Isn't that kind of related to dithering, too? And the reason why someone like Chris from Airwindows wrote a dozen different dither algos, each with their own unique sonic fingerprint? Then again, to be honest I never quite wrapped my head around dithering, but I know that it involves noise being added to a signal for addressing bit resolution.
Right. Well, dithering is more subtle - or I should say the dither noise itself is, for one because it's at a much lower level than hiss, but mainly because it's designed carefully not to be heard.
The basic idea is that when you reduce the number of bits - typically when you go down from the 24-bit production standard to 16 for CDs (I know, what are those?) - what you lose is low-level detail (because you're truncating the bottom bits) and just cutting that off can sound harsh.
So dither is very quiet noise added to the signal that causes it to jump over the 16-bit lower threshold some of the time, thus smoothing the transition to silence. The best analogy I've heard is kids jumping up to see over the fence at a baseball game - they get a view every time they peer over.
However, the different flavors of dither get very tweaky. Not only is it noise-shaped, meaning it's at frequencies the ear is least sensitive to, it's designed to be decorrelated from the signal to help mask the effect.
You can hear the effect by recording a decaying note at 24 bits, possibly lowering it way down to make sure it falls into the bits you're losing, and bouncing that to disk at 16 bits both dithered and truncated.
Then raise the level way up and compare the two. You'll hear one fade out smoothly and the other cut off abruptly.