Installation¶
Preparing Your Keep¶
Before TaleKeeper can chronicle your adventures, you'll need a few things in place.
Prerequisites¶
- Apple Silicon Mac (M1+) — Recommended for ML features (transcription, diarization, image generation)
- Python 3.12+ — python.org
- ffmpeg — Required for audio processing
- Ollama (optional) — For AI-powered text summaries and session naming
brew install ffmpeg
brew install ollama # optional, for AI text features
sudo apt install ffmpeg
# For Ollama, see https://ollama.ai
Install TaleKeeper¶
pip install talekeeper
Launch¶
talekeeper serve
This starts the server at http://localhost:8000 and opens your browser automatically.
Command Options
| Flag | Description |
|---|---|
--host <address> |
Bind to a specific address (default: 127.0.0.1) |
--port <number> |
Use a different port (default: 8000) |
--reload |
Restart automatically when settings change (for development) |
--no-browser |
Don't open the browser on startup |
Optional: Set Up Ollama¶
If you want AI-powered summaries, character journals, and session naming:
ollama serve
ollama pull llama3.1:8b # For text generation
No Ollama? No Problem
Recording, transcription, speaker identification, and image generation all work without Ollama. You only need an AI assistant for summaries, session naming, and scene description crafting.
Optional: HuggingFace Token for Speaker Identification¶
Automatic speaker identification (figuring out who said what) requires a free HuggingFace token:
- Create an account at huggingface.co
- Accept the pyannote model license at pyannote/speaker-diarization-3.1
- Add your token in TaleKeeper's Settings under Providers → HuggingFace
Without a token, transcription still works but all speech will be attributed to a single speaker.
Next: Run the Setup Wizard →