# MixFake
This is the official repository for the paper "MixFake: Benchmarking and Enhancing Audio Deepfake Detection in Diverse Real-world Mixed Audio".

2026/06/03 News: Our Code has been released.

2026/05/25 News✨: Our paper has been accepted as a **Spotlight** paper at ICME 2026.

2026/05/17 News: MixFake dataset is available at : [HuggingFace](https://huggingface.co/datasets/Tnxts/MixFake/tree/main)

# Usage
Run the scripts in `sh/` to train or evaluate a model, for example:

```bash
sh sh/xxx.sh