Audio Copy-Move Forgery Detection

Overview

This project detects audio copy-move forgeries, where a segment of an audio file is copied and pasted elsewhere within the same file to alter its content. It transforms audio into spectrograms, extracts keypoints, identifies high-frequency ranges, generates graph images using advanced graph techniques, and classifies them as forged or genuine using a Convolutional Neural Network (CNN).

Features

• Converts audio to super-resolution spectrograms using Short-Time Fourier Transform (STFT).
• Extracts keypoints with SIFT and identifies high-frequency ranges.
• Applies bandpass filtering and spiral pattern extraction.
• Constructs visibility graphs and converts them to images.
• Uses a fine-tuned MobileNetV2 CNN for forgery classification.

Requirements

• Python 3.8+
• Libraries: tensorflow, librosa, opencv-python, networkx, scipy, numpy, pandas, joblib, community, tqdm, matplotlib, scikit-learn

Usage

• Clone the repository

git clone https://github.com/pavank-v/Audio-Forgery-Detection.git

• Create a new python environment

python3 -m venv env
# Activate the environment
source env/bin/activate

• CD to Backend

cd Backend

• Start the Project

python manage.py runserver

• Preview

File Structure

.
├── Backend
│   ├── api
│   │   ├── admin.py
│   │   ├── apps.py
│   │   ├── __init__.py
│   │   ├── models.py
│   │   ├── templates
│   │   │   └── api
│   │   │       └── index.html
│   │   ├── tests.py
│   │   ├── urls.py
│   │   └── views.py
│   ├── Backend
│   │   ├── asgi.py
│   │   ├── __init__.py
│   │   ├── settings.py
│   │   ├── urls.py
│   │   └── wsgi.py
│   ├── manage.py
│   └── static
├── notebooks
│   ├── ACM_MODEL.ipynb
│   └── audio_forgery_detection_model.joblib
├── README.md
└── requirements.txt

Contributing

Contributions are welcome! Please fork the repository and submit pull requests. For major changes, open an issue to discuss potential improvements.

How to Contribute

1. Fork the repository.
2. Create a new branch (git checkout -b feature-branch).
3. Make your changes and commit them (git commit -m "Improved the model accuracy").
4. Push to the branch (git push origin feature-branch).
5. Open a pull request.

---

Acknowledgments

• Libraries Used: This project leverages several libraries such as TensorFlow, Pandas, Librosa, OpenCV, and more.
• Inspiration: Special thanks to the open-source community for providing valuable resources and documentation.