Sebastian Krzyszkowiak authored Feb 01, 2021 and Guido Gunther committed Feb 14, 2021

module-echo-cancel uses WebRTC AEC algorithm to filter out acoustic echo.
It's mostly meant for desktop use with speakers and microphone, therefore
by default it does additional processing for things like voice detection
or noise suppression in order to filter out things like keyboard typing.
However, it appears that at times this filter may be a bit too overzealous
for our mobile phone use case, with users reporting the voice coming out
on the other side as "muffled".

Since echo cancellation is all that we cared about when enabling the
filtering, let's disable any kind of processing that's not related to
removing the echo to make sure it doesn't have any unintended consequences.
In the end nothing prevents us from selectively reenabling particular
features after enough testing.

Note: module-echo-cancel also supports AECM (mobile) variant of the algorithm
that theoretically should better match our use case. However, so far I wasn't
able to make it work (my current suspicion is that we have to reduce the overall
latency in order for it to actually filter out the echo).