Re: zita-a2j and capture alignment

On Sat, Sep 07, 2013 at 04:21:35PM +0100, Ben Bell wrote:

> Zita-a2j always resamples, it doesn't know that the cards are synced.
> I think I vaguely remembering you once wondering out loud about making
> an option for disabling resampling. Any chance of that? :)

When my new PC arrives (should be early next week), I'll look into
both issues. Currently I can't test anything...

Zita-ajbridge tries to set a well-defined and stable latency -- it's
the latency that controls the resampling control loop. So it should
be possible to set port latency automatically, with a command line
option to modify it.

The only problem I see with the non-resampling mode is that the latency
is fixed to the value set initially, which may be less accurate. When
resampling that is not problem, as it is adjusted constantly and will
reach the target value after some time. A longer start-up delay
(allowing more time to calculate the initial value) may solve this.

> Yes. In the mean time there's the possibility of using two zita-a2j
> instances talking to a jack with a dummy back end, but then everything
> would be resampling which really does undermine the whole idea of jack
> giving sample-accurate recording.

That's relative. The values you get from the pcm interface are not
the original ones from the real A/D converter anyway - most interfaces
use oversampling converters, so what you see is already filtered and
resampled by the digital hardware. What happens there is essentially
the same as in a resampling lib such as zita-resampler or SRC.
 
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)