transcribe-podcast.py

import sys
from pathlib import Path

import openai
from pydub import AudioSegment

podcast_file = sys.argv[1]

openai.api_key_path = Path("~/.openai-api-key.txt").expanduser()

podcast = AudioSegment.from_mp3(podcast_file)
ten_minutes = 10 * 60 * 1000
first_10_minutes = podcast[:ten_minutes]

for segment in podcast[::ten_minutes]:
    transcript = openai.Audio.transcribe(
        "whisper-1",
        segment.export("/tmp/segment.mp3", format="mp3")
    )
    print(transcript.text)