[MPlayer-dev-eng] Practical outcomes of the ALSA thread

[Attila Kinali]

On Fri, 4 May 2007 21:49:35 +0200
Nicolas George <nicolas.george at ens.fr> wrote:

> Le quintidi 15 floréal, an CCXV, Attila Kinali a écrit :
> > Yes, but this means you have to ensure that this config
> > snipped is used on all systems with alsa on which MPlayer
> > (or any other audio application for that matter) is used.
> 
> Now, that, I do not agree. The decision to use automatic resampling, for
> example, is a matter of cost vs. benefits evaluation: with automatic
> resampling, audio output always works without manual configuration, but
> sometimes the quality may be suboptimal; on the other hand, without
> automatic resampling, audio output may need to be tweaked by hand sometimes,
> but quality will always be optimal.

I agree here, it's up to the user what he wants. But resampling
has to be done at one point, and only at one point.
For me it doesn't matter whether ALSA does the resampling or
MPlayer. But if ALSA does it, it has to support all the sampling
rates that MPlayer throws at it and not just provide a selection
of possible sampling rates.
 
> Given that most of the time, there will not be double-resampling, and that
> most users would not hear the double-resampling anyway, I think that
> trusting the default setup of the distribution plus a FAQ entry would quite
> enough.

No. Resampling can be heard quite well if it isn't done
with a high quality algorithm. If it is done twice, you will
notice it even if it is both time a good algo.

			Attila Kinali

-- 
Linux ist... wenn man einfache Dinge auch mit einer kryptischen
post-fix Sprache loesen kann
                        -- Daniel Hottinger