fanahova/smol-podcaster

smol-podcaster is your autonomous podcast production intern 🐣

smol-podcaster

We use smol-podcaster to take care of most of Latent Space transcription work. What it will do for you:

• Generate a clean, diarized transcript of the podcast with speaker labels and timestamps
• Generate a list of chapters with timestamps for the episode
• Give you title ideas based on previous ones (modify the prompt to give examples of your own, it comes with Latent Space ones)
• Give you ideas for tweets to announce the podcast

To run:

python smol-podcaster.py AUDIO_FILE_URL GUEST_NAME NUMBER_OF_SPEAKERS

The URL needs to be a direct download link, it can't be a GDrive. For files <100MB you can use tmpfiles.org, otherwise Dropbox. For example:

python smol-podcaster.py "https://dl.dropboxusercontent.com/XXXX" "Tianqi" 3

The script will automatically switch https://www.dropbox.com to https://dl.dropboxusercontent.com in the link.

Environment Setup

Activate virtualenv with

source venv/bin/activate

Install dependencies with

pip install -r requirements.txt

Make a copy of the .env.sample and replace it with your keys:

mv .env.sample .env

Run with the command above

License

MIT License

smol-podcaster's Issues

Python error with Replicate

Hey,
just tried running this script but came across an error:

  File "/Users/jan/ai/smol-podcaster/smol-podcaster.py", line 226, in <module>
    main()
  File "/Users/jan/ai/smol-podcaster/smol-podcaster.py", line 193, in main
    transcript = transcribe_audio(url, name)
  File "/Users/jan/ai/smol-podcaster/smol-podcaster.py", line 17, in transcribe_audio
    output = replicate.run(
  File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages/replicate/client.py", line 146, in run
    raise ModelError(prediction.error)
replicate.exceptions.ModelError: [Errno 2] No such file or directory: '1692044373.3206804-recording.wav'

I'm not sure where the .wav file comes from or should get created?