Any Idea for feature update for targeted speaker transcription? about whisper_streaming

[Gldkslfmsd]

OK, I understand.

This is not an issue of streaming, but of audio pre-processing or ASR modelling. You need to ask elsewhere to have such model, and then integrate it for streaming, the same way as Whisper.

[Gldkslfmsd]

hi, what do you mean by "targeted speaker transcription"?
I think that if you use an underlying model that supports it, instead of the default Whisper, than it will work.

[Jeevi10]

Thank you for the prompt response. I mean, suppose multiple speakers speak in an environment ( noisy environment where people are speaking in the background)
I would like to transcribe the main speaker (who is close to mic) only, I tested the "mic_test_whisper_simple.py" it works very well, however it tend to capture all the noises ( still speech but far from mic) in-between my speech.