End-to-end multi-talker audio-visual ASR using an active speaker attention module

Re-assign community