When Compact Discs were invented they encoded analog music using 16 bits of resolution at a sample rate of 44.1 kHz. A rate of 44.1 kHz gives the Nyquist frequency of 22.05 kHz (half the sample rate) and supports unaliased reproduction of frequencies from 0 to the Nyquist frequency. 22.05 kHz covers the hearable frequency range of human beings. As we age, our frequency range reduces considerably, as well.
So why encode audio at higher sample rates and with greater bit depths? Does it sound better? In general, it should not, but there are good reasons for mixing and mastering at higher sample rates and bit depths. Specifically, high frequencies outside our hearing range might get aliased down into that range if they are not filtered out, but during mixing and mastering they can be preserved by using higher rates. Digital Audio Workstations (DAWs) offer a range of options but 48 kHz and 96 kHz are widely used. The latter provides extra headroom with a Nyquist frequency of 48 kHz when using effects and other types of processing. Greater bit depth grants greater dynamic range, as well.
There is a side note, also: much re-encoded legacy audio is just upsampled from 16/44.1 content anyway, so there is no new information available in the porkier files.
However, there are some studies that suggest that human hearing might be able to distinguish timings that exceed the Nyquist frequency in some cases. Whether this is sufficient to lead to perceivable differences in music listening is not clear, however. As mentioned in the linked study, it is also possible that filter roll-off properties may be the determinant factor. A curious suggestion is that ultrasonic sound might be perceivable in some cases via bone conduction. With great claims, though, there are always doubts. Check out the analysis of the work here that casts doubts on the validity of the research as well as additional understanding of the masking effects that are in actual music (rather than square waves).
So let’s keep this at Audiomyth Level 3. There are some counterarguments and evidence but it remains difficult to make a solid determination of how that evidence might result in better audio recordings given the limitations of the recording capabilities themselves.
