We have deployed jitsi meet 1.0.4025 & jitsi videobridge 2.1.183 on our own servers.
Jitsi Meet and Videobridge servers are running on separate Virtual Machines as described in scalable installation tutorial.
Each VM size is 4 CPU Cores and 8GB RAM. Each VM can go upto 10GB of network throughput.
We have also implemented jwt token based authentication mechanism.
We are trying to tune video and audio parameters to achieve best experience for end users.
We have primarily two use cases as described below.
We should be able to hold video conferencing meeting of around 25 participants in each room. We can have 100s of such meetings running in parallel. We can horizontally scale number of videobridges and meet servers as per requirement.
Hosting a webinar with 1 Host, 1 or 2 moderators, 4 to 5 panelists and upto 100 attendees. Our next goal is to have 1000+ attendees in one session.
Hosts and Panelists will have their audio / video streams on and they can mutu / unmute themselves whenever required. Moderator will be able to mute anyone or everyone.
Attendees will always start audio and video muted. However they can raise hand, unmute themselves and ask queries to host or panelists.
We have been playing with following parameters to achieve best experience in terms on audio and video.
Modified video constraint from minimum of 480p to maximum of 720p.
We were looking for tuning audio related parameters such as audio bitrate, however could not find much in configuration files.
Above settings gave very good video quality however many people were not able to turn on video due to limitations in bandwidth.
Also as number of participants increased beyond 5 audio jitters and breaking was observed and subsequently video was also getting jittery and unstable.
With stereo = true lot of echo and bass was observed.
Finally we had to revert back to original settings. However overall video and audio experience looks to be very degraded even for people with very good bandwidth (1 Gbps).
Need your advice in terms of configuration settings or some pointers to look for in order to improve overall audio and video experience in conferences.
Please let me know if you require any further inputs from my side.
Your setup is a bit above my knowledge in terms of systems administration. However I thought I’d share a previous post I wrote about best practices. I will help a LOT if your participants and moderators agree on some simple participation rules.
It also occurs to me for large audiences you may consider streaming to another service instead of accepting active (and possibly disruptive) participants, just a thought. Good luck with your setup.
I agree that the best practices mentioned in the post will help to save bandwidth and improve overall experience.
However these can be useful for smaller group of people and will always not be practical in large group scenarios such as webinars due to timing issues( not everyone will join at the start of event or go through set of instructions) and is also dependent on each person’s understanding of such systems.
Can you provide us guidance in terms of technical parameters on platform side which we can tune or optimize to have overall better experience in terms of audio and video.
As mentioned previously the priority would be to get audio to be stable and crystal clear. So far we have not found any settings similar to video constraints where we could mention minimum and maximum video resolution.
Based on the statistics shown in the room what we have observed is audio is always around 34kbps ( 74kbps if we enable stereo ). Is there any settings to modify audio codec or increase the audio bandwidth?
Similarly for video is support for VP9 or VP8 is stable? Do you recommend to try these out in production systems?