A python script that combines OpenAI's Whisper and Chat-GPT to summarize audio from a class/lecture (or a custom task)
- Custom prompt (task)
- Custom chunk size (how many words you feed GPT-3 at once)
- Generate or load transcription (for reuse with the same audio)
Use the instructions from these two github repos to install Whisper and Chat-GPT Wrapper in one or two commands:
https://github.com/mmabrouk/chatgpt-wrapper
https://github.com/openai/whisper
- Open up notes.py to customize your settings (change loadTranscription to True if you want to run it using previously generated transcription)
- Place your audio in the makeNotes folder name it "samp.mp3" (or wav, flac, aiff, whatever's in your settings)
- Run the script and enjoy
In most cases where there's issue with the Chat-GPT Wrapper, you can do the following to fix it:
- Run 'pkill firefox' in your command line
- Run 'chatgpt install' in your command line, log out and back into your OpenAI account
- Quit the browser and rerun the script