I've tried using appr.tc
and looks like when there is a sudden delay in network, audio will become tearing / robot-sounded, jitter in webrtc-internals for auido looks like this:
After reviewing the code of webrtc.audio modules in neteq, I noticed the peak detection was removed (M93), and default quantile for IAT becomes 97%:
Since based on my understanding, if the sudden jitter exceeds 97% of the IAT values, jitter buffer in neteq is not able to handle such packets, leading to poor audio quality.
And what is more, adding playout delay seems not working as well, since it doesn't handle the jitter buffer in neteq.
So, I've got questions:
1. Is my understanding correct, if we have sudden delay time to time (sudden jitter and recovered), do we have any strategy to further improve such case?
2. If my understanding is correct, since the configuration of audio jitter buffer is only available in native webrtc, but NOT in js API, do we have any plan to support setting jitter buffer size?
Thanks for your help and time~