AI-powered Anti-Scam Call Defense System. Built with Vue.js, Python Flask, OpenAI, and ElevenLabs for Nexora Hacks 2026.

✨ iPhone-style Call Simulation UI - Realistic call interface with real-time scam detection 🤖 AI-Powered Analysis - Uses OpenAI GPT-4 to detect scam patterns and threats 🔒 PII Redaction - Automatically redacts credit cards, SSN, and other sensitive data 🎯 Threat Detection - Identifies urgency tactics, impersonation, payment requests, and more 🔊 Audio Warnings - Generates spoken alerts using ElevenLabs text-to-speech 🔐 Privacy Toggle - Blur numbers in transcripts for added privacy

VocalGuard/
├── backend/
│   ├── app.py              # Flask API with /analyze endpoint
│   ├── scam_detector.py    # PII redaction and threat detection
│   └── .env.example        # Environment variables template
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   │   └── CallScreen.vue   # iPhone-style call UI
│   │   ├── App.vue
│   │   ├── main.js
│   │   └── style.css
│   ├── index.html
│   ├── package.json
│   ├── vite.config.js
│   ├── tailwind.config.js
│   └── postcss.config.js
└── requirements.txt        # Python dependencies

1. Install Python dependencies:

   pip install -r requirements.txt

2. Configure environment variables:

   cd backend
   cp .env.example .env
   # Edit .env and add your API keys

3. Get API Keys:

   • OpenAI API Key: https://platform.openai.com/api-keys
   • ElevenLabs API Key: https://elevenlabs.io/

4. Run the backend:

   cd backend
   python app.py

   The backend will run on http://localhost:5000

1. Install Node.js dependencies:

   cd frontend
   npm install

2. Run the development server:

   npm run dev

   The frontend will run on http://localhost:3000

3. Build for production:

   npm run build

1. Open the frontend at http://localhost:3000
2. The call screen displays an iPhone-style interface with a demo scam transcript
3. Click "Analyze Call for Scams" to analyze the transcript
4. The system will:
   • Detect PII and redact sensitive information
   • Analyze the transcript using OpenAI for scam patterns
   • Flag threats using keyword detection
   • Display a scam alert with confidence level and detected threats
5. Use the "Blur Numbers" toggle to hide digits in the transcript for privacy

Analyze a call transcript for scam detection.

Request Body:

{
  "transcript": "Call transcript text",
  "generate_audio": true
}

Response:

{
  "is_scam": true,
  "confidence": 0.95,
  "threat_level": "HIGH",
  "detected_threats": ["urgency", "payment", "personal_info"],
  "redacted_transcript": "Text with [CREDIT CARD REDACTED]",
  "detected_pii": ["credit_card", "phone"],
  "warning_message": "⚠️ HIGH RISK SCAM DETECTED!...",
  "audio_url": "/audio/warning.mp3"
}

Health check endpoint.

• Python 3.x - Programming language
• Flask - Web framework
• OpenAI GPT-4 - AI-powered scam detection
• ElevenLabs - Text-to-speech for audio warnings
• Regular Expressions - PII pattern matching

• Vue.js 3 - JavaScript framework
• Tailwind CSS - Utility-first CSS framework
• Vite - Build tool and dev server

• Credit Card Redaction - Detects and redacts card numbers (e.g., 4532-1234-5678-9012)
• SSN Protection - Identifies and redacts social security numbers
• Phone Number Privacy - Redacts phone numbers from transcripts
• Email Masking - Removes email addresses
• Privacy Toggle - Allows users to blur all numbers in the UI

• Urgency - Pressure tactics and time-sensitive demands
• Payment - Requests for wire transfers, gift cards, cryptocurrency
• Personal Info - Requests for SSN, passwords, account numbers
• Impersonation - Claims to be IRS, police, tech support, etc.
• Threats - Mentions of arrest, lawsuits, frozen accounts
• Too Good to Be True - Lottery wins, prizes, inheritance claims

# Backend tests
cd backend
pytest

# Frontend tests
cd frontend
npm test

• Backend follows PEP 8 Python style guide
• Frontend uses ESLint with Vue.js recommended rules

MIT License - Built for Nexora Hacks 2026

Pull requests are welcome! For major changes, please open an issue first to discuss proposed changes.

• OpenAI for GPT-4 API
• ElevenLabs for text-to-speech technology
• Vue.js and Tailwind CSS communities