How much sample rate and bit depth is enough? Well, according to the Nyquist Theorem, you need twice as many samples per second as the frequency you're trying to digitize. Human hearing peaks out at around 20-24 KHz, so it would stand to reason that we need a sample rate of 40-48 KHz to reproduce the entire range of human hearing. But that's only half the story. It may only take two samples per sound wave to reproduce it digitally, but the quality of taking this minimum approach is less than desirable. Bumping it up to four samples or more per wave, however, creates truly compelling audio. A sampling rate of 96 KHz is used in the DVD-Audio standard and by most professional digital recording equipment. This allows four samples for each wave in the upper limit of human hearing, and six to twelve samples for waves in the 8-6 KHz frequency range, where most of the music we hear is.
Bit depth is pretty good at 16 bits per sample, which is most common, but most professional audio equipment and the DVD-A standard use 24 bits per sample. More bits are good in audio for the same reason they're good in graphics. 16-bit audio allows for 65,536 different "levels" per audio sample, while 24-bit boosts that fidelity to 16.7 million. This huge increase helps preserve nuances and overtones and prevents quantization errors when mixing tracks, in much the same way as 24-bit color on your monitor prevents the banding artifacts prevalent with 16-bit color.