Hacker News Reader

Comment by anthk | Hacker News Reader

CD's had no distortion.

IIRC CDs were the first use of dithering in audio. It's not distortion, but both it and distortion change the signal.

https://en.wikipedia.org/wiki/Dither

While its true that dithering changes the original signal, dithering is a necessarily evil that prevents quantization errors as a result of downsampling 24-bit+ source material due to CDs being 16-bit. It’s a choice between some barely audible noise, or some weird spectral artifacts..

I was under the impression that the dithering level is well below our threshold for hearing. Can someone confirm or deny that?

Dithering modulates the least significant bit of the audio signal, so at normal listening levels it should be inaudible. Of course, one can amplify the silence by 120dB or so and hear the noise.

Sources: http://www.dspguide.com/ch3/1.htm

https://www.bruce.audio/post/2022/05/26/dithering/

I often wondered about the noise I was hearing from my Discman with headphones. You'd hear this hiss when you pressed play, even before the CD loaded, so it was probably the cheap consumer audio hardware rather than the CD format itself.

I think an additional layer of hiss would show up when it actually started playing.

Indeed, no inexpensive portable digital electronics was going to give you perceptible-noise-free 16-bit audio.

Depending on how the original audio was mastered, it could also have it's own hiss in the sound data if the CD itself. You can tell if you load the raw CD data into a wave editor and see low amplitude noise where you'd expect silence.

This can be a contentious area as a lot of different sources will throw different figures for the noise floor of a CD. I recall anywhere from -98 to -130dB. Either value, at the minimum level, is something like 4.6dB. Which is completely inaudible in normal listening conditions.

Your comment has strong “well, actually …” energy. We’re talking about faulty blocks from physical abrasions and the two common error mechanisms (skipping or repeating). They were both “weird, ugly, uncomfortable, and nasty.”

I recall reading a music artist (Akufen, maybe?) would scribble over CDs with a marker to get them to skip, then use the resulting sounds to make music.

Records would skip as well due to worn tracks, worn needle or someone bumping the turntable.

CD Reed-Solomon forward error correction did make that pretty rare though. You had to abuse a disc a fair amount to overcome the coding.

"Vinyl" would make more sense in this context.

When I first used CD music it was ugly to my ears. It was full of high frequency sound compared to the hissy tape. I got used to it and all this digital sound DSP was available to change the sound to your liking.

In theory yes, but in reality it has a plenty of distortion from various sources.

For instance the jitter - the early CDs didn't have a super stable clocks that governed the conversion, so the slight changes of the pace when D/A conversion takes place would create slight distortion and it would also affect intersample peaks making it more apparent. Then you have the non-linearity of D/A converters themselves that were essentially acting like waveshapers.

Reply on news.ycombinator.com