Transcribing mono audio with voice separation.

Question

I'm trying to use the Transcribe Call Analytics service to transcribe call center audios. However, when the audio is mono, the separation between AGENT and CUSTOMER doesn't work because I imagine it's using each channel side of the audio for this separation (I'm passing via API that Agent = 0 and Customer = 1).

Is there any way to transcribe with the correct separation when the audio is mono? Or any other API or service that I can add or use in conjunction to solve this problem?"

Answer

I dont think that is a option for this on Call Post Analytics

Answer

Hi,

Did you try Partitioning speakers (diarization) ? See https://docs.aws.amazon.com/transcribe/latest/dg/diarization.html

See examples with console or CLI at bottom of page

Best

Didier

Transcribing mono audio with voice separation.

Relevant content