This paper describes our proposed system for the spoofingaware speaker verification challenge (SASV Challenge 2022). The system follows an integrated approach that uses speaker verification and antispoofing embeddings extracted from specialized neural networks. Firstly, a shallow neural network, fed with the test utterance’s verification and spoofing embeddings, is used to compute a spoof-based score. The final scoring decision is then obtained by combining this score with the cosine similarity between speaker verification embeddings. The integration network was trained using a one-class loss to discriminate between target and unauthorized trials. Our proposed system is evaluated over the ASVspoof19 database and shows competitive performance compared to other integration approaches. In addition, we compare our approach with further state-of-theart speaker verification and antispoofing systems based on selfsupervised learning, yielding high-performance speech biometric systems comparable with the best challenge submissions.
Cite as: Martín-Doñas, J.M., González Torre, I., Álvarez, A., Arellano, J. (2022) The Vicomtech Spoofing-Aware Biometric System for the SASV Challenge . Proc. IberSPEECH 2022, 66-70, doi: 10.21437/IberSPEECH.2022-14
@inproceedings{martindonas22_iberspeech,
author={Juan Manuel Martín-Doñas and Iván {González Torre} and Aitor Álvarez and Joaquin Arellano},
title={{The Vicomtech Spoofing-Aware Biometric System for the SASV Challenge }},
year=2022,
booktitle={Proc. IberSPEECH 2022},
pages={66--70},
doi={10.21437/IberSPEECH.2022-14}
}