KAVACH - Scam Detection Application

Overview

This is a Flask-based web application designed to detect potential scam calls in real-time using speech recognition with the Vosk model and phone number lookup via a Telegram bot. The app monitors audio input for scam-related keywords in Hindi and alerts users if a threshold is exceeded. It also allows users to look up phone numbers for additional details.

Project Structure

scam-detection-app/
├── app.py
├── vosk-model-small-hi-0.22/
│   └── [Vosk model files]
├── static/
│   └── style.css
├── templates/
│   └── index.html
└── README.md

Prerequisites

• Python 3.8+
• Vosk model (vosk-model-small-hi-0.22) downloaded and placed in the project root
• Telegram API credentials (API_ID, API_HASH, PHONE_NUMBER)
• Node.js (for SocketIO client, if needed)
• Git for version control

Setup Instructions

1. Clone the Repository

   git clone https://github.com/shreyash729/Kavach---ReTime-Phone-Scam-Detection-System.git
   cd Kavach---ReTime-Phone-Scam-Detection-System

2. Install Dependencies Install the required Python packages:

   pip install flask flask-socketio eventlet vosk pyaudio numpy telethon

3. creates a virtual environment:

   python -m venv venv

4. Configure Telegram environment Variables:

   • Obtain your Telegram API credentials from my.telegram.org.

   set TELEGRAM_API_ID=YOUR_TELEGRAM_API_ID   # your telegram api ID

   set TELEGRAM_API_HASH=YOUR_TELEGRAM_API_HASH   # your telegram api hash

   set PHONE_NUMBER=PHONE_NUMBER   # Your phone number Include country code (e.g., +91xxxxxx)

5. Run the Application

   python app.py

   The app will start on http://localhost:5000.

Usage

• Access the Web Interface: Open http://localhost:5000 in a browser.
• Phone Number Lookup: Enter a phone number and click "Lookup" to fetch details via the Telegram bot.
• Call Monitoring: Click "Accept Call" to start real-time audio monitoring. The app will detect scam keywords and alert if the threshold (2 keywords) is reached.
• Reject Call: Stops the audio monitoring.

Features

• Real-time scam detection using Vosk speech recognition for Hindi keywords.
• Phone number lookup via Telegram bot integration.
• Responsive UI with SocketIO for real-time alerts.
• Customizable scam keyword list and detection threshold.

Dependencies

• Flask: Web framework
• Flask-SocketIO: Real-time communication
• Eventlet: Asynchronous I/O
• Vosk: Speech recognition
• PyAudio: Audio input
• NumPy: Audio data processing
• Telethon: Telegram API client

Notes

• Ensure a working microphone for audio input.
• The Vosk model path must be correctly set in app.py.
• Telegram bot (@TrueCaller_Z_Bot) must be accessible and responsive.
• Replace vosk-model-small-hi-0.22 with vosk-model-hi-0.22 for better Accuracy
• Download Vosk NLP model from https://alphacephei.com/vosk/models

Contributing

1. Fork the repository.
2. Create a feature branch (git checkout -b feature-branch).
3. Commit changes (git commit -m "Add feature").
4. Push to the branch (git push origin feature-branch).
5. Create a pull request.

License

This project is licensed under the MIT License.