[MPlayer-dev-eng] SB Live/emu10k1 driver capabilities

[Anders Johansson]

Hi,

Adding support for 6 channels is mainly a codec side problem. Adding
it to audio_out is planned work that is in my pipe (but I won't stop
you from helping me :). I've got a 6 channel card myself (creative
audigy) and am dying to see it run using all 6 channels. I have
planned writing output plugins to synthesise the missing channels when
only stereo is provided, I have also planned to write a anything to
ac3 encoder to use with external ac3 decoders. A fairly large patch is
in for testing but is not applied yet, so if you are planning on making
any changes to the ao_xxxx drivers perhaps you should wait until the
patch is applied.

//Anders


> On Thu, 22 Nov 2001, Arpi wrote:
> 
> > > b) Simply call SNDCTL_DSP_CHANNELS to request 6 channels, and write the
> > sounds interesting. but doe sit work for non-5+1 cards too?
> > I have "old" sblive, with only 2 stereo analog output on it.
> 
> I've got two SBLive cards - one is a "SBLive Player 5.1" and that has
> altogether 6 analogue outs - 4 on a pair of stereo minijacks, the
> subwoofer and centre on a "merged" digital/analogue
> 
> The other is a SB Live with 4 analogue outs on a pair of minijacks,
> but no analogue/digital output.  Is that like yours?
> 
> I'm no expert - but I expect that mplayer could output full 6 channels to
> either - non 5.1 users could set up the card to mix the LFE and centre
> channels into the others.
> 
> > > Is it an easy change to do the ac3 decoding and pass all the channels
> > > discretely?  It looks like "skip the downmixing" in libac3, perhaps some
> > > tweaks in the libao2 drivers to deal with >2 channels?
> > yes it needs changing both. but it should be easy.
> > just define a new AFMT_ for 6channel out and extend libac3 (and dec_audio.c)
> > to use this format.
> 
> I did a quick hack and made an ao_sblive that sets the card to 6 channel
> mode.  Didn't change the actual audio - so the actual result was 3x speed
> chipmunks in all 6 channels.  (Far out!)  But does seem to show that that
> emu10k1 driver behaves as advertised.
> 
> Steve
>