pipeline/utils.py

import ffmpeg

def extract_audio(video_path: str, out_audio: str = "audio.wav") -> str:
    (
        ffmpeg
        .input(video_path)
        .output(out_audio, ac=1, ar=16000, format="wav")
        .overwrite_output()
        .run(quiet=True)
    )
    return out_audio