SPDIF bit depth

You seem to be confusing the long standing and opinionated arguments about sampling rates higher than the Nyquist frequency with the issue about bit depth. The bit depth encodes the dynamic range of the sampled audio, the sampling rate encodes the frequency. High sampling rates can accurately encode frequencies that are too high for humans to hear, or rather too high for anyone but bats and audiophiles to hear. But bit depth encodes the power of the signal i. e. the number of decibels that can be encoded without clipping. There is a very practical use of a higher bit depth, and that is to avoid clipping, which even us normal people can easily hear. Given the choice most recordists will choose higher than 16 bits because it gives them some wiggle room in the strength of the recorded and processed signal. As a rule of thumb a bit of depth provides about 6 dB of encoding, and so 16 bits will provide more than enough dynamic range for music at ~96 dB. If your 0 dBspl is the threshold of hearing, the threshold of pain is about 120 dBspl, and except for club music most music is not supposed to hurt. But it is not uncommon to capture a loud sound and amplify it through a couple of stages to the point that your recording is in the red with 16 bits. Careful attention to recording levels and gain stages can produce a good recording at 16 bits, but the extra bits at the top end make clipping less likely. And if the source you are actually recording has a range from 0 to 96 dB, you are likely to be picking up electrical noise at the quiet points adding hiss or crackles where you intend to have silence. So the advantage to more bits is in what you will not hear, rather than what you will. Twenty to twenty four bits is probably plenty to avoid most problems. But I guarantee that most everyone can appreciate the difference between a sound at 96 dB and one at 140. Then again at 140 dBspl distortion is probably the least objectionable issue with the sound.