Hi there, we’ve recently switched to ParticipantEgress-based recording, migrating from RoomComposite-based recording. We’ve discovered a pretty big issue:
When the participant publishes only audio, with no video, there are gaps in the recording, roughly every 110ms there is a portion of audio missing. This results in very choppy recordings, and the missing portions are totally lost. Over the course of an entire call, between 40-50% of the entire audio is lost
Here’s an egress example ID: EG_i9Yk44cFvDDG (though ALL Participant egresses are showing the same issue).
I’ll point out that during the actual call itself, audio is totally fine. So this is only an issue on the egress recording side.
I narrowed it down to this specific cause (only happens when publishing audio with no video), I tried:
- publish screenshare video + screenshare audio stream: totally fine
- publish camera video + audio stream: totally fine
- publish only audio: broken
- publish audio-only at first: broken; then start publishing camera video: becomes totally fine; turn off video: broken again.
We’ve worked around it by publishing a synthetic black video when the user is publishing only audio, though this is a kind of complicated and potentially breakey workaround, so we’d appreciate a quick proper fix on this!
