from faster_whisper import WhisperModel

model = WhisperModel('medium', device='cpu', compute_type='int8')

videos = [
    ('video1_audio.mp3', 'video1_transcript.txt'),
    ('video2_audio.mp3', 'video2_transcript.txt'),
]

for inp, out in videos:
    print(f'Starting: {inp}', flush=True)
    segs, info = model.transcribe(inp, language='zh', beam_size=1)
    segs = list(segs)
    with open(out, 'w', encoding='utf-8') as f:
        for seg in segs:
            f.write(f'[{seg.start:.1f}s] {seg.text}\n')
    print(f'Done: {out} ({len(segs)} segments)', flush=True)

print('ALL DONE')