About MeetMe voice quality question

Hello everybody,
i met a boring issue of Meetme.i made a test,using cell phones or telephones call directly into Meetme through a cisco GW(as5300). 4 calls voice(codec 711ulaw) were mixxed by Meetme. the CPU or Memory is powerful enough (2*2.5GHZ and 4G memory)
then i capture the RTP stream on the Meetme server by tcpdump command.
download to my computer,and recovered 4 RTP stream into 8 .au files.one call became 2 .au files(one is in stram,another is out).
i split a short time slot,in this time-slot, A talk,B listen.In moral A-in should almost equal B-out.
But in fact,i found there a about 0.04second slience time-slot was insert into B-out .au file.that mean,MeetMe insert a short slience time-slot during mixing the calls voice.and made the Normal rtp stream delay.
i have a voice graphic image made by cooledit.but i dont know how to upload to show.
but those kind of issue was not alway happend when MeetMe mix calls voice.i tested many times,some time MeetMe work well ,no slience time-slot inserted.very strange.
do i express clearly? thank you.

I don’t think I understand, but I’ll offer this:

MeetMe maintains an audio buffer in order to perform mixing. As audio arrives, a buffer is filled into the mixer, and then audio is sent out. So, there’s going to be delay associated with using a MeetMe conference bridge.



I can’t be sure from your description, but if you’re testing this with there was a recent fix for an issue were Meetme was dropping 2ms of audio every 4 ms.

See r10205 on trunk:

svnview.digium.com/svn/dahdi/lin … iew=markup

and r10207 on the 2.5 branch:

svnview.digium.com/svn/dahdi/lin … iew=markup

These revisions have not yet been rolled into a release but will be soon.

Is this the issue you’re running into?