How to implement continuous speech recognition same as other speech recognition and display it as streaming?

[Bruce337f]

I checked most of the discussion topics about continuous recognition and found that they are basically about recording audio on the fly through a microphone to return streaming recognition results, but there is no continuous recognition method for the audio file itself.
I want to achieve the same continuous recognition effect as Microsoft speech recognition. What do I need to modify?

[phineas-pta]

u shouldn't

whisper work best with 30s audio input