Comments by "Lawrence D’Oliveiro" (@lawrencedoliveiro9104) on "The Compact Disc: An Introduction" video.

  1. Are you sure it was FEC? I remember being in my uncle’s place years ago (he was a radio ham) and hearing some digital data communication being monitored on his gear: you could distinctly hear the packet being transmitted, followed by the acknowledgement, then another packet etc. Seemed like a waste of bandwidth. I told them they should do a TCP-style sliding-window protocol, to allow multiple packets to be sent before receiving the acknowledgements, to get around the channel latency.
    6
  2. There were never “CD+R/RW” formats. You’re thinking of DVD+R/RW.
    4
  3. Here in Mathland, quaternions break into you.
    3
  4. CD-R brands varied widely in longevity, in my experience; Verbatim, Sony and TDK were good, Imation was crap.
    3
  5. CDs were designed for audio first, and general data files (CD-ROM) later. DVDs were designed to be a computer data format from the beginning, with a proper computer file system (actually two, ISO-9660 and UDF), with all the actual applications (DVD-Video, DVD-Audio etc) being done basically as file formats on top of that.
    3
  6. 12:37 Surely the bumps would form pits (cavities), not lands.
    2
  7. There was an extra spec called CD-TEXT, that added track title/artist information using the “P” and “Q” subcode fields. My car stereo can display this, so my CD-burning script includes the info in the control file for cdrdao.
    2
  8. Yes, Fuad. That is very funny.
    2
  9. CD-RWs worked by a phase-change method. I think this meant that the material changed its refractive index, and that is how it generated the same interference pattern you got with the pits and lands on a pressed CD.
    2
  10. I just realized the, ahem, connection with Burke’s “Connections” series. Thanks for pointing it out.
    1
  11. 7:21 Why jumble up the data? Because there was an expectation that a scratch on the disc, for example, could destroy a whole lot of bits in a row. But because of the shuffling, these destroyed bits end up spread out among a bunch of other good bits, lessening the impact of the run of errors.
    1
  12. 7:49 OK, you mentioned that point. ;)
    1
  13. Reed-Solomon has never been improved upon. ;)
    1
  14. FLAC is good, because it is a) lossless, and b) can include metadata (artist, title etc) that common players like VLC can recognize.
    1
  15. This was a frequently-commented-on drawback of CDs, that they had less room for that lovely album cover art.
    1
  16. 09:18 Actually, the real reason for EFM encoding is to reduce the frequency bandwidth of the signal that the circuitry has to cope with. Remember that each 1 represents a waveform transition. So two successive 1s in a row represent a frequency of 14 bits per encoded byte × 2352 bytes per sector × 75 sectors per second ÷ 2 transitions per cycle = 1234800Hz. By insisting on at least 3 0s between each pair of 1s, this is reduced to a quarter of that. At the bottom end of the frequency range, the maximum number of 0s you can have in a row is dependent on how stable you can maintain your clock sync.
    1
  17. Sorry, since it’s a minimum of 2 0s in a row, that’s a third of that frequency, not a quarter. The number of combinations of 14 bits that satisfy these constraints is 267, just slightly more than the 256 needed to encode 8 bits of data. (Which explains why 14 bits are used, and not 13 or 15, say.) Some are used for special marker purposes. There is also a 3-bit “packing period” between each of these 14-bit words, which can contain an extra 1-bit as necessary to even out the total numbers of 1s and 0s to avoid the overall signal having a DC offset. Source: John Watkinson, The Art Of Digital Audio , of which I have the 2nd edition from 1994.
    1
  18. That description applies to sign-magnitude as well. The difference is, consider 16-bit integers: unsigned they can represent values from 0 to 65535. In twos-complement, you interpret the encodings after 32767 to actually represent that value minus 65536. So the encoding that would be 32768 becomes -32768, 32769 becomes -32767, and so on up to 65535 which now represents -1. The clever thing is, if you add two such numbers as though they were unsigned, you can interpret the result as signed and the answer is correct. Conversely, to subtract two numbers a and b, just add a and -b instead.
    1
  19. 3:09 That’s the opening to Steven Spielberg’s Amazing Stories -- not that I’ve seen the series, but I recognize it from some of the vintage CG clips here on YouTube.
    1
  20. The EU remembers. It’s their anthem.
    1
  21. And for CD-ROMs, there’s an extra layer of error correction on top of that.
    1
  22. Using something in its entirety can indeed be fair use: https://www.techdirt.com/articles/20090810/1913245833.shtml https://www.techdirt.com/articles/20120530/17333319132/myths-realities-about-fair-use.shtml
    1
  23. “Sample rate” ≠ “Bit rate”
    1
  24. My previous car CD player couldn’t cope with zero pregaps. The current one can, though.
    1
  25. 5:17 I thought the “I” stood for “Incremental” (as in it’s the changes that are important, not the levels).
    1
  26. CIRC can correct a whole lot of errors -- a lot more than Hamming. But you need a lot of redundant bits to do it.
    1
  27. Doesn’t that make it sound “digital”?
    1
  28. I rest my case.
    1
  29. 09:40 You could have any number of 0s in a row. Or any number of 1s in a row.
    1
  30. Look at the part of the video I’m referring to.
    1