Multi-task Learning-Based Spoofing-Robust Automatic Speaker Verification System

Abstract Spoofing attacks posed by generating artificial speech can severely degrade the performance of a speaker verification system. Recently, many anti-spoofing countermeasures have been proposed for detecting varying types from synthetic to replay presentations. While there are numerous effective defenses reported on standalone solutions, integration and spoofing detection systems has obvious benefits. In this paper, we propose spoofing-robust automatic system diverse based multi-task learning architecture. This deep learning-based model is jointly trained with time-frequency representations utterances provide recognition decisions both tasks simultaneously. Compared other state-of-the-art ASVspoof 2017 2019 corpora, substantial improvement combined under different conditions be obtained.