[pulseaudio-discuss] Patch to work around some codec's phase inversion when remap_channels mapping stereo to mono

Alban Browaeys prahal at yahoo.com
Tue Jun 22 10:11:53 PDT 2010

> On Mon, 21.12.09 17:26, Bu, Long (long.bu <at> intel.com) wrote:
> > Hi all, 
> > There are some Inel HDA codecs like ALC272 that inverse phase
> > of one channel when capturing..  remap_channels remaps stereo to
> > mono by simply add one channel to another. the sum is always very
> > very close to zero.
> > 
> > ALSA works around this by using "copy" route policy for all
> > Intel-HAD card which is equivalent with PA_RESAMPLER_NO_REMAP set in
> > PA's resampler.
> > 	
> > Attach is a patch that set a REMAP flag for source when its
> > underlying H/W is ALC 272. Then set REMAP flag for source-outputs
> > associated with this source.

I hijack this thread as nothing came out it seems ... and I need more clue as 
to what you meant by telling it was up to the alsa userspace layer to manage this issue.
Ie alsa userspace already handle this via route copy ... though pulseaudio ignores those
alsa plugins. So what do we expect from the alsa userspace layer ?

> Hmm, I am not convinced this is the right approach. The NO_REMAP flag
> simply disables any kind of remapping. That includes disabling
> remapping for surround playback and for example if you have a stereo
> stream with R and L reversed that this is reorder when you play it
> back on a device which has L first and R second.
> I think the effect of this patch is way more than you actually need,
> right?

Could at least a switch that we could enable on a card be acceptable ?

> Also, I am quite confident that the right place to fix this is
> actually ALSA userspace and not PA. I'd prefer not to see
> device-specific hacks in PA and also I believe that this problem needs
> fixing for all ALSA clients, not just PA?

It is already fixed for all alsa clients it seems except PA .

Best regards

More information about the pulseaudio-discuss mailing list