Hello!
I need access from python to real time PCM data from PJSIP and I've found
here, in the list, a patch to create a custom port to make callback calls
to python when RTP frame arrived and parsed:
http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/2014-March/037650.html
And it works. At least on a developer machine. But not in the Kubernetes.
No incoming audio.
At first i thought the NAT was causing this, but tested with pjsua
softphone in echo mode and there is no problem - rtp flows in both
directions. I've added log outputs on the RTP packet path from
transport_udp.c to conference.c, sound_port.h and finally python UAC. And
when it ran in kubernetes pod I don't see any output. (Well, in python i
receive empty packets). Looks like some races when running PJSIP
kubernetes.
Maybe someone encountered problems like this? I also encounter it with
latest PJSIP release but only in 1 of 20-30 calls.
A second question. Is there another way to get near real time PCM data from
PJSIP to python to process it on the fly?
--
Vovk Donets
python developer