Use case:
After wakeup word is detected, I'd like to do TDOA estimation to tell the position of the speaker. TDOA algrithm generally needs 2+ channel's data (depending on MIC array) to process.
Current status:
After AFE processing, only one channel data (the channel which contains speech. But even all channels contain speech, only one channel is returned) is retained.
Requirement:
AFE returns all channels which contain speech.
Use case: After wakeup word is detected, I'd like to do TDOA estimation to tell the position of the speaker. TDOA algrithm generally needs 2+ channel's data (depending on MIC array) to process.
Current status: After AFE processing, only one channel data (the channel which contains speech. But even all channels contain speech, only one channel is returned) is retained.
Requirement: AFE returns all channels which contain speech.