I'm more or less copying what I posted on "the other" thread:
I'm seeing this from a numerical point of view. For digital recording there is one thing that you do want and one thing that you don't want:
- You WANT high integer numbers, because that increases precision. If your maximum is 65535 it means that you have 65536 "possible levels of expression". If you set your maximum to 100 you only have 101.
- You DON'T WANT integer or float overflows (that translates to clipping), because at that point you're loosing (killing) information.
So, leisuremuffin, I think you're right with your initial idea: as long as you don't get any clipping (overflow), you can crank up your faders as much as you like. This is especially (or better: exclusively) possible with pre-recorded clips where you can go through several cycles and turn faders down until you don't get reds anymore. In that kind of situation you don't need any headroom.
BUT: If you are in a live situation where you have to deal with the unknown, then you don't have any headroom. And that means: you're doomed.
I'd need to know some facts:
- Is it true that as soon as the data enters Live everything is converted from int to float until the data leaves Live via the master track?
- What about VST or AU plugins and rewire apps? Do they get/return int or float data? Or can they process/deliver either?
- What's the bit-depth of the master track's "exit" when you're not rendering to disk but simply sending to your D/A device?
Who knows the answers?
