is aec3 suitable for audio source separation

[Andrew Wang]

Hi,

I try to use webrtc/aec3 to do semi-blind-audio-source-separation task in the following scenario:

Some vocal speech and one piece of music are mixed into one audio track via sound card, which should do some signal processing. Then I try aec3 to separate vocal speech and music from the mixed audio, where mixed audio is input for audioproc_f, and vocal speech is reference input. But I can still hear some echo residual, i.e. vocal speech.

Is aec3 suitable for the scenario? Thanks!

Regards

andrew