J.A.R.V.I.S AI 🤖

A voice-activated AI assistant inspired by Iron Man's J.A.R.V.I.S., featuring a stunning holographic Arc Reactor interface and real-time audio responses.

🔴 Live Demo: Try JARVIS now →
(Note: Free tier may take ~30s to wake up)

✨ Features

• 🗣️ Voice Input/Output - Talk to JARVIS using your microphone, get audio responses
• ⚛️ Arc Reactor HUD - Procedurally generated 3D interface with audio-reactive animations
• 🧠 Google Gemini AI - Fast, intelligent responses powered by Gemini LLM
• 🎙️ Custom Voice - British accent using the community-trained JARVIS voice model
• 🚀 Real-time Streaming - WebSocket-based sentence-level audio streaming for low latency
• 💜 Stunning UI - Purple/cyan neon theme with glassmorphism effects

🛠 Tech Stack

• Backend: FastAPI (Python) + WebSockets
• AI Brain: Google Gemini API
• Voice (TTS): Piper TTS with custom JARVIS model from Hugging Face
• Voice (STT): Web Speech API (built into Chrome)
• Frontend: Vanilla JavaScript + Three.js
• Deployment: Docker-ready, deployed on Render.com

📦 Installation

Prerequisites

• Python 3.11+
• Git

1. Clone the Repository

git clone https://github.com/yourusername/jarvis-ai.git
cd jarvis-ai

2. Set Up Python Environment

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install -r requirements.txt

3. Configure Environment Variables

Create a .env file in the root directory:

GOOGLE_API_KEY=your_gemini_api_key_here

Get your Gemini API key from: https://ai.google.dev/

4. Run the Application

uvicorn src.app.main:app --reload --port 8000

Visit http://localhost:8000 in your browser (Chrome recommended for best compatibility).

🐳 Docker Deployment

Build and Run with Docker

docker build -t jarvis-ai .
docker run -p 8000:8000 -e GOOGLE_API_KEY=your_key_here jarvis-ai

Deploy to Render.com

1. Push your code to GitHub
2. Create a new Web Service on Render.com
3. Connect your repository
4. Add GOOGLE_API_KEY as an environment variable
5. Deploy!

Render will automatically detect the Dockerfile and build your app.

🎮 Usage

1. Text Input: Type your message in the input field and click "SEND"
2. Voice Input: Click the microphone button (🎤) and speak
3. Voice Output: JARVIS will respond with both text and audio
4. Interactive Avatar: The Arc Reactor pulses and rotates in sync with audio

🔧 Project Structure

jarvis_ai/
├── src/
│   ├── app/
│   │   ├── main.py              # FastAPI server & WebSocket endpoint
│   │   └── services/
│   │       ├── llm_service.py   # Gemini integration
│   │       ├── tts_service.py   # Piper TTS with HuggingFace model
│   │       └── websocket_manager.py  # WebSocket connection manager
│   └── static/
│       └── index.html           # Frontend UI (HTML + Three.js)
├── Dockerfile                   # Docker configuration
├── requirements.txt             # Python dependencies
├── fly.toml                     # Fly.io deployment config
└── README.md

🎨 Customization

Change the Voice Model

Edit src/app/services/tts_service.py:

# Switch from 'medium' to 'high' for better quality (requires more CPU)
self.model_filename = "en/en_GB/jarvis/high/jarvis-high.onnx"
self.config_filename = "en/en_GB/jarvis/high/jarvis-high.onnx.json"

Adjust JARVIS Personality

Edit src/app/services/llm_service.py to modify the system prompt.

Customize UI Colors

Edit the CSS in src/static/index.html to change the color scheme.

⚡ Performance Notes

• Local Development: The high-quality voice model runs instantly on decent hardware (M1 Mac, modern Intel/AMD)
• Render Free Tier: Uses the medium-quality model to balance quality and speed on limited CPU
• Recommended: For production, deploy on a VPS with 1GB+ RAM (DigitalOcean, Fly.io, etc.) for best performance

⚠️ Disclaimer

This project is for educational and personal entertainment purposes only.

The voice model mimics the character J.A.R.V.I.S. (voiced by Paul Bettany) from Marvel Studios. It is a community-trained model and is not licensed for commercial use.

• ❌ Do not use for commercial products
• ❌ Do not use to impersonate real individuals
• ✅ Respect intellectual property rights

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Piper TTS - Fast neural TTS engine
• jgkawell/jarvis - Custom JARVIS voice model
• Google Gemini - Powerful LLM API
• Three.js - 3D graphics library
• Marvel Studios & Paul Bettany - Original JARVIS inspiration

🤝 Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.

📧 Contact

For questions or suggestions, open an issue on GitHub.

---

Built with ❤️ by a fan who just wanted to talk to JARVIS