@TheVogon... Mastering
The point of the high bit-count high-sample rate audio in the studio is that the signal gets filtered, scaled, and maybe even sped up/slowed down before outputting to CD.
Just like calculating with rounded-off numbers, starting with 16-bit 44kHz data would lead to rounding errors in the output, which increases quantisation noise. The Nyquist filter for 96kHz sampling can be a fairly simple affair, as it doesn't need to roll off particularly quickly. The recovery filter in CD players needs to be quite steep to prevent >22KHz sidebands in the output, which, whilst being inaudible themselves, can cause intermodulation distortion in less-that-perfectly linear devices downstream. Like speakers, for example. As a result, the phase behaviour can be a bit squirrely.
Biting the hand that feeds IT
