[Alsa-user] Channel reduction problems

[Duncan S. <dun...@ma...>]

Hi all, I have the following problem when reading
from /dev/dsp (oss-emulation): when recording, if
the source has two channels but the destination
(i.e. my program) only wants one channel, then the
sound I record is drowned out in a loud pulsing sound.
If the destination has two channels as well, then there
is no problem.

Now, in the situation of 2 channels -> 1 channel, the
oss-emulation layer does "channel reduction".  I had
a look at the source code but it made my head hurt...
Does anyone else have this problem?  Or could someone
with a harder head than mine explain how the channel
reduction is supposed to work, so I can make sense of
the spaghetti?

Thanks for any help!

Duncan Sands.

PS: The sound is S16_LE.
PPS: The sound read into the buffers is ok before
the channel reduction.
PPPS: Since this is kernel code, it uses integer
arithmetic, rather than float arithmetic (which is
supported in the code - presumably because the
code is shared with the oss-lib or something like
that).