[Asterisk-Users] IAX Jitter Buffer
We have a customer who is connected to our PSTN gateway using IAX and noticing that even when the traffic from their site is modest their outbound audio has short dropouts. Inbound audio is fine. (They have ADSL so it is expected that outbound audio would be the first to experience problems.) We have several questions to pose to the collective wisdom of this list. Q1: Are there any statistics collected/available or diagnostics tools to tell us how much of this can be attributed to packet loss and how much to packet jitter and to measure quantitatively how bad this is? The use of the jitterbuffer in iax.conf seems to have problems. Extensive searching turns up comments such as: 1. jitterbuffer, unfortunately, is buggy and don't work as expected. [asterisk-users/2003-July/016029.html.] 2. This supports my thinking that there is some sort of broken logic in the IAX jitter buffer - HRH Mark Spencer [http://www.marko.net/asterisk/archives/0302/0077.html] When we enabled jitterbuffer the sound quality seemed to improve but we noticed some problems: (a) sometime we would get only one-way audio; (b) other times we would experience no audio in one direction for between 1 and 4 seconds and then things would seem to work fine; (c) some times users reported a clipped and almost half duplex sound quality as the flow of the conversation shifted back and forth. We also noticed some wingnut values for Lag and Jitter such as: Lag: -65476ms Jitter: 12897799ms PSTN gateway is CVS-04/20/04-01:11:29 Client machine is CVS-HEAD-06/02/04-07:56:41 Searching the Asterisk bug lists shows some significant fixes (1696, 1643). Q2: Is jitterbuf working well enough to try again? Q3: Any other suggestions for improving voice quality with IAX links? Thanks. g. ___ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] IAX Jitter Buffer
On Thu, 17 Jun 2004, George Pajari wrote: Q1: Are there any statistics collected/available or diagnostics tools to tell us how much of this can be attributed to packet loss and how much to packet jitter and to measure quantitatively how bad this is? Q2: Is jitterbuf working well enough to try again? Q3: Any other suggestions for improving voice quality with IAX links? Hi George, I'm looking at the jitter buffer and will persevere until it works right for me. (Here in South Africa Internet quality is not of US standard!) I did find one small problem and have a fix which hopefully will go into CVS. But I think further tweaking is also desirable. I see on bugs.digium.com stevek has also submitted some adjustments which have stimulated discussion. So check asterisk-dev, check bugs.digium.com and I think we'll get the jitter buffering right. Steve ___ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] IAX Jitter Buffer
On Friday 18 June 2004 02:46, George Pajari wrote: (b) other times we would experience no audio in one direction for between 1 and 4 seconds and then things would seem to work fine; I just had this problem with my * setup: KSU - Adit600 - T100P - IAX2(Office) - IAX2(Colo) - IAX2(Nufone) The *Colo box never steps out of the way since *Office is not routeable to *Nufone (no NAT, but rather two network interfaces at *Colo, one going directly to *Office. The Colo box also has a TE405P in it going to the telco PRI for local calls, but dropouts never occured on those calls; only on calls to Nufone. I turned off jitter buffer and moved to the GSM codec at the request of Nufone's technical support department (and turned on IAX2 trunking, I had it disabled since calls between *Office and *Colo would exhibit bursty audio) and the problem went away. So no, I don't think jitter buffer's quite there yet, although I *never* had that problem before this week. Perhaps it's a recent CVS fix. :-) Regards, Andrew ___ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] IAX Jitter Buffer
When we enabled jitterbuffer the sound quality seemed to improve but we noticed some problems: (a) sometime we would get only one-way audio; (b) other times we would experience no audio in one direction for between 1 and 4 seconds and then things would seem to work fine; (c) some times users reported a clipped and almost half duplex sound quality as the flow of the conversation shifted back and forth. We also noticed some wingnut values for Lag and Jitter such as: Lag: -65476ms Jitter: 12897799ms PSTN gateway is CVS-04/20/04-01:11:29 Client machine is CVS-HEAD-06/02/04-07:56:41 Searching the Asterisk bug lists shows some significant fixes (1696, 1643). Q2: Is jitterbuf working well enough to try again? Q3: Any other suggestions for improving voice quality with IAX links? A google search of the asterisk-cvs list indicates there has been several iax changes in the last several months. Iax2 with gsm is working very well between * systems using the current cvs Head. I was told specifically by Mark to include jitterbuffer=no in the iax.conf, but with no explanation as to why. Although I'm not a programmer, causual browsing of the source code would seem to suggest that some sort of dynamic jitter buffer function is in use and attempts to over-ride it might not be a reasonable thing to do. I'd suggest bumping both systems up to current cvs Head, add the statement, and eval the result. ___ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] IAX Jitter Buffer
On Fri, 18 Jun 2004, Rich Adamson wrote: A google search of the asterisk-cvs list indicates there has been several iax changes in the last several months. Iax2 with gsm is working very well between * systems using the current cvs Head. I was told specifically by Mark to include jitterbuffer=no in the iax.conf, but with no explanation as to why. Although I'm not a programmer, causual browsing of the source code would seem to suggest that some sort of dynamic jitter buffer function is in use and attempts to over-ride it might not be a reasonable thing to do. I'd suggest bumping both systems up to current cvs Head, add the statement, and eval the result. jitterbuffer=no turns off that dynamic jitter buffer function. People recommend to turn that off because it doesn't work 100% at the moment. Steve ___ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
