DAW 32-bit float - converters 24-bit: should I use dither?

[sound_forward]

My monitoring signal path is Cubase -> RME Multiface -> Genelec monitors.

Cubase is 32-float, RME DA converters are 24-bit.

I set the master channel fader in cubase usually bewtween -10 and -20 while working.

Should I use dither (apogee uv22hr) at master effect (post fader)?

Thanks

[DAS]

Interesting question...

The whole idea of trying to dither floating point signals is a pretty big can of worms to begin with. I don't exactly know what UV22 would be doing in this application -- maybe someone from Apogee or Steinberg can speak to that more specifically. I wouldn't think you'd need to dither because you are turning your master fader down. The floating point math should be able to scale accordingly. I also don't think you should have to dither to get from 32 bits down to 24 because only 24 of those 32 bits are used for the audio wave -- the other 8 are used to scale it up and down. (Yes, I know this is a simplistic and crude explanation.) Now, if you were going down to 16 bit I would say that you should use it...even though I'm still not exactly sure how it gets applied to a floating point signal (the aforementioned can of worms is still applicable here).

My guess would be that you do not need it, but the specific implementation in their software may dictate differently. Of course you could try it with and without and see what, if anything, you hear.

[Dan]

32 bit applies to the internal mathmatics, it outputs 24bit. No need to dither.

[16/44.1]

Originally Posted by sound_forward

My monitoring signal path is Cubase -> RME Multiface -> Genelec monitors.

Cubase is 32-float, RME DA converters are 24-bit.

I set the master channel fader in cubase usually bewtween -10 and -20 while working.

Should I use dither (apogee uv22hr) at master effect (post fader)?

Thanks

Only in that case.....yes, you should.
If you reducing resolution by going from 32 to 24 or lower, always dither.

[DAS]

You aren't reducing the resolution though (or at least not much). In 32 bit you have 24 bits of precision. (Actually 25, but that's not so important right now) In a 24 bit converter you have 24 bits of precision. You shouldn't need to dither to make that step, but I qualified it by observing that I don't know exactly how things are implemented in those two programs.