Modifying streaming sound in realtime

I'm working on an application that is streaming audio synthesized on
the fly.  When a new synthesized sound begins, the response is very
rapid because I can grab a free source and immediately begin queuing
up buffers.  The difficulty arises when the currently streamed audio
needs to change.  In the worst case, the system will have just pushed
a newly filled buffer onto the queue and so the change will not be
realized until both the current buffer and queued buffer are
processed.  It seems that even if I shrink my buffers down to as small
as they can be without causing starvation problems, there is still
noticeable lag.  If we assume that I can "rewind" the synthesizer back
to the present time (it needs rewinding because it has already spooled
off a frame of sound into the future without accounting for the change
that just occurred), I could stop the source, de-queue the buffers and
then refill and re-queue with the newest data but obviously that will
result in an undesirable hitch in the sound stream.  Is there a better
way?  How can I change (not just stop) a streaming source with minimal
lag?

jc
_______________________________________________
Openal mailing list
[email protected]
http://opensource.creative.com/mailman/listinfo/openal