dithering in digital audio has to be one of my favorite little tricks of all time. i know why it works but the fact that you can still create detectable signals below the quantization resolution feels like magic
@vaporeon_ xiph.org has a great episode 1 on the fundamentals of digital audio
the bit depth (that is, number of depth purr sample) determines the resolution of audio, in the sense of how many pawsible diffurent voltages relative to max can be repurresented. (theoretically, the analog signal being captured has more or less infinite resolution.) so, inevitably, there is something of a quantization error when sampling audio due to necessarily imprecise “loudness”, as well as a floor fur how quiet a signal can be pawsibly repurresented
HOWEVER. dithering the signal, by just adding a quiet amount of pseudorandom noise, increases the effective resolution of digital audio by a significant amount.
(fur a thought experiment, consider a bit depth of one, and a target signal of exactly 0.5 (in the range from 0 to 1). obviously, with standard quantization error, the only pawsible repurresentation of the target signal is full silence or full loudness at all times. but with a succulently loud, pseudorandim, dithering noise, sometimes the target signal + dither will cross the threshold fur 1, and other times 0, but the target signal being captured will bias the resulting capture so that the mean recorded signal over time is 0.5)
@vaporeon_ the xiph.org video, if you have half and hour and ever feel like getting utterly inundated with infurmation
@aescling There's dithering for audio and not just for pictures? :O
Please tell me more, I am quite clueless about audio...