I've been trying to incorporate music into my conferences. Even while conversational voice is consistently good, music is consistently terrible.
I think I narrowed the cause to audio processing. In the process of testing, I passes the audio=echoCancellation=false parameter to an apprtc conference and found that to totally fix the issue.
When I set the disableAEC and disableNS config parameters to true in Jitsi, however, I get some but not all the improvement versus Apprtc. I'm assuming the difference has to do with Jitsi explicitly setting all audio constraints on or off while Apprtc only explicitly turns AEC off when passed that parameter (no audio constraints at all when starting Apprtc session without any parameters).
Is it possible to expose the additional constraints (High Pass Filter and Auto Gain Control) in the same way AEC and NS are or are there issues with that? Or is there another approach?