Acoustic echo cancellation (AEC) support - feedback loops or erroneous recognition of spoken commands

[marto]

Hi,

I had looked at the docs and the code and I noticed there is no support for Acoustic echo cancellation. I'm playing about with the library and using listen_in_background to obtain segments of speech picked up from a microphone. This works fine, but when sounds are emitted from the system's own speakers, such as in a conferencing or voice assistant scenario the callback function is being called because the system is capturing its own output as input, leading to issues like feedback loops or erroneous recognition of spoken commands.

I looked at integrating this myself in the callback method (via speexdsp), but quickly realised that the callback will be out of sync with what is being played out of the speakers as the callback is called multiple times (due to pauses).

I think it kind of needs to be part of this library? What do you think?