3 ms·

>Combine that with the very limited bits provided by CDs, preventing any possibility of good playback 16 bits gives you 96dB SNR (and subjectively more with di

by mrob 8y ago

>Combine that with the very limited bits provided by CDs, preventing any possibility of good playback 16 bits gives you 96dB SNR (and subjectively more with dithering), which is plenty at non-deafening listening volumes. It's more than even the highest quality vinyl ever had. High bit depth is mostly useful so you've got some headroom during editing.

MrTonyD 8y ago
That is absurd. Anybody who has read about the history of CD bit depth would know just how absurd that is. And I even personally investigated CD accuracy in burning using an electron microscope to understand how the laser burning process worked on the plastic media. It was absurdly inaccurate. All these claims about how great digital is are just crazy - claims made by people who are relying on old data about hearing accuracy and a religious faith in technology. Anybody who has actually designed instruments using the known laws of physics understands the incredible limitations - and those limitations apply to both characterizing hearing as well as reproducing sound digitally.
- mrob 8y ago
  >I even personally investigated CD accuracy in burning using an electron microscope to understand how the laser burning process worked on the plastic media. It was absurdly inaccurate. I can believe that, but the standard was designed to tolerate high bit error rates. 25% of the bits are redundant, as part of a error correction code: https://en.wikipedia.org/wiki/Cross-interleaved_Reed%E2%80%93Solomon_coding https://en.wikipedia.org/wiki/Cross-interleaved_Reed%E2%80%9... This scheme works very well, as proved by people consistently ripping bit-for-bit identical copies of CDs pressed or burned with the same audio.
pps43 8y ago
16 bits give you 96 dB SNR only if your signal is one-bit. In a quiet room with decent headphones set to comfortable level I can easily hear one-bit flips in otherwise silent WAV file. Not a problem for popular music that's consistently loud, but a real problem for classical music with wide dynamic range.
- mrob 8y ago
  I tried it and I couldn't hear it. Are you sure you got the endianness right and it was really a least significant bit?
  - pps43 8y ago
    Yes, pretty sure. Had to zoom in a lot in audio editor to even see it. Could hear it nevertheless. It's important for the room to be quiet, as the sound is very faint and easily masked, but it's there. It's possible that Bluetooth headphones could lose it because of compression, or cheap DAC in the sound card simply ignores least-significant bits.

by mrob 8y ago

MrTonyD 8y ago

That is absurd. Anybody who has read about the history of CD bit depth would know just how absurd that is. And I even personally investigated CD accuracy in burning using an electron microscope to understand how the laser burning process worked on the plastic media. It was absurdly inaccurate. All these claims about how great digital is are just crazy - claims made by people who are relying on old data about hearing accuracy and a religious faith in technology. Anybody who has actually designed instruments using the known laws of physics understands the incredible limitations - and those limitations apply to both characterizing hearing as well as reproducing sound digitally.

mrob 8y ago
>I even personally investigated CD accuracy in burning using an electron microscope to understand how the laser burning process worked on the plastic media. It was absurdly inaccurate. I can believe that, but the standard was designed to tolerate high bit error rates. 25% of the bits are redundant, as part of a error correction code: https://en.wikipedia.org/wiki/Cross-interleaved_Reed%E2%80%93Solomon_coding https://en.wikipedia.org/wiki/Cross-interleaved_Reed%E2%80%9... This scheme works very well, as proved by people consistently ripping bit-for-bit identical copies of CDs pressed or burned with the same audio.

pps43 8y ago

16 bits give you 96 dB SNR only if your signal is one-bit. In a quiet room with decent headphones set to comfortable level I can easily hear one-bit flips in otherwise silent WAV file. Not a problem for popular music that's consistently loud, but a real problem for classical music with wide dynamic range.

mrob 8y ago
I tried it and I couldn't hear it. Are you sure you got the endianness right and it was really a least significant bit?
- pps43 8y ago
  Yes, pretty sure. Had to zoom in a lot in audio editor to even see it. Could hear it nevertheless. It's important for the room to be quiet, as the sound is very faint and easily masked, but it's there. It's possible that Bluetooth headphones could lose it because of compression, or cheap DAC in the sound card simply ignores least-significant bits.