[asterisk-dev] IAX Voice Framing

[Tilghman Lesher]

On Friday 05 February 2010 13:31:16 Manivasagam Sivaraman wrote:
> I'm trying to pack 3 or 4 audio voice frames into one IAX audio packet and
> send it out. The receiving end will also be the same asterisk that would
> expect the 3 or 4 voice frame in one IAX audio packet. Before touching the
> code I want to know if this is feasible. I'm new to asterisk and I want to
> have a heads up before I begin. Please let me know if there will be some
> risk involved in this. I see a function in frame.h that allows framing of
> multiple voice frames in one RTP packet for instance in sip and h323
> channels. However no such function is used in chan_iax2.c file. Could you
> please expalin why ? Why iax channel is coded a bit different ?

Clearly, because the IAX channel does not use RTP.  This is a feature, which
helps IAX to more cleanly go through NATs, and it may work in places where
SIP does not, such as double and triple NAT situations.

I really don't understand why you'd want to pack multiple frames into a single
packet.  That increases latency, which for a real time protocol is rather bad.

-- 
Tilghman Lesher
Digium, Inc. | Senior Software Developer
twitter: Corydon76 | IRC: Corydon76-dig (Freenode)
Check us out at: www.digium.com & www.asterisk.org