The shortest path from an empty checkout to one transcribed, summarised episode. For every install option and prerequisite, see Installation.
git clone https://github.com/willtheorangeguy/Shell-Game-Transcripts.git
cd Shell-Game-Transcripts
pip install -r requirements.txt
pip install git+https://github.com/openai/whisper.gitYou also need ffmpeg on your PATH and Ollama running locally.
ollama pull llama3.1:8bThe summarisation stage calls Ollama over HTTP and fails if the model has not been pulled. This is the most common first-run failure.
python main.py 2025- Audio downloads first, then transcription, then summarisation, then grammar cleanup.
- Transcription is the slow part. On CPU it can take longer than the episode itself. See Installation for the CUDA build of PyTorch.
- Each stage appends to a log as it finishes an item, so interrupting and re-running resumes rather than starting over.
- Output lands beside the audio: a transcript, a summary, and a corrected transcript.
If something fails, Troubleshooting covers the usual causes.