This is the Bumblebee project — a system for communicating through song snippets. It has two main parts: a library bootstrapper and a lyric search engine.
This project includes a Pi skill at .pi/skills/bumblebee-voice/ that lets you
(the Pi agent) speak through song snippets after completing tasks.
To give a Bumblebee voice summary:
# After completing a task, find a voice line
.pi/skills/bumblebee-voice/scripts/voice.py "task description"
# Export it as MP3
.pi/skills/bumblebee-voice/scripts/voice-line.py export --song "Song Name" --start "01:23.45" --duration 5Present voice lines in this format:
*activates vocal synthesizer...*
🐝 Bumblebee says: "{voice line}"
— {song}, {artist} ({timestamp})
*deactivates vocal synthesizer*
Ask the user first: "Want a Bumblebee voice summary?"
bumblebee/
├── bumblebee-bootstrap/ # Library bootstrap CLI (Python + uv)
│ ├── bootstrap/ # Python package
│ └── pyproject.toml
├── bumblebee/ # Bumblebee Engine (Python + uv)
│ ├── bumblebee/ # Python package (search, audio, API)
│ ├── web/ # React web UI (Vite + Tailwind)
│ └── pyproject.toml
├── .pi/skills/bumblebee-voice/ # ← THIS SKILL (Pi voice integration)
└── AGENTS.md # This file
- Backend: Python 3.10+, uv, pydub, SQLite FTS5, FastAPI
- Frontend: React 19, Vite, Tailwind CSS, wavesurfer.js
- Package manager: uv (Python), npm (web)
- License: AGPL-3.0
# Test bootstrap
cd bumblebee-bootstrap && uv run bootstrap/__init__.py --help
# Test engine
cd bumblebee && uv run -m bumblebee --help
# Test web UI
cd bumblebee/web && npm run build
# Test voice skill
.pi/skills/bumblebee-voice/scripts/voice.py "completed successfully"
.pi/skills/bumblebee-voice/scripts/voice.py --listWhen appropriate (ask user first), after completing tasks, search for a lyric that matches the context and present it as a Bumblebee voice line. See the bumblebee-voice skill for full instructions and the curated voice line database.
"When words aren't enough... speak through song."