[mythtv] [mythtv-commits] Ticket #1608: Proper Handling of Alsa AC3-PassThru

[Markus Schulz]

Am Samstag, 1. April 2006 14:46 schrieb Martin Ebourne:
> On Sat, 01 Apr 2006 00:29:20 +0000, MythTV wrote:
> > #1608: Proper Handling of Alsa AC3-PassThru
> > -------------------------------+-----------------------------------
> >--------- Reporter:  msc at antzsystem.de  |       Owner:  rkulagow
> >      Type:  patch              |      Status:  new
> >  Priority:  minor              |   Milestone:
> > Component:  mythtv             |     Version:  head
> >  Severity:  medium             |
> > -------------------------------+-----------------------------------
> >--------- Fixed ac3 pass thru code with help from mplayer code.
> >
> >  Now the samples will be passed thru, only if passthru is enabled,
> > to iec958 alsa plugin. This way all normal mpg stereo output can go
> > to alsa:default (mostly pcm0 or dmix plugin or similar) and ac3
> > will be passed thru iec958 plugin.
>
> No, that's not right. Your code forces the output on soundcard 0
> whereas my passtrhu output is on soundcard 1 for instance. My
> soundcard (Soundblaster Live 24-bit External) doesn't need any
> special options set for passthru, and I don't think even supports the
> iec958 plugin (I can check and confirm that for certain if it helps).

What have you configured as alsa output? ALSA:spdif? or similar alias?
IIrc the iec958 is used all the time if you pass thru, cause iec958 is 
the sample format for sp/dif output. Check "aplay -L" for all 
pcm-output device strings.

The problem is, that we should be sure to send normal uncompressed pcm 
data to another pcm output then compressed ac3 streams. Best way will 
be an additional input field (next to passthru checkbox) where you can 
enter the alsa-pcm-device string for ac3 streams differently to 
uncompressed pcm data.

Current SVN Code only appends the iec958 specific options ( ":{ AES0 
0x02 }" ) to the device string you have been given. When this is 
ALSA:default and default is pcm0 Out, this option is totally wrong and 
results in errors.

-- 
Markus Schulz