以下、指定方法
sample.py
transcript = openai.Audio.transcribe(model="whisper-1",
file=audio_file,
response_format="text",
temperature=0.2,
language="ja",
prompt="This is a transcription of the live broadcast of a European bicycle race.")
#json, text, srt, verbose_json, or vtt.
print(transcript)
(参考)
OpenAIの文字起こしWhisper。試したらすごかった。
https://note.com/santarou/n/nbc43e3d8349e