A real-time voice emergency detection system that uses advanced voice processing to identify emergency situations and provide appropriate responses.

• Real-time voice recording and processing
• Emergency classification based on keywords (Fire, Medical, Violence, Accident)
• Text-to-speech response generation using ElevenLabs
• Persistent event storage with MongoDB
• Real-time event feed via WebSocket
• Responsive web interface built with React

• Python 3.8+
• Node.js 14+
• MongoDB Atlas account
• ElevenLabs API key

1. Navigate to the backend directory:

   cd backend

2. Install Python dependencies:

   pip install -r requirements.txt

3. Configure environment variables in .env:

   MONGO_URL="your_mongodb_connection_string"
   DB_NAME="voice_assistant_db"
   CORS_ORIGINS="*"
   

4. Start the backend server:

   uvicorn server:app --host 0.0.0.0 --port 8000 --reload

1. Navigate to the frontend directory:

   cd frontend

2. Install Node dependencies:

   yarn install

3. Configure environment variables in .env:

   REACT_APP_BACKEND_URL=http://localhost:8000
   

4. Start the development server:

   yarn start

This application is configured for deployment to:

• Frontend: Vercel (free tier)
• Backend: Render (free tier)

📖 Complete deployment guide: See DEPLOYMENT.md for detailed step-by-step instructions.

Copy .env.example files and update with your credentials:

Backend (backend/.env):

MONGO_URL=your_mongodb_connection_string
DB_NAME=voice_assistant_db
ELEVENLABS_API_KEY=your_elevenlabs_api_key
GEMINI_API_KEY=your_gemini_api_key
CORS_ORIGINS=*

Frontend (frontend/.env):

REACT_APP_BACKEND_URL=http://localhost:8000

Run the deployment script:

./deploy.sh

This will:

• Install all dependencies
• Build production versions of both frontend and backend
• Start both services in the background
• Log output to the logs/ directory

1. Open your browser and navigate to http://localhost:3000
2. Click the microphone button to start recording
3. Speak an emergency phrase (e.g., "There's a fire in the building")
4. View the classified emergency and AI response
5. Monitor real-time events in the event feed

• POST /api/voice - Process voice recording
  • Form data: audio (webm audio file)
  • Returns: Emergency event object

• GET /api/events - Get recent emergency events
  • Query param: limit (default: 50)
  • Returns: List of events

• GET /api/status - Get system status checks
• POST /api/status - Create new status check

• WS /ws - Real-time event streaming

┌─────────────┐    HTTP    ┌──────────────┐
│   React     │ ──────────▶│   FastAPI    │
│  Frontend   │            │   Backend    │
└─────────────┘            └──────────────┘
                                │
                           ┌────▼────┐
                           │ MongoDB │
                           │   Atlas │
                           └─────────┘

• ElevenLabs STT: Speech-to-text conversion with fallback to mock STT
• Keyword Classifier: Classifies emergencies based on transcript keywords
• ElevenLabs TTS: Text-to-speech response generation
• MongoDB Event Store: Persistent storage of emergency events
• WebSocket Manager: Real-time event broadcasting

backend/
├── routes/          # API route handlers
├── services/        # Business logic and external services
├── websocket/       # WebSocket connection management
├── server.py        # Main application entry point
└── requirements.txt # Python dependencies

frontend/
├── src/
│   ├── components/  # React UI components
│   ├── pages/       # Page components
│   ├── services/    # API service clients
│   └── App.js       # Main application component
└── package.json     # Node dependencies

This project is proprietary and confidential.