/fake-audio-detector

Simple fake audio detector

Primary LanguagePythonGNU Affero General Public License v3.0AGPL-3.0

fake-audio-detector

Simple fake audio detector made for CVPR 2022 - Workshop on Media Forensics.

⚠️ TODO: I have to update the models, the current ones could not work properly ⚠️, but the based logic and the feature extraction is the one described in the paper.

Requirements

Install Python3.9.12+ and the related dependencies using:

pip3 install -r requirements.txt

Usage

Run the main.py file giving as input the path of the audio that you want to check through the trained model ADC.

Example:

python3 main.py "data/sample.wav"

The result will be The sample audio is fake or The sample audio is real. Feel free to re-use the feature extraction script, if you have any issues or questions about this script open an issue.

Reference

Is synthetic voice detection research going into the right direction?

If this paper helped you in the research, please cite:

@inproceedings{borzi2022synthetic,
  title={Is Synthetic Voice Detection Research Going Into the Right Direction?},
  author={Borz{\`\i}, Stefano and Giudice, Oliver and Stanco, Filippo and Allegra, Dario},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={71--80},
  year={2022}
}

Credits

  • Stefano Borzì
  • Oliver Giudice
  • Dario Allegra
  • Filippo Stanco