[Freeswitch-users] Correlation between jitter/discarded_packets with choppy audio streams

Unai Rodriguez unairodcas at gmail.com
Mon May 18 10:20:22 UTC 2020


Dear List,

We're experiencing choppy/distorted audio streams at times and have been
looking into understanding the definition and units of measure of these
values reported on the main FS log (i.e. /var/log/freeswitch/freeswitch.log)

- r = received
- l = lost
- j = jitter
- p = playout_delay
- d = discarded_packets
- i = ignored_packets

The following log lines seemed to correspond with choppy/distorted audio:

========================================================
...
2020-05-15 10:18:05.202714 [INFO] mpf_rtp_stream.c:541 () Close RTP
Receiver 10.252.148.75:18800 <- 10.252.148.73:17592 [r:403 l:0 j:651 p:0
d:87 i:0]
...
2020-05-15 10:25:53.622708 [INFO] mpf_rtp_stream.c:541 () Close RTP
Receiver 10.252.148.75:22660 <- 10.252.148.56:18572 [r:402 l:0 j:143 p:0
d:84 i:0]
...
2020-05-15 10:43:42.642722 [INFO] mpf_rtp_stream.c:541 () Close RTP
Receiver 10.252.148.75:18940 <- 10.252.148.72:20800 [r:356 l:0 j:838 p:0
d:79 i:0]
...
2020-05-15 11:00:52.102707 [INFO] mpf_rtp_stream.c:541 () Close RTP
Receiver 10.252.148.75:20952 <- 10.252.148.61:16540 [r:582 l:0 j:806 p:0
d:66 i:0]
...
========================================================

At this point we believe that having a (d)iscarded value of more than 60
correlates with choppy calls and most of the time those come with (j)itter
value of several hundreds.

Would it be possible for someone to point out where we could find the
definition and unit of measure of those values (i.e.: r, l, j, p, d, i)? Is
there any baseline that could be references to determine which calls are
being choppy/distorted based on these numbers? We've been trying to
understand FS source code and find documentation to no avail.

Thank you so much
Unai Rodriguez
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20200518/8108fb9a/attachment-0001.html>


More information about the FreeSWITCH-users mailing list