16 bit or 24 or 32 float for ACX?

Is there any connection between the noise floor and 16, 24 or 32 bit recording?

Missed one. Audacity operates in 32-bit, floating. It does that so there is no sound damage no matter which effects you use or in what order.

The one place where Audacity actually creates noise is sound file export. When you make a sound file from your show, Audacity has to convert from 32-bit floating back to normal sound files. It helps that process by adding a tiny bit of special background noise to the show. It’s way quieter than performance noise.


In short, the more “bits”, the lower the “possible” noise floor.

In real life, the actual noise floor is considerably more that the “lowest possible” for 16, 24 or 32 bit.

For technical reasons, it’s best to work in 32-bit float, then export in the required format (usually 16-bit).

That’s the practical information. Wikipedia has pretty good account of the technical information: https://en.wikipedia.org/wiki/Audio_bit_depth

Thank you everybody for your help!