On Sun, 11 Nov 2007, Jared D. McNeill wrote:
>> ... or sox? Mixing multiple audio streams in userland comes to mind.
>> Would that be possible? How?
>
> Sure, wouldn't be hard..
>
> The device driver currently always outputs the same format (44kHz 16-bit 
> stereo slinear_le). You could have an application reading from 
> /dev/sound[1-8], mix the results together, and feed them back to /dev/sound0 
> (the real hardware device).
>
> You could even get fancy and make /dev/sound a symlink to another pseudo 
> device that selects the first available pseudo audio device..

Sounds cool. And besides some frobbing to get volume for all/each source 
right, I wonder if the actual mixing couldn't be put into the kernel too, 
after that...


  - Hubert