AI-TruthScan: Detect AI-Generated Interview Responses

AI-TruthScan is a web application that allows users to analyze audio and video recordings to determine whether the responses are likely generated by AI or created by humans. The application uses Google Gemini for AI analysis and integrates speech recognition for transcription.

The primary objective is to assist recruiters and HR professionals in verifying the authenticity of candidates' responses, ensuring fair and reliable hiring decisions.

Table of Contents

• Features
• Technologies Used
• Installation
• Usage
• API Endpoints
• Database
• Tests
• Future Improvements
• License

Features

• Record audio and video or upload existing files for analysis.
• Transcribe audio using speech recognition.
• Analyze transcriptions using AI models to classify them as "Real (Human-Created)" or "Fake (AI-Generated)".
• Gaze tracking to assess whether the speaker is reading from a script.
• Provides real-time analysis and confidence scores.
• Store and retrieve analysis results from a SQLite database.
• User-friendly UI for seamless interaction.
• Detailed insights on response authenticity.
• Can be integrated with existing HR tools.

Technologies Used

• Frontend: Streamlit
• Backend: Flask
• Database: SQLite
• AI Model: Google Gemini
• Audio Processing: SpeechRecognition, SoundDevice, Wavio
• Video Processing: OpenCV, FFmpeg
• Gaze Tracking: MediaPipe

Installation

Prerequisites

• Python 3.7 or higher
• pip (Python package installer)

Clone the Repository

git clone https://github.com/yourusername/AI-TruthScan.git
cd AI-TruthScan

Install Dependencies

You need to install the required packages for both the frontend and backend.

1. Frontend (Streamlit) Navigate to the frontend directory and install the required packages:

cd frontend
pip install -r requirements.txt

2. Backend (Flask) Navigate to the backend directory and install the required packages:

cd ../backend
pip install -r requirements.txt

Set Up Google API Key Replace the placeholder 'api_key' in 'backend.py' with your actual Google API key.

Usage

Running the Backend

1. Navigate to the backend directory:

cd backend

2. Run the Flask application:

python backend.py

Running the Frontend

1. Open a new terminal window.
2. Navigate to the frontend directory:

cd AI-TruthScan/frontend

3. Run the Streamlit application:

streamlit run app.py

Accessing the Application

Open your web browser and go to 'http://localhost:8501' to access the Streamlit application.

API Endpoints

Analyze File

• Endpoint: '/analyze'
• Method: POST
• Description: Analyzes an uploaded audio or video file.
• Parameters:
  • 'file': The audio or video file to analyze.
  • 'file_type': Type of the file (audio or video).
  • 'context': Context of the speech (default: "Formal Interview").
  • 'threshold': AI-generated probability threshold (default: 50).

Get Results

• Endpoint: '/results'
• Method: GET
• Description: Retrieves saved analysis results.

Database

The application uses SQLite to store analysis results. The database file is named 'analysis_results.db' and will be created automatically when the application runs.

Project Contributors

• Anusha G E - Lead Developer
• Kashif - UI Designer
• Keerthana H - PPT Designer and Feedback
• Prajwal - UI and Report
• Ayush Kumar - Tester
• Reetish Kulkarni - Documenter

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any improvements or bug fixes.

Tests

The system has been tested using multiple AI-generated and human responses to ensure accuracy.

https://drive.google.com/file/d/1eUJZOWN5XeO3HeA1ej7Afe2amaSt7ijZ/view?usp=drive_link

Future Improvements

• Enhancing model accuracy for better AI response detection.

• Expanding dataset coverage to include more response variations.

• Integrating with HR software for seamless recruitment workflows.

License

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

### Notes:
- Replace `yourusername` in the clone URL with your actual GitHub username.
- Ensure that you have a `requirements.txt` file in both the `frontend` and `backend` directories that lists all the necessary Python packages.
- Adjust any sections as needed to fit your specific project structure or additional features