Transcribe audio files using OpenAI's whisper-1

transcribe-audio-files.py

#!/usr/bin/env python3


import pathlib
import pprint
import sys

import cachier
import openai


audio_filenames = sys.argv[1:]

home_folder = pathlib.Path.home()
api_key_filename = f'{home_folder}/.config/openai/api-key.txt'
openai.api_key = open(api_key_filename).read().strip()


@cachier.cachier()
def get_transcript(audio_filename):
    with open(audio_filename, 'rb') as audio_file:
        response = openai.Audio.translate('whisper-1', audio_file)
        return response['text']



transcripts = []
for audio_filename in audio_filenames:    
    transcript = get_transcript(audio_filename)
    print(transcript)
    transcripts.append(transcript)


print(' '.join(transcripts))