Transformers and their variants have achieved great
success in speech processing. However, their multi-head selfattention mechanism is computationally expensive. Therefore, one
novel selective state space model, Mamba, has been proposed
as an alternative. Building on its success in automatic speech
recognition, we apply Mamba for spoofing attack detection.
Mamba is well-suited for this task as it can capture the artifacts
in spoofed speech signals by handling long-length sequences.
However, Mamba’s performance may suffer when it is trained
with limited labeled data. To mitigate this, we propose combining
a new structure of Mamba based on a dual-column architecture
with self-supervised learning, using the pre-trained wav2vec
2.0 model. The experiments show that our proposed approach
achieves competitive results and faster inference on the ASVspoof
2021 LA and DF datasets, and on the more challenging In-theWild dataset, it emerges as the strongest candidate for spoofing
attack detection.
2. Setup Environment
You need to create the running environment by Anaconda.
First, create and activate the environment:
The pretrained model XLSR can be found at this link.
4. Results
Dataset
EER (%)
min t-DCF
ASVspoof2021 LA
0.93
0.208
ASVspoof2021 DF
1.88
-
In-The-Wild
6.71
-
5. Citation
If you find our repository valuable for your work, please consider giving a star to this repo and citing our paper:
@article{xiao2024xlsr,
title={{XLSR-Mamba}: A Dual-Column Bidirectional State Space Model for Spoofing Attack Detection},
author={Xiao, Yang and Das, Rohan Kumar},
journal={arXiv preprint arXiv:2411.10027},
year={2024}
}
This model has been pushed to the Hub using the PytorchModelHubMixin integration:
