This repository provides codes to reproduce our paper On the potential of jointly-optimised solutions to spoofing attack detection and automatic speaker verification accepted to the IberSPEECH 2022 conference.
Codes were tested using a GeForce RTX 3090 GPU with CUDA Version==11.2. Please try to install the dependencies first:
conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge
Then:
pip install -r requirements.txt
The ASVspoof 2019 database can be downloaded from here.
Please change the database
defined in each .conf
file from '/path/to/your/LA/'
to its actual path.
To jointly-optimise the ASV and CM sub-systems:
python main.py --config ./config/ResNetSE34V2_AASIST_OC.conf
To train a back-end classifier and keep ASV and CM fixed:
python main.py --config ./config/ResNetSE34V2_AASIST_OC_fixed.conf
We provide pre-trained models in 'pre_trained_models/'
.
You can try to calculate the evaluation score by:
python save_sasv_score.py
If you find this repository useful, please consider citing:
@inproceedings{sasv_joint,
author={Wanying Ge and Hemlata Tak and Massimiliano Todisco and Nicholas Evans},
title={{On the potential of jointly-optimised solutions to spoofing attack detection and automatic speaker verification}},
year=2022,
booktitle={Proc. IberSPEECH 2022 (to appear)},
}
Codes are based on the implementations of SASVC2022_Baseline, voxceleb_trainer, aasist and AIR-ASVspoof.