Summary:ASTERISK-05274: iax trunking with ilbc has audio distortion with vm only
Reporter:radamson (radamson)Labels:
Date Opened:2005-10-10 09:15:41Date Closed:2006-05-09 09:54:50
Versions:Frequency of
Environment:Attachments:( 0) msg0000.WAV
Description:Two asterisk boxes 150 miles apart, both cvs-head as of Oct 7th (same since Sept 27th at least), connected via iax2 with low-utilized ds3 internet, C7960 calls exten on remote system (also C7960), and call goes to VM. No other calls in either system (eg, no load).

Both boxes have iax config'ed as:
Recorded VM messages are very distorted. Normal calls (non-vm) sound fine. Use of trunktimestamps=yes had no impact on the distortion.

Changing only jitterbuffer=no (and * restart), recorded VM messages are very clean. With jitterbuffer=yes and trunk=no, messages are very clean.

Issue can be recreated easily and 100% of the time.


If both boxes config'ed as:
Recorded VM messages are very clean.

Conclusion: looks like the combination of trunk=yes and jitterbuffer=yes with ilbc is causing the distorted VM messages. Possible timing issue? Normal answered calls have no distortion. The distortion sounds like a jitterbuffer problem. As noted above, easy workaround is to use gsm instead of ilbc.
Comments:By: BJ Weschke (bweschke) 2005-10-13 21:51:00

I was able to reproduce this in the code zone this evening. Uploaded an example WAV file.

By: stevekstevek (stevekstevek) 2005-11-21 12:34:12.000-0600

does this also happen if you use trunktimestamps?

By: Brian West (bkw918) 2005-11-21 14:23:49.000-0600

I bet this has to be because ilbc is a 30ms codec.


By: stevekstevek (stevekstevek) 2005-11-21 14:33:52.000-0600

Probably, and I think that trunking without trunktimestamps still sucks when the trunk interval != the codec interval..  It should be OK with trunktimestamps, though.

By: BJ Weschke (bweschke) 2005-12-12 20:17:20.000-0600

radamson9: pls respond as to whether or not trunktimestamps fixed your problem so we can close out this bug.

By: radamson (radamson) 2005-12-12 20:36:39.000-0600

As was stated in the origianl summary, "Use of trunktimestamps=yes had no impact on the distortion".
I've not tried to replicate the problem since I've not seen anything committed that referred to bug 5420. Did I miss something? (Currently running cvs head from Nov 26th.)

By: BJ Weschke (bweschke) 2005-12-12 20:49:12.000-0600

It's possible it's been fixed since then. I'm really not sure. I've not tried to reproduce since my original note on 10/13. Does your current version from 11/26 still exhibit the problem?

By: radamson (radamson) 2005-12-12 20:51:19.000-0600

I'm out of town and won't be able to test until later this week. Will post back in a couple of days.

By: radamson (radamson) 2005-12-16 07:51:21.000-0600

Just retested to identify any differences since this problem was originally reported. Both asterisk systems running cvs head from Nov 26, 2005.
The problem remains as originally documented. Switching between trunktimestamps=yes and =no has no impact on the problem. With trunktimestamps=yes and allow=gsm (only changed the codec), the problem does disappear. So, still an issue with ilbc, jitterbuffer and voicemail in wav format.
I'm not able to update these two systems to current cvs head at this time.

By: Olle Johansson (oej) 2006-02-03 02:22:40.000-0600

Any updates to this issue?


By: Serge Vecher (serge-v) 2006-05-01 14:29:11

is this still an issue?


By: Serge Vecher (serge-v) 2006-05-09 09:54:50

no update from reporter. Please reopen the bug if the problem still exists in recent trunk. Thanks.