pipeline_tag: audio-classification
WaveSP-Net: Learnable Wavelet-Domain Sparse Prompt Tuning for Speech Deepfake Detection
This repository contains the WaveSP-Net model, a novel architecture for speech deepfake detection, presented in the paper WaveSP-Net: Learnable Wavelet-Domain Sparse Prompt Tuning for Speech Deepfake Detection.
WaveSP-Net introduces parameter-efficient front-ends that fuse prompt-tuning with classical signal processing transforms and combines a Partial-WSPT-XLSR front-end with a bidirectional Mamba-based back-end. This design effectively enhances the localization of subtle synthetic artifacts without altering frozen XLSR parameters, leading to superior performance on challenging benchmarks with low trainable parameters.
For the official code and more details, please refer to the GitHub repository.
Citation
If you find WaveSP-Net or the associated code helpful for your research, please kindly cite our paper:
@misc{xuan2025wavespnet,
title={WaveSP-Net: Learnable Wavelet-Domain Sparse Prompt Tuning for Speech Deepfake Detection},
author={Xi Xuan and Xuechen Liu and Wenxin Zhang and Yi-Cheng Lin and Xiaojian Lin and Tomi Kinnunen},
year={2025},
eprint={2510.05305},
archivePrefix={arXiv},
primaryClass={eess.AS},
url={https://arxiv.org/abs/2510.05305},
}