Can this be used to stream audio over websockets?

[timmolter]

... rather then record and entire chunk and at the end send it all at once?

It would be cool if it started streaming the words back, while you were still talking.

[rpdrewes]

This could be done, and I have played with it. The interface between the phone (from Konele) and the server supports incremental results. Basically the accumulated audio is reportedly submitted to transcription as audio is gathered. I personally do not like this incremental recognition approach from the client side; I find it distracting. If there is enough demand for this I could add this feature to whisper-websocket-server but I don't plan to right now. You could do it and let us know!