Question about Jibri recordings & audio channels

First, thanks a lot for your work with jitsi. Amazing, really!

Is there a way to configure jibri to preserve per participant audio channels in the generated recording file (goal: precise speaker identification) or should I use/fork https://github.com/jitsi/jigasi to do that ?

Best regards
Nicolas

I finally switched to jigasi to do that but, despite many many hours of digging, I cannot get it work.

I though I found a workaround customizing https://github.com/jitsi/jigasi/blob/master/src/main/java/org/jitsi/jigasi/transcription/Participant.java#L527 to store per participant data on dedicated wav files but once a speaker mute itself recordings get de-synchronized since no data is received.

I also looked at https://github.com/jitsi/jirecon but it is not clear to me how to split audio packets coming through the MediaDevice, based on participant SSRC, and manage situation where a speaker is muted.

@damencho Could you please give me some advice on the right way to go?
Alternatively, is there a way to let jigasi know when a participant mute/unmute itself?

Best regards
Nicolas

when user mute/unmute its audio, jigasi receives a
ChatRoomMemberPresenceChangeEvent with

  • type "MemberUpdated"
  • associated sourceRoom ChatRoomJabber
  • associated member ChatRoomMemberJabber
  • reason null

Is there a way to get some audiomuted attribute for associated member ?