Does model.refine() work with faster-whisper model?

[dgoryeo]

I am trying to use the refine method with the faster-whisper model. However, I encounter the following error:

Traceback (most recent call last):
    ...
    model.refine(movie_path, result)
AttributeError: 'WhisperModel' object has no attribute 'refine'

My code:

import stable_whisper

model = stable_whisper.load_faster_whisper('large-v2', device="cuda", compute_type="int8_float16")
result = model.transcribe_stable(movie_path, **kwargs)
model.refine(movie_path, result)

Am I missing something? It appears that the refine method is not available for the faster-whisper model. Is the refine method supposed to be available for the faster-whisper model? If not, is there an alternative approach to refine the transcription results using faster-whisper?

Thank you!

[jianfch]

That error is expected because refine is not implemented for the faster-whisper models. For more faster refinement, an alternative to use refine with a smaller model than the one used for the transcription (e.g. medium if transcription was done with large).