[jitsi-users] Problem with setting a mixer mode for audio channels


#1

Hi,

My colleagues filed today an issue about a problem with setting a "mixer" mode for audio channels:
https://github.com/jitsi/jitsi-videobridge/issues/9

It would nice to hear if it is really a bug or intended behavior.
If it is a bug, it would be nice to get it fixed as it has broken our setup. Hopefully, fixing this is pretty straight-forward.

Thanks,
Leo


#2

На 3.07.2014 19:54, Leo Romanoff написа:

Hi,

My colleagues filed today an issue about a problem with setting a "mixer" mode for audio channels:
https://github.com/jitsi/jitsi-videobridge/issues/9

It would nice to hear if it is really a bug or intended behavior.
If it is a bug, it would be nice to get it fixed as it has broken our setup. Hopefully, fixing this is pretty straight-forward.

Thanks,
    Leo

Thank you. Does https://github.com/jitsi/jitsi-videobridge/commit/03398da2940a3c08b543e4a61bca179ece3b87d6 fix it for you?


#3

Hi Lyubomir,

Thanks for looking into this!

https://github.com/jitsi/jitsi-videobridge/commit/03398da2940a3c08b543e4a61bca179ece3b87d6 is a step in the right direction. It fixes one part of the problem. But there are still some issues with the "source" elements inside colibri payloads. More detailed explanation and traces can be found in the our comments on github.

Thanks,
Leo


#4

На 4.07.2014 09:40, Leo Romanoff написа:

But there are still some issues with the "source" elements inside colibri payloads. More detailed explanation and traces can be found in the our comments on github.

I didn't know there was a problem with the source element.

Devs, I see that the pre-announcement of the audio SSRC has been disabled. Why is that?


#5

На 4.07.2014 09:40, Leo Romanoff написа:

But there are still some issues with the "source" elements inside colibri payloads. More detailed explanation and traces can be found in the our comments on github.

Thank you! Does https://github.com/jitsi/jitsi-videobridge/commit/6504887c5fb844389dfc7ced1feac1da1436b498 fix it for you?


#6

Hi Lyubomir,

Thanks! The new fix improves the situation a lot. Now mixed audio can be received, but behaves in a very strange way still. Please see our comments on github for more details.

BTW, we also provided in our github comment a few observations about lastN and active speaker detection functionality, which was part of our tests as well. May be it is useful. Specifically, the lastN events and active speaker events are not sent in some situations, where in our opinion one could expect them to be sent (e.g. when a new participant joins, etc). And even in a stable mode (i.e. no new or leaving participants) they are sent not as often and as in-time, as we expected them to be sent...

Thanks,
Leo

···

Lyubomir Marinov <lyubomir.marinov@jitsi.org> schrieb am 9:54 Freitag, 4.Juli 2014: На 4.07.2014 09:40, Leo Romanoff написа:

But there are still some issues with the "source" elements inside colibri payloads. More detailed explanation and traces can be found in the our comments on github.

Thank you! Does
https://github.com/jitsi/jitsi-videobridge/commit/6504887c5fb844389dfc7ced1feac1da1436b498
fix it for you?


#7

No real reason -- I didn't take into account that we should support
audio mixing and it is needed in that case. Sorry, and thank you for
taking care of it.

Boris

···

On 04/07/14 10:02, Lyubomir Marinov wrote:

На 4.07.2014 09:40, Leo Romanoff написа:

But there are still some issues with the "source" elements inside colibri payloads. More detailed explanation and traces can be found in the our comments on github.

I didn't know there was a problem with the source element.

Devs, I see that the pre-announcement of the audio SSRC has been
disabled. Why is that?


#8

На 4.07.2014 13:22, Leo Romanoff написа:

Hi Lyubomir,

Thanks! The new fix improves the situation a lot. Now mixed audio can be received, but behaves in a very strange way still. Please see our comments on github for more details.

Thank you for the feedback! Please report the new issue(s) on the dev mailing list in a new thread because this thread is concerned with setting the mixer rtp-level-relay-type attribute value.

BTW, we also provided in our github comment a few observations about lastN and active speaker detection functionality, which was part of our tests as well. May be it is useful. Specifically, the lastN events and active speaker events are not sent in some situations, where in our opinion one could expect them to be sent (e.g. when a new participant joins, etc). And even in a stable mode (i.e. no new or leaving participants) they are sent not as often and as in-time, as we expected them to be sent...

This is a totally different topic, please report that to the dev mailing list in a new thread.