[linux-audio-user] Capture with DIGI9636 Hammerfall Light andplayback with PCI 128

[Paul Perkins]

Mark,

Thanks, your message did indeed suggest to me a way that using different
sound cards for input and output makes things more complicated for the
sound system. Only if the input analog-to-digital converter (ADC) and
the output digital-to-analog (DAC) use the same clock, will the number
of input samples available for reading match the number of output
samples that should be written, over a given span of time. If there are
two clocks, one will be slightly faster than the other. If some sound
processing application is in a loop of reading X number of samples,
processing them, and then writing X number of (different) samples,
eventually the faster interface has to wait for the slower one to catch
up. Waiting means lost samples, and lost samples sometimes cause audible
clicks or pops.

On Thu, 2003-05-01 at 09:54, Mark Knecht wrote:
> > On Wed, 2003-04-30 at 07:13, Steve Harris wrote:
> > >
> > > You cant easily because the soundblaster and RME (unless you
> > have wordclocked
> > > them) will be running out of sync. If you did manage to make a virtual
> > > device that read from the RME and wrote to the SB you would get lots of
> > > clicks and pops.
> >
> > Is this still an issue if all the recording is with one card and all the
> > playback is through the other card? And all the playback, at least, is
> > analog? I don't see where the problem would come from in that situation.
> >
> Paul,
>    I'm going to guess that the problem will be in the way the output data is
> handed to the Sound Blaster. Let's say that the underlying sound system
> (Alsa or whatever) is running off of the RME's clock. The data is brought
> into the system using the RME clock. It gets processed in the Linux
> environment using the RME clock, and it gets handed, as a block of data, to
> the Sound Blaster using the RME clock. This is the problem I think. The
> sound Blaster needs to have the block handed to it based on it's own clock
> rate. If not, then there will be times when the SB clock comes a little bit
> early and the data won't be ready to go since the RME clock hasn't happened
> yet. This will result in a click.
> 
>    I hope this helps.
> 
> Mark
> 
> 
> 

-- 
sigmotto: Liberty is theft.