Quote:
Originally Posted by Dean Roddey  I read some of that, but I'm still not convinced that we are talking about the same thing. The question isn't whether 16 bits or 24 bit sample signals can reproduce the original signal or not. Clearly it can. And it's not whether lowering the fader in the digital domain reduces the quality, since it doesn't.
It's whether a signal sampled at the full available range of sample levels at a given sample rate, has represented the amplitude of each sample more accurately. I fail to understand how it cannot be the case, given that you have spread the signal over 8 times fewer possible sample values, therefore it has to be more heavily quantized. |
Because as you descend in amplitude, the noise level in the signal is already dwarfing the loss of resolution...the quantization error is less than the noise level.
So yes, you should strive to maintain as much dynamic range as you can. However, in practice, once again, the noise floor of your preamp is already going to be louder than that of your converter, therefore, as long as you print so that noise floor is higher than the converter's (very easy with good ADCs), you lose nothing.
Nothing!
