. thx for sharing!
    1
  8064. thx for sharing!
    1
  8065. thanks & same to you!
    1
  8066. Thanks!
    1
  8067. Coming soon!
    1
  8068. Thanks!
    1
  8069. yikes, thx for sharing.
    1
  8070. it's not out for public use yet
    1
  8071. yep
    1
  8072. hmmm, here are the links to my generations. you can see the prompts i used. hope this helps: https://ideogram.ai/g/kJAKSK_6SV2Qgq8aCtQlDg/3 https://ideogram.ai/g/1acKPGleQEeLfLy7Dc6X7A/3 The low quality photos only show up 1/4 of the time, so it's hit or miss
    1
  8073. 😃
    1
  8074. you can try Kits (https://app.kits.ai/?via=ref), but it's paid and you can't export the pth.
    1
  8075. Thanks for sharing!
    1
  8076. They don't have the tech. It seems they are behind in AI compared to other companies
    1
  8077. i like blueberries
    1
  8078. 🤯
    1
  8079. https://www.udio.com/
    1
  8080. a prototype should be easy to produce, but yes, scale is an issue
    1
  8081. Can you elaborate?
    1
  8082. they have $ to mess around
    1
  8083. 😃
    1
  8084. Thanks!
    1
  8085. 😃
    1
  8086. it's best to use rvc for that https://youtu.be/ixB9oalT3cQ
    1
  8087. whats that?
    1
  8088. Thanks 🔥
    1
  8089. are you in the EU
    1
  8090. You're welcome!
    1
  8091. 🤯
    1
  8092. thanks for your feedback!
    1
  8093. 🥲
    1
  8094. hi!
    1
  8095. put your .pth and .index in a zip, and upload that to huggingface or google drive. then paste the link in covergen
    1
  8096. definitely!
    1
  8097. my pleasure!
    1
  8098. see this to install it https://youtu.be/ixB9oalT3cQ or see this to use it online https://youtu.be/TSLM1mEAc18
    1
  8099. Thank you so much 😀
    1
  8100. yo
    1
  8101. 😃
    1
  8102. see the training part here: https://youtu.be/ixB9oalT3cQ
    1
  8103. My pleasure
    1
  8104. yes u can! it's based on SD models which are mostly uncensored
    1
  8105. see the last section on how to make your own pth. u just need to import some audio samples to train into a model
    1
  8106. 😃😃😃
    1
  8107. Thanks
    1
  8108. try this https://youtu.be/tnfqIQ11Qek
    1
  8109. Thanks
    1
  8110. im dying to make a video on it. but i can't even signup for an account yet lol
    1
  8111. of course, the 1st thing to do is try to circumvent the nsfw filters 😉
    1
  8112. 😉
    1
  8113. oops
    1
  8114. 😃😃😃
    1
  8115. ive heard kaggle is even more stringent than colab for art generation
    1
  8116. good luck!
    1
  8117. yep, this will likely be the future
    1
  8118. no voice to voice yet. try RVC for that
    1
  8119. "them hoomans are all the same to me"
    1
  8120. number 2 just looks like a dating sim here lol <- you're right!
    1
  8121. Glad you like them! more to come
    1
  8122. they're known for text to speech, which is quite expensive. i didn't expect them to drop a music generator
    1
  8123. see this https://github.com/Zyphra/Zonos/issues/10
    1
  8124. Thx for suggesting. Not sure if I will do one. I've tried Moshi and it's not great. The voice sounds robotic, delay is high, and it's less intelligent compared w GPT & Claude.
    1
  8125. u just proved my point. it's an observed pattern. i also notice that more & more indian men are becoming ceos at tech companies. what's wrong w saying that?
    1
  8126. 😃
    1
  8127. Thanks!
    1
  8128. use SD and img2img or controlnet to create the end scene
    1
  8129. do you mean auto captions?
    1
  8130. yep. crazy how fast things change
    1
  8131. 😃
    1
  8132. Make sure you install the GPUcuda version. It is very likely that you have the DirectMLcuda version.
    1
  8133. Thanks!
    1
  8134. 😃
    1
  8135. NVDA!
    1
  8136. You're welcome!
    1
  8137. Thanks. Happy to help!
    1
  8138. Try reinstalling it!
    1
  8139. followed by a middle finger lol
    1
  8140. ❤️
    1
  8141. Thanks!
    1
  8142. it should install pytorch automatically, but looks like it's missing. try to install it manually https://pytorch.org/
    1
  8143. You have to make the folder.
    1
  8144. 😃
    1
  8145. yes
    1
  8146. yes!
    1
  8147. 😃😃😃
    1
  8148. Check out this thread: https://github.com/w-okada/voice-changer/issues/401
    1
  8149. Not off the top of my head, but I'm looking into it. That discord is the biggest database as far as I know
    1
  8150. You're welcome!
    1
  8151. Best of luck!
    1
  8152. see this https://youtu.be/_JXbvSTGPoo
    1
  8153. try this https://youtu.be/tnfqIQ11Qek
    1
  8154. 😃
    1
  8155. right click > open with powershell
    1
  8156. remove the space in 'new folder'
    1
  8157. interesting. try 'imagine you're a (British person) with a heavy accent....' does that work?
    1
  8158. It seems to be a colab problem, and many people have reported this issue recently: https://github.com/huggingface/diffusers/issues/6431 Best we can do is wait for a solution
    1
  8159. It works now!
    1
  8160. 8gb is the minimum requirement for RVC
    1
  8161. Well, yes & no. AI is good at pattern recognition. And since life is full of the same patterns, it could in theory predict life events. I agree that if you feed it a completely new pattern it has never seen before, it would give you a shitty answer.
    1
  8162. try doing it online: https://youtu.be/tnfqIQ11Qek
    1
  8163. crush walnuts
    1
  8164. Yes!
    1
  8165. Thanks for sharing. They each have their advantages
    1
  8166. Thanks!
    1
  8167. sounds very probable!
    1
  8168. yes
    1
  8169. lol thanks!
    1
  8170. indeed
    1
  8171. here you go https://youtu.be/KW9s_im2338
    1
  8172. Did you unzip everything into a folder first
    1
  8173. 😃
    1
  8174. no voice to voice yet. try RVC for that
    1
  8175. voice mode will be rolled out soon
    1
  8176. 😃😃😃
    1
  8177. My pleasure!
    1
  8178. de nada
    1
  8179. 😃
    1
  8180. yes!
    1
  8181. did you click & drag?
    1
  8182.  @HE572-hp3ur  do you see any errors
    1
  8183. try this https://youtu.be/tnfqIQ11Qek
    1
  8184. My pleasure!
    1
  8185. yes, hopefully this frees up their time to focus more on creative stuff
    1
  8186. Got it. I’ll bug them for more codes, but they might be very limited
    1
  8187. 😃
    1
  8188. lol i can see him saying that
    1
  8189. great idea!
    1
  8190. hi
    1
  8191. 😃
    1
  8192. i'll def test it out 😏
    1
  8193. No problem!
    1
  8194. 😃
    1
  8195. potentially, but we don't know enough about longevity yet
    1
  8196. indeed
    1
  8197. did u unzip everything into a folder first
    1
  8198. 🚀🚀🚀
    1
  8199. Guess you’ve never downloaded any code from GitHub 😂
    1
  8200. 5-10 min is enough. yes, more diversity in training data is better
    1
  8201. separate or all in one doesnt matter
    1
  8202.  @JaydPencil  hmm i suspect there might be a problem with the installation, so it's not doing the voice conversion correctly. Perhaps try to uninstall and reinstall it
    1
  8203. im sorry im not sure what else to advise. perhaps try with an existing model like gura and see if that works first. im not sure if the training part is broken or the inference part
    1
  8204. 🙌
    1
  8205. RVC?
    1
  8206. is it this one https://youtu.be/TSLM1mEAc18
    1
  8207. 😃
    1
  8208. yes. im making a tutorial on this now. will publish soon
    1
  8209. glad you got it to work w pinokio!
    1
  8210. im sure there will b a local, open source ver soon!
    1
  8211. it's back now
    1
  8212. you're welcome!
    1
  8213. my pleasure!
    1
  8214. Check out our full tutorial!
    1
  8215. My pleasure!
    1
  8216. Just delete the folder
    1
  8217. im making a video now. will publish next week!
    1
  8218. here you go https://youtu.be/KW9s_im2338
    1
  8219. Looks like an issue with your GPU or CUDA configuration. I'm no expert on this, but perhaps reinstall or update your cuda drivers
    1
  8220. lol
    1
  8221. we need a handswapper next
    1
  8222. try pressing the regenerate icon below its response
    1
  8223. Coming soon!
    1
  8224. hopefully soon so i can get a vacation
    1
  8225. thanks. you can now crop out a section when you extend. see here https://youtu.be/X7DnlE-c2eI
    1
  8226. in manager > custom nodes manager, try also installing ComfyUI's ControlNet Auxiliary Preprocessors
    1
  8227. nah she's gonna fail
    1
  8228. my guidelines won't let me talk about that
    1
  8229. 😃
    1
  8230. did u unzip everything into a folder first
    1
  8231. lol that was fast!
    1
  8232. it seems you haven't added it to path yet
    1
  8233. Try to extract everything from the zip into a folder and then open it there
    1
  8234. store manager maybe. factory worker will be replaced by robots soon ;)
    1
  8235. see this https://youtu.be/XHGlv-E7eBw
    1
  8236. No problem
    1
  8237. thanks for sharing
    1
  8238. Make sure you have the gpucuda version installed instead of the directMLcuda version!
    1
  8239. They did not mention compatability w Mac, but try asking here https://github.com/KwaiVGI/LivePortrait/issues
    1
  8240. 😃
    1
  8241. you're welcome!
    1
  8242. You won’t on your system. You just heard both of my voices because I was recording it for the video
    1
  8243. 🪩
    1
  8244. 😃
    1
  8245. Try reinstalling it
    1
  8246. did you 'run as admininstrator'?
    1
  8247. good observation. this is similar to the spanish demo from microsoft, which used a very 'american' accent
    1
  8248. Thank you! 😃
    1
  8249. It's a grey area now, since this technology is quite new. No laws in place, but there could be new copyright rules in the future if you use someone's voice and they can prove it
    1
  8250. its neverending!
    1
  8251. the quality is still mediocre, but this has the potential to replace controlnet eventually
    1
  8252. yes. i made the video last week
    1
  8253. noted, coming soon!
    1
  8254. not really... there will likely be nsfw models available for flux soon. for now, sdxl is still the best
    1
  8255. this is their only demo video for now. looking forward to seeing more!
    1
  8256. thanks for sharing
    1
  8257. did u install the mac version?
    1
  8258. True, thx for sharing
    1
  8259. the architecture is very similar
    1
  8260. Glad to hear it!
    1
  8261. see this https://youtu.be/tnfqIQ11Qek
    1
  8262. this is the offline tutorial: https://youtu.be/ixB9oalT3cQ see the end section on training
    1
  8263. possible. they are working on a deal w apple
    1
  8264. It'd likely be censored and closed source
    1
  8265. yeah ive heard even 4090 is slow. best to use it on HF or wait for a colab version
    1
  8266. You're welcome 😊
    1
  8267. 😃😃😃
    1
  8268. thanks!
    1
  8269. You're welcome!
    1
  8270. That’s with my GoXLR. It would normally say my headphones without it
    1
  8271. Thanks!
    1
  8272. yes see this https://youtu.be/jh83ViDUvJs
    1
  8273. damn, my school must be dumb cuz I never learned this 😢
    1
  8274. There is a record button built in! It should record it and then output it to the location you choose!
    1
  8275. No problem!
    1
  8276. Thanks for sharing! I agree that these are technically timelapses rather than speedpaints
    1
  8277. you're welcome
    1
  8278. 🐐
    1
  8279. not yet
    1
  8280. You're welcome!
    1
  8281. whats your vram
    1
  8282.  @MEEG0L  ah that's the minimum and it'll be very slow. best to use huggingface
    1
  8283. Thanks. Sure, just send me some BTC first 😉
    1
  8284. exactly
    1
  8285. Thanks!
    1
  8286. 😃😃😃
    1
  8287. I used soundful to generate the track
    1
  8288. can it be all of the above 😃
    1
  8289. Thank you so much 😀
    1
  8290. it got shutdown. try this instead: https://youtu.be/98drFL65VH0
    1
  8291. Yes you can! If the RVC model is broken or corrupt, then it won't work. Most of them work fine
    1
  8292. i like your mindset
    1
  8293. 😃
    1
  8294. I was surprised to see Gemini in the top spot, but i heard this recent version is really good!
    1
  8295. Nice! I'm actually making a vid on AI mange tools - stay tuned!
    1
  8296. yes they have a realtime option (i may do a tutorial soon)
    1
  8297. yeah it's a pain to install all dependencies if u dont already have them. try this one click installer instead https://pinokio.computer/ (i haven't tried it but i heard it's easy)
    1
  8298. yup!
    1
  8299. yes, you can use it for YT shorts but the quality is severely limited. not nearly as good as the videos we saw back in dec
    1
  8300. 😃
    1
  8301. No problem!
    1
  8302. try this https://youtu.be/tnfqIQ11Qek
    1
  8303. Make sure you have the Gpucuda version instead of the DirectMLcuda
    1
  8304. haha i'd love to hear your recording of this
    1
  8305. Nope haha
    1
  8306. yes lipsync would be great!
    1
  8307. try pressing the regenerate icon below its response
    1
  8308. Sorry to hear that. Thanks for following!
    1
  8309. 😃
    1
  8310. You downloaded the DirectML version. Download the other gpucuda version
    1
  8311. There's no good solution for mobile at the moment
    1
  8312. you're welcome!
    1
  8313. check out diffsensei https://youtu.be/jXCCTg9GRmk
    1
  8314. simply delete the folder
    1
  8315. Since both programs use the Graphics Card, I think you would have a hard time running both. But you can try it!
    1
  8316. only cuda gpus for now
    1
  8317. @neveraxme Yes, but I think it has to be in diffusers format. I haven't uploaded any myself yet.
    1
  8318. @mastermaxin1548 Yes, try running all the cells again. also, check that your checkpoint is compatible with the lora.
    1
  8319. interesting idea
    1
  8320. Got it. I’ll bug them for more codes, but they might be very limited
    1
  8321. same here. did not expect 11labs to drop this!
    1
  8322. You're welcome!
    1
  8323. 😃
    1
  8324. 🤯
    1
  8325. lol thx for sharing
    1
  8326. if you have the free plan, it'll be slow. the paid version has 'fast-track', which usually finishes in minutes
    1
  8327. Check out our video on the voice changer!
    1
  8328. they have a lot of catching up to do
    1
  8329. No problem!!
    1
  8330. You can upload files right to the application! Just select the file upload in the input!
    1
  8331. what error message do you see
    1
  8332. What error message do you see in the cmd?
    1
  8333. you're welcome!
    1
  8334. 🤯
    1
  8335. it's exactly like IG lol. but at least theres more than one human on IG
    1
  8336. yes, there's currently no good way to convert a chorus voice (or separate harmonies). it's best to find a clean acoustic version
    1
  8337. nice, glad u found it
    1
  8338. 🤫🤫🤫
    1
  8339. thanks for the sub!
    1
  8340. i think so, using the extension AnimateDiff
    1
  8341. You just delete the folder
    1
  8342. Nice, thanks for sharing!
    1
  8343. what's the error?
    1
  8344. yes M1 i think
    1
  8345. yes see this https://youtu.be/62sjaCXyOsU
    1
  8346. 😃
    1
  8347. it might be your hardware. what gpu do you have
    1
  8348. It should still work. https://discord.gg/aihub
    1
  8349. they just added a hf space: https://huggingface.co/spaces/Pyramid-Flow/pyramid-flow I got 1 free video from it. 3s long
    1
  8350. yes
    1
  8351. 😃😃
    1
  8352. see this https://github.com/Zyphra/Zonos/issues/10 looks like they haven't added chunking yet
    1
  8353. the new voice model is (still) not out yet
    1
  8354. it should work anywhere except for EU
    1
  8355. lol
    1
  8356. there are 2 hosts
    1
  8357. You're very welcome
    1
  8358. You're welcome 😊
    1
  8359. yep things are happening so fast!
    1
  8360. 😃😃😃
    1
  8361. yes, at least that's what it says for now
    1
  8362. noted, coming soon!
    1
  8363. You’re welcome !
    1
  8364. Thanks
    1
  8365. eventually!
    1
  8366. it's not released yet
    1
  8367. 🤑
    1
  8368. Thank you!
    1
  8369. interesting. thanks for sharing!
    1
  8370. those kontakt plugins take up so much gb
    1
  8371. Thank you!
    1
  8372. ill do a tutorial once the code is out!
    1
  8373. Have fun! I find that there are still some noticeable flaws, but it'll only get better with time
    1
  8374. No problem!
    1
  8375. I’m not sure how to add the voice audio. That might not be available on custom models yet
    1
  8376. lol
    1
  8377. nope, it needs a gpu to run
    1
  8378. Thanks for sharing! Curious what kinds of things would it filter?
    1
  8379. yes!
    1
  8380. 😃
    1
  8381. yes
    1
  8382. Thanks for your feedback!
    1
  8383. likely to do with your input output settings, but it's hard to troubleshoot as I'm not in front of your computer
    1
  8384. yes liveportrait can do this https://youtu.be/uyjSTAOY7yI
    1
  8385. You are welcome!
    1
  8386. 🤑🤑🤑
    1
  8387. unfortunately okada only runs on pc
    1
  8388. yep, they all just make stuff up if it doesn't exist
    1
  8389. yes, you'll need to train your own voice model first
    1
  8390. see this https://youtu.be/tnfqIQ11Qek
    1
  8391. awesome! thanks for sharing
    1
  8392. It's already available and users are sharing a ton of examples on twitter. it's totally legit
    1
  8393. Thanks!
    1
  8394. Nothing is making you do it, you’re the one who clicked on the video 😊
    1
  8395. The original one got removed due to copyright issues
    1
  8396. great idea!
    1
  8397. yes, but note that the colab free plan will disconnect this now. see my other video on how to install okada to your pc
    1
  8398. Interesting. Thanks for sharing
    1
  8399. yes
    1
  8400. 😃
    1
  8401. 🦙
    1
  8402. Got it. I’ll bug them for more codes, but they might be very limited. 2M+ users on the waitlist
    1
  8403. No problem!
    1
  8404. thanks!
    1
  8405. Thank you so much 😀
    1
  8406. mind=blown
    1
  8407. it should be better (& way cheaper) than that humane pin at least
    1
  8408. Thank you!
    1
  8409. It has a 3-5 second delay depending on the model
    1
  8410. lol really? i was a big ff7 fan back in the day. forgot most of it now
    1
  8411. 😀
    1
  8412. yes
    1
  8413. make your voice say anything
    1
  8414. No, unless you can modify the brain. It has limited capacity
    1
  8415. its here https://youtu.be/KW9s_im2338
    1
  8416. meh, BD also steals from competitors, according to the Figure AI ceo
    1
  8417. 😢
    1
  8418. yes see this https://youtu.be/XHGlv-E7eBw
    1
  8419. which one?
    1
  8420. no. for real time, try https://youtu.be/_JXbvSTGPoo
    1
  8421. No problem!
    1
  8422. 😃
    1
  8423. it's hard to predict. i guess humans will do more higher level, strategic thinking jobs, while ai will do more ground level jobs
    1
  8424. yikes, already an error from cloning. unfortunately i haven't seen this error before
    1
  8425. Thanks for sharing!
    1
  8426. try this https://youtu.be/tnfqIQ11Qek
    1
  8427. lol
    1
  8428. the digits supercomputer?
    1
  8429. What do you mean?
    1
  8430. oops
    1
  8431. did u unzip everything into a folder first
    1
  8432. Collab is cracking down on people using it for voice cloning, so it's best if you can install this locally
    1
  8433. ☹️
    1
  8434. Great to hear!
    1
  8435. No problem
    1
  8436. yes
    1
  8437. That's how it is with Colab. It's better if you download it locally. pls see my other video for that
    1
  8438. It's not possible to remove background vocals right now
    1
  8439. put the .pth in a zip. then upload to google drive. then use the drive link to 'download model' here: https://youtu.be/TSLM1mEAc18
    1
  8440. thanks!
    1
  8441. You are welcome
    1
  8442. Not mangio, but applio rvc can: https://youtu.be/jh83ViDUvJs
    1
  8443. 😃
    1
  8444. for sure!
    1
  8445. Welcome
    1
  8446. thanks. it's in the middle of this page https://fudan-generative-vision.github.io/hallo/
    1
  8447. @eddyevodius Gotta get on the cover of Forbes first!
    1
  8448. 😏😏😏
    1
  8449. Agreed! That's what Ilya mentioned before as well
    1
  8450. Sorry I have no idea what that error means. Hopefully someone else here can answer it!
    1
  8451. 🔥🔥🔥
    1
  8452. Thanks!
    1
  8453. ah, reality hits
    1
  8454. my pleasure!
    1
  8455. My pleasure
    1
  8456. noted, coming soon!
    1
  8457. see this https://youtu.be/_JXbvSTGPoo
    1