AI Voice Detection API

This project implements a robust REST API for detecting AI-generated voice samples. It uses an Ensemble of 3 State-of-the-Art Deep Learning Models to achieve high accuracy and provide detailed explanations.

Features

• Ensemble Detection: Combines predictions from 3 different Hugging Face models.
• Dynamic Explanations: Provides context-aware reasons for classification.
• Format Support: Accepts Base64 encoded MP3 files.
• Language Support: Optimized for Tamil, English, Hindi, Malayalam, Telugu.

Prerequisites

• Python 3.8+
• ffmpeg (Required for audio processing)

Installation

1. Install System Dependencies (Linux/Ubuntu):

   sudo apt update && sudo apt install ffmpeg -y

2. Install Python Dependencies:

   pip install -r requirements.txt

Running the API Server

Start the FastAPI server using uvicorn:

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

The server will start at http://localhost:8000.

Testing

Automated Test Script

We have provided a script test_api.py that automatically finds all .mp3 files in the current directory and tests them against the API.

python3 test_api.py

Manual Testing (cURL)

You can verify the API manually using curl:

curl -X POST http://localhost:8000/api/voice-detection \
-H "Content-Type: application/json" \
-H "x-api-key: AI_DETECT" \
-d '{
    "language": "English",
    "audioFormat": "mp3",
    "audioBase64": "<YOUR_BASE64_STRING_HERE>"
}'

API Documentation

Once the server is running, you can access the interactive API docs at:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc