---
license: mit
tags:
  - audio
  - deepfake-detection
  - anti-spoofing
  - wav2vec2
  - xlsr
  - speech
  - asvspoof
datasets:
  - asvspoof2019
  - asvspoof2021
metrics:
  - equal_error_rate
pipeline_tag: audio-classification
language:
  - en
library_name: pytorch
---

# XLS-R + SLS Classifier for Audio Deepfake Detection

Reproduction of **"Audio Deepfake Detection with XLS-R and SLS Classifier"** (Zhang et al., ACM Multimedia 2024).

The Selective Layer Summarization (SLS) classifier extracts attention-weighted features from all 24 transformer layers of [XLS-R 300M](https://huggingface.co/facebook/wav2vec2-xls-r-300m) (wav2vec 2.0), then classifies bonafide vs. spoofed speech via a lightweight fully-connected head. [RawBoost](https://arxiv.org/abs/2301.00693) (algo=3, SSI) data augmentation is applied during training.

## Available Checkpoints

| File | Experiment | Description |
|------|-----------|-------------|
| `v1/epoch_2.pth` | v1 (baseline) | Best cross-domain generalization. Patience=1, no validation, 4 epochs. |
| `v2/epoch_16.pth` | v2 (val-based) | Validation early stopping. Patience=10, ASVspoof2019 LA dev validation, 27 epochs. |

**Recommended**: Use `v1/epoch_2.pth` — it generalizes better to unseen attack types (DF, In-the-Wild).

### Original authors' pretrained models

The original pretrained checkpoints from Zhang et al. are available from:
- [Google Drive](https://drive.google.com/drive/folders/13vw_AX1jHdYndRu1edlgpdNJpCX8OnrH?usp=sharing)
- [Baidu Pan](https://pan.baidu.com/s/1dj-hjvf3fFPIYdtHWqtCmg?pwd=shan) (password: shan)

## Results

| Track | Paper EER (%) | v1 EER (%) | v2 EER (%) |
|-------|--------------|------------|------------|
| ASVspoof 2021 DF | 1.92 | **2.14** | 3.75 |
| ASVspoof 2021 LA | 2.87 | 3.51 | **3.47** |
| In-the-Wild | 7.46 | **7.84** | 12.67 |

v1 closely reproduces the paper results. v2 improves LA slightly but degrades DF and In-the-Wild due to overfitting to the LA validation domain — a well-documented cross-domain generalization problem in audio deepfake detection ([Muller et al., Interspeech 2022](https://arxiv.org/abs/2203.16263)).

## Training Configuration

Both experiments share the following setup:

| Parameter | Value |
|-----------|-------|
| Training data | ASVspoof2019 LA train (25,380 utterances) |
| Loss | Weighted Cross-Entropy [0.1, 0.9] |
| Optimizer | Adam (lr=1e-6, weight_decay=1e-4) |
| Batch size | 5 |
| RawBoost | algo=3 (SSI) |
| Seed | 1234 |
| SSL backbone | XLS-R 300M (frozen feature extractor) |
| GPU | NVIDIA RTX 4080 (16 GB) |

### v1 specifics
- Early stopping: patience=1 on training loss
- No validation set
- 4 epochs trained, best at epoch 2 (train loss = 0.000661)

### v2 specifics
- Early stopping: patience=10 on validation loss
- Validation: ASVspoof2019 LA dev (24,844 trials)
- 27 epochs trained, best at epoch 16 (val_loss = 0.000468, val_acc = 99.99%)
- Bug fixes: `torch.no_grad()` in validation loop, correct `best_val_loss` tracking

## Usage

### Download checkpoint

```python
from huggingface_hub import hf_hub_download

# Download v1 checkpoint (recommended)
checkpoint_path = hf_hub_download(
    repo_id="sukhdeveyash/XLS-R-SLS-Deepfake-Detection",
    filename="v1/epoch_2.pth"
)

# Download v2 checkpoint
# checkpoint_path = hf_hub_download(
#     repo_id="sukhdeveyash/XLS-R-SLS-Deepfake-Detection",
#     filename="v2/epoch_16.pth"
# )
```

### Load and run inference

```python
import torch
from model import Model  # from the GitHub repo

device = "cuda" if torch.cuda.is_available() else "cpu"

model = Model(device=device, ssl_cpkt_path="xlsr2_300m.pt")
model.load_state_dict(torch.load(checkpoint_path, map_location=device))
model = model.to(device)
model.eval()
```

Full training and evaluation code: [GitHub Repository](https://github.com/Yash-Sukhdeve/XLS-R-SLS-Deepfake-Detection)

## Requirements

- Python 3.7+
- PyTorch 1.13.1 (CUDA 11.7)
- fairseq (commit a54021305d6b3c)
- XLS-R 300M base checkpoint (`xlsr2_300m.pt`) from [fairseq](https://github.com/pytorch/fairseq/tree/main/examples/wav2vec/xlsr)

See `environment.yml` in the [GitHub repo](https://github.com/Yash-Sukhdeve/XLS-R-SLS-Deepfake-Detection) for the full environment.

## Citation

```bibtex
@inproceedings{zhang2024audio,
  title={Audio Deepfake Detection with XLS-R and SLS Classifier},
  author={Zhang, Qishan and Wen, Shuangbing and Hu, Tao},
  booktitle={Proceedings of the 32nd ACM International Conference on Multimedia},
  year={2024},
  publisher={ACM}
}
```

## Acknowledgements

- [XLS-R](https://github.com/pytorch/fairseq/tree/main/examples/wav2vec/xlsr) (Babu et al., 2022)
- [RawBoost](https://arxiv.org/abs/2301.00693) (Tak et al., Odyssey 2022)
- [ASVspoof Challenge](https://www.asvspoof.org/)