<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
<title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
<br>
<br>
I'm also interested in this, as the other Steve knows.<br>
<br>
Anyone in the re-worked jitter-buffer/PLC/DTX crowd besides me going to
be at astricon?<br>
<br>
We can at least start working there on requirements. I think I've
wrote this before, but here's what I'd _really_ like to see as
requirements for a re-worked jitter-buffer:<br>
<br>
- Channel Support:<br>
IAX2 in asterisk<br>
IAX2 in libiax2<br>
Other IP channels in asterisk (RTP-based ones, I guess are all that
is left).<br>
<br>
- DTX Support: Sending a single CN packet (in IAX2, this should
probably be sent reliably) <br>
would probably be good.<br>
- PLC Support: For codecs that support this natively, use that
support (iLBC, speex, others). <br>
For those that don't, we can add some kind
of hack to the codec. For example, <br>
in app_conference, I repeat even GSM frames
when I detect loss.<br>
<br>
- Configurability by applications: It seems that some applications
(app_fax) might do better without any of this; we should consider what
different applications may want to configure in their use of this, and
allow them to change settings while the channel is in the application.
So, app_fax might want to disable PLC, and also disable any
jitter-buffer shrinkage (and dropped frames that it would cause).<br>
<br>
- For IAX2 (and RTP, I think) Interleaved frame support: For
single-channel calls, in IAX2, and a compressed codec, you have about
100% overhead with 20ms frames. Using larger frame sizes of course
drops this considerably, but without interleaving, it makes PLC much
less effective. Using interleaved frames, where in a single packet you
have, for example<br>
<br>
Packet 1, frame 0,2<br>
Packet 2, frame 1,3<br>
Packet 3, frame 4,6<br>
[...]<br>
<br>
And PLC, means that if you drop packet 2, you can do a much better job
concealing that loss.<br>
<br>
<br>
-SteveK<br>
<br>
<br>
<a class="moz-txt-link-abbreviated" href="mailto:steve@daviesfam.org">steve@daviesfam.org</a> wrote:
<blockquote
cite="midPine.LNX.4.58.0408271453150.10553@bob.daviesfam.org"
type="cite">
<pre wrap="">
On Fri, 27 Aug 2004, Michael Manousos wrote:
</pre>
<blockquote type="cite">
<pre wrap="">I hope that the above issues will start a discussion and result to a
solution, no just for PLC, but also for the DTX operation.
</pre>
</blockquote>
<pre wrap=""><!---->
Yeah - my goal for a reworked jitter buffer includes DTX and PLC. And
other TLAs ;-)
Steve
_______________________________________________
Asterisk-Users mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Asterisk-Users@lists.digium.com">Asterisk-Users@lists.digium.com</a>
<a class="moz-txt-link-freetext" href="http://lists.digium.com/mailman/listinfo/asterisk-users">http://lists.digium.com/mailman/listinfo/asterisk-users</a>
To UNSUBSCRIBE or update options visit:
<a class="moz-txt-link-freetext" href="http://lists.digium.com/mailman/listinfo/asterisk-users">http://lists.digium.com/mailman/listinfo/asterisk-users</a>
</pre>
</blockquote>
<br>
</body>
</html>