model.py

from faster_whisper import WhisperModel

model_size = "turbo"

model = WhisperModel(model_size, device="cpu", compute_type="int8")

def audio_transcribe(audio_path) -> str:
    segments, info = model.transcribe(audio_path, beam_size=5)
    full_transcript = " ".join([segment.text for segment in segments])
    
    return full_transcript

if __name__ == "__main__":
    audio_path = "T0055G0007S0001.wav"
    transcript = audio_transcribe(audio_path)
    print(transcript)