[pulseaudio-discuss] Bluetooth HSP sink is more likely to cause buffer underrun when moving a live stream to it from ALSA, how to ask application to write more data?

Lin, Mengdong mengdong.lin at intel.com
Fri Jun 17 09:17:04 PDT 2011

> As far as I understand it, this information is already supplied to the
> application via the timing information and underrun callbacks.
> When an underrun is hit, it would make sense that the applications
> checks the latency information and adjusts it's flow rate accordingly.

Thank you, Col! I'll check the application code to see if there is something I can do.

But, I still have question on pulseaudio:
1. why buffer underruns are more likely to happen for a Bluetooth HSP sink, especially after moving sinks?
If I connected the BT headset at first, and then launch the VOIP application, the sink input latency is about hold 20 ~ 30ms although there are still underruns from time to time. 
But if I launch VOIP and then move to BT sink, the sink input latency is 0, frequent underruns and rewind flood happens. Why moving cause such a difference?

2. But if I use ALSA, sink input's latency is usually 50m ~ 80ms. Why it works better than a BT sink?


More information about the pulseaudio-discuss mailing list