Does anyone want to learn some of the terminology involved and how that matters for setting input levels? If so, this is what I picked up in EE-638 from Purdue University many moons ago.

The quantization signal to noise ratio is 48 dB better in a 24 bit recording compared to a 16 bit recording. The rule of thumb is that the QNSR improves 6 dB for every bit of the A/D conversion at hand. To hear how this works in the reverse, take a nice clean 24 bit signal and then do bit reduction to 16 bit (most won't notice), then bit reduce to 12 bit (you will probably hear this) and then 8 bit, and you'll definitely harken back to the old Castle Wolfenstein game days with it's crudely rendered words and sounds.

Quantization noise is what happens with the signal that is effectively zero and the A/D has to decide if the signal is actually all zeros, or all zeros except for a 1 in the LSB.

2^16, or 65,536 'shelves' to stick the signal on with 16 bit recording, and 2^24 or 16,777,216 potential shelves to stick the signal on with 24 bit recording.

The quantization of the signal for a 24 bit at such a finer resolution is what accounts for the 48 dB better QSNR.

Translation: There is no harm in recording near full-scale use at 24 bit resolution, but there's also WAY WAY WAY more wiggle room (maybe another WAY in there) with 24 bit vs. 16 bit. I state that there's no harm because you can scale down the full scale tracks after you've made the recording, but the chance to be introducing quantization noise at let's say -18 dBfs, is for all intents and purposes non-existent.


Last edited by rockstar_not; 09/21/15 04:27 PM.
My Songs, photos, writing,etc.
My Sound Cloud