Re: [LAU] questions about resampling - was - jack and the merging of soundcards

On Fri, Nov 21, 2014 at 06:17:43AM -0800, Len Ovens wrote:
 
> Ok, so open question then. Does it make a difference to quality what
> the original rate is? For example: the direct card with no resample
> is a 48k, the second card will be resampled, is it better to be as
> close as possible to 48k or would starting at 96 (or whatever) work
> better?

First, assuming the resampling is done correctly, there is no loss
of 'quality' in the sense of distortion, added jitter, etc. etc.
The only differency will be some change in frequency response, and
then only for frequencies close the half the sample rate.

In practical cases, if all sample rates are at leat 44.1 kHz, the
differences are so tiny that they become completely academic.

> In all this I would assume, stereo pair inputs should be kept to one
> card. Or a suround group for that matter. Or does the internal
> resample for filtering make that not an issue anyway?

Any digital resampling done by a single device or software resampler
will be identical on all channels, so it doesn't change the relative
phases of the channels. So yes, stereo pairs, AMB signals etc. should
be on the same device.
 
> One more... if a group of inputs that start synced are resampled, do
> they have a better post resample channel to channel sync that
> accross two intefaces with different sample rates before resampling?
> (this is sort of what the above questions are getting at though
> there amy be other issues as well)

I don't understand the question... could you rephrase it ?

> In the end, most people are comparing these questions to analog with
> the idea that theis no phase shift in analog... I can see that
> two analog channels through two eqs with the same circuit, set the
> same way, will still likely have some phase shift between them.

Maybe, but if the amplitude responses are the same the phase responses
will be the same as well. And even if the amplitude responses are not
exactly the same but still similar, the differences in phase response
will be very small. This is so because for the simple circuits used
for audio EQ the amplitude and phase responses are linked by mathematical
relations. It is actually quite difficult to control them separately
with analog circuits. Whatever phase differences there will be between
'identical' EQs will be very small and of no consequence. One example:
if the phase difference between two equal amplitude signals is 30
degrees, then the sum will be 0.3 dB down. That will be a tiny fraction
of the amount of EQ you are applying anyway.

> So is the channel differences caused by digital manipulation better
> than analog? The same? Worse?

Unless you do something very weird, there will be no differences
in the digital case.

Ciao,

-- 
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user@email-addr-hidden
http://lists.linuxaudio.org/listinfo/linux-audio-user