https://github.com/vintellx/verivoice
Implementation of a Graph-Enhanced Audio Transformer for detecting deepfake speech.
https://github.com/vintellx/verivoice
deepfake deepfake-audio deepfake-audio-detection deepfake-detection deepfakes deepfakes-classification
Last synced: 5 months ago
JSON representation
Implementation of a Graph-Enhanced Audio Transformer for detecting deepfake speech.
- Host: GitHub
- URL: https://github.com/vintellx/verivoice
- Owner: VintellX
- License: apache-2.0
- Created: 2025-05-10T10:51:58.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-05-11T03:52:26.000Z (about 1 year ago)
- Last Synced: 2025-06-26T05:35:38.552Z (12 months ago)
- Topics: deepfake, deepfake-audio, deepfake-audio-detection, deepfake-detection, deepfakes, deepfakes-classification
- Language: Jupyter Notebook
- Homepage:
- Size: 4.61 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
```bash
git clone --recursive https://github.com/VintellX/VeriVoice.git
cd VeriVoice
```
## Overview
VeriVoice is a DeepFake Audio Detection model, this is a PyTorch-based framework for detecting cloned or syntthesized speech. It uses a Graph-Enhanced Audio Transformer to capture both local spectral artifacts and global temporal patterns that distinguish real voices from their deepfakes or clones.
## Quick Start
1. **Install dependencies**:
```bash
conda create -n VeriVoice python=3.10
conda activate VeriVoice
pip install -r requirements.txt
```
2. **Prepare data** (check out [`usage.md`](USAGE.md) for details).
3. **Train the mdoel**:
```bash
python -m ipykernel install --user --name=VeriVoice --display-name "VeriVoice-Env"
jupyter notebook
```
4. **Test an audio file**:
```bash
python3 -m predicto path/to/audio.wav
```
## License
This project is licensed under the **Apache License 2.0**. See [LICENSE](LICENSE) for details.