RE: [Sipping] Use cases of multi-transcoding

On Tue, 21 Feb 2006, Henry Sinnreich wrote:

> The numerous wireline legacy ITU-T G.7xx codecs and other countless codecs
> deployed for mobile phones are a disgrace IMHO.

Tell me about it, I have been working for 2 years to get CPE manufactures
to support some updated CODECs. The big problem has always been that the
manufactures sell CPE to service providers with a PSTN leg in over 90% of
there calls. As a service provider you have two options to connect to the
PSTN, you can go out and buy some trunking gateways and do it yourself or
use a GLBX, XO, BWING of the world and hand them SIP and let them deal
with that. So the issue is that even if you support Speex at the edge,
there is no way your going to had that off to a Sonus or other gateway.

> The industry would be better off adopting "Internet codecs":

We have started to see some light at the end of the tunnel on this, some
of the SBC vendors are now supporting transcoding. This would let a
service provider support say Speex at the edge of their network and then
hand off G.711 to origination/termination partners. If a call was CPE to
CPE you could just use wideband.

> The iLBC RFC 3951 and RFC 3952 as the default codec and SPEEX for variable
> rate wideband
> (ftp://ftp.rfc-editor.org/in-notes/internet-drafts/draft-ietf-avt-rtp-speex-
> 00.txt).

I have been very disappointed in iLBC, all my testing puts it just below
G.729 in quality. If you look at that night graph from the guys are GIPS
it shows that it does much better with packet loss, but rumor is that they
got that data by using very sub optimal G.729 settings. Speex however has
shown to be a great CODEC along the full range of bit rates. I think the
best way to deal with packet loss with CODECs that take up small amounts
of bandwidth is to send duplicate frames.

><>
Nathan Stratton
nathan at robotics.net
http://www.robotics.net

_______________________________________________
Sipping mailing list  https://www1.ietf.org/mailman/listinfo/sipping
This list is for NEW development of the application of SIP
Use sip-implementors@cs.columbia.edu for questions on current sip
Use sip@ietf.org for new developments of core SIP