[QLab] Sample rate and Qlab

[Christopher Ashworth]

On Nov 6, 2006, at 10:00 PM, Clay Benning wrote:
>
> My question for Chris, and others who want to chime in, how does the
> locked sample rate of the output device affect all this? If I have an
> external firewire box with a default sample rate of 44.1kHz, is this
> messing up the nice quality of my 96kHz sampled music file? Should I
> really set the master device clock to 96kHz and let it up-sample all
> of the 44.1kHz stuff? Or are we splitting hairs since most of the
> quality is in the file itself, rather than the D/A?
>
> What is the CoreAudio (and Qlab) take on this?

That's a great question.  I must admit my DSP math skillz are falling  
short on the upsampling aspect.  Naturally you're not going to get  
more information than you started with, and the upsample will only  
interpolate between the samples you've got.  I'm unsure whether that  
interpolation necessarily screws with the frequencies present in the  
signal, and if so, if it does so to an audible degree.

Equally obviously, downsampling throws away information, and  
certainly risks aliasing.

If Apple publishes the particular algorithms they use to do sample  
rate conversions, I'm not currently aware of it, so it would seem an  
analysis of the quality would have to be done empirically.

A higher sample rate would, of course, require more processing power,  
so that should be factored in too.

At the end of all this, I can't give a complete answer because I  
can't speak with certainty about upsampling.  However, since  
downsampling will certainly degrade quality, my gut reaction is:   
match the sample rate of the device to the highest sample rate of  
your files.  Use fleshy appendages on head to sanity check; lower  
device rate if upsampling is introducing audible artifacts, or if CPU  
usage becomes a concern.

Sorry I can't be more authoritative at the moment.  I should go dig  
back into my image processing textbooks....

Anyone else got a better answer than this?

Cheers,
C