Jarvis listens for a wake word, transcribes speech in real time on the GPU, understands what you want via Claude, and talks back — while controlling apps, windows, Spotify, timers, and system settings hands-free, in English or Spanish.
- Voice control — wake word detection, real-time GPU transcription (Whisper), natural language understanding (Claude), text-to-speech replies, and interrupt handling mid-speech
- App & window management — open/close apps by voice, move windows across monitors, auto-discovers your shortcuts
- Spotify integration — play songs, playlists, or artists by voice search; full playback control
- System control — shutdown/restart/lock, volume, brightness, screenshots, clipboard, run shell commands
- Timers, reminders & memory — set/cancel timers and reminders, remember and recall facts across sessions
- Advanced controls — screen reading via OCR (Claude vision), mouse/keyboard automation, gaming mode (auto-closes distracting apps)
git clone <repo-url>
cd Jarvis
pip install -r requirements.txt
python main.pyRequires Python 3.10+, a microphone, and API keys in .env (Anthropic, optionally Spotify) — full
setup guide in SETUP.md. Voice command examples, configuration, architecture,
performance notes and troubleshooting: docs/USAGE.md.
Python, Claude (Anthropic API) for language understanding and screen OCR, Whisper (faster-whisper, GPU-accelerated) for speech-to-text, Edge TTS, Spotify API (spotipy), OpenWakeword
See LICENSE.