Attention is currently required from: Hoernchen, fixeria, pespin.
1 comment:
Patchset:
I think it's currently lacking some effective control to avoid unbounded buffer queue growth (sorry if I missed it)
The sample clock of your ALSA device and the GSM frame clock will inevitably have independent sources, and one will be running faster than the other. If the GSM clock is faster than your audio clock, your buffer queue will grow more and more, and audio delay increase.
A simple hack might be to have a limit for the queue. We'd then drop a frame once it overflows. Maybe that's "good enough" for a first implementation, and simple enough to do already in the initial version of the patch
The smarter approach would of course be to simply drop some of the decoded audio samples here or there before playing them, causing smaller/shorter disruptions in the audio.
To view, visit change 30337. To unsubscribe, or for help writing mail filters, visit settings.