recognizer.recognize_whisper doesn't work
Bae-ChangHyun opened this issue · comments
It works fine when call the whisper by installing the whiper directly on the local, but if i call the whisper through thie speech_recognition, it will occurs error.How can i do
When calling whisper directly, model = whisper.load_model(var_s.model_type)
result = model.transcribe("./mp3/tmp.wav",fp16=False)
Like this,
When i call whisper through speech_recognition:
Recognizer.recognize_whisper (audio, language="ko-KR") Is this the right way to run it? there is no enough explanaion. Is it possible to directly input an audio object rather than an audio file into the whisper?
For reference, audio=recognizer.listen (source, timeout=30).