Can it be used for transcribing streaming audio?

#20
by AB200 - opened

Hi, I was looking for solution of ASR for steaming. While canary1-b and parakeet 1.1b is good for non steaming use-cases, if I wan to use them for streaming the accuracy reduce significantly, I understand it happens due to lack of future context, but this problem can be overcome by audio buffer method, while I found
documentation for buffer method of "stt_en_conformer_ctc_large" model, there are no such docs for these 2 models (canary1-b and parakeet 1.1b). They don't give any output with the current implementation for buffer method.

So I want to if canary model can be used for streaming, if yes, is there any documentation/example of that?

Any news on this?

Of course it can be used, but it is up to external program to provide buffering and transcription.

Sign up or log in to comment