Audition sets the sampling rate and bit depth for incoming recorder signal?

[OttoC]

Novice question:

I have a Sound Device MixPre 6 II connected to my MacBook Pro. I set the bit rate to 32 (float) but the unit will not allow me to set the sampling rate. 

Is this because Adobe Audtion (AA) set the sampling rate (which I want to be 48kHz)? I assume this because AA is doing the actual recording and thus must sample the incoming 32bit signal. 

Lastly, since I am here, if the MixPre is sending a 32bit signal and I accidentally set AA to 24bit, will it downsample the incoming signal or will the MixPre override AA in the case of bit rate?

thanks, Otto

p.s - first post. nice to be here.

[SteveG_AudioMasters_]

I think that everything Sound Devices makes is USB class-compliant - which essentially means that whatever controls the interface (and it's unlikely to be Audition) gets to say what sample rate it runs at. That's certainly the case with my MixPre-D, anyway. If I want that to run at 48k, then I have to tell the Windows Sound control system that this is its new default setting. I'm not sure how this works out on a Mac but I would imagine that something similar is going to happen. Audition records what it's sent...

As far as 32-bit recording is concerned, don't sweat it. What happens with 32-bit recording is that it's a Floating Point format that can easily be rescaled, and that's what seven of those bits do - the rescaling. Twenty four of them represent the signal normalized into a range between 0 and 1, and the other bit is a sign bit, saying whether that's a negative value or a positive one. Put simply, there's nothing to be gained by recording in 32-bit mode - the only advantages come in signal processing. Essentially it's a 24-bit signal with some flag-waving. The Laws of Physics dictate that you can't use more than 20 bits of that anyway; it's not possible to build anything with a noise floor below about -118dB and you can sample that fully with a 20-bit digitiser. They only make them in multitples of 8 bits though, and that's the reason we've got 24-bit ones - nothing at all with being able to actually use all that range! What actually gets recorded when you set a device to 32-bit is an integer value that's converted to fractional values, and a string of zeros in all seven of the range bits. The huge advantage comes in the processing later - you can ramp a 32-bit signal up and down with impunity. If you did this with an integer signal and stored the results, you'd lose most of the bit depth - permanently. You can store 32-bit files at -115dB so they look as though nothing is there, and if you gain-boost them again on reopening, everything's back exactly as it was. Do that with an integer-based file and you'd end up with nothing except noise.

[OttoC]

Excelllent, thanks Steve!