[pulseaudio-discuss] Mixing streams

[Lennart Poettering]

On Mon, 19.11.07 17:45, Dennis Fleming (arsantiqua at sbcglobal.net) wrote:

> I missed a few of the responses.  My questions came faster than my
> registration.  Anyway, I don't think DRC is the answer in my case.
> I am working on an embedded system and expect pulseaudio to route
> and mix as needed.  The driver I have now is quite limited and not
> conducive to mixing streams.  I envision applications sending data
> at full resolution and since they would not have knowledge of other
> streams there is no way for them to decrease their dynamic range in
> anticipation of the final mix.  It seems that the simplest means for

Yes, what I was proposing is to add DRC to the PA mixer. That way
clients wouldn't need to know anything about how many streams are
active. The fact whether DRC is applied or not would not be visible to
the clients. And PA would apply DRC only to the mixer output if
required to make sure that we never clip. As long is only one stream
is played DRC will never be needed.

> limiting would be to divide the output pf the mixer by the number of
> streams present inside pulse.  I haven't fully thought out what this

This is not really an option, because it would decrease the volume of
all active playback streams if a new stream connects -- even if that
one just plays silence.

> might mean when some of the streams are close to silence, but as a
> first stab it might make sense.  I'm a little afraid of the
> artifacts from rapid volume change in each stream, but ramping the
> changes would help.
>
> I'll play with it a bit.

Lennart

--
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net         ICQ# 11060553
http://0pointer.net/lennart/           GnuPG 0x1A015CC4