Using Deep Learning for Detecting Spoofing Attacks on Speech Signals

TL;DR

This paper presents deep learning-based systems for detecting speech spoofing attacks, achieving high accuracy on a standard benchmark by using neural networks as classifiers and feature extractors.

Contribution

It introduces deep neural network models for spoofing detection that outperform previous methods on the ASVSpoof2015 dataset.

Findings

Achieved less than 0.5% EER on known attacks

Demonstrated effectiveness of neural networks as classifiers and feature extractors

Validated approach on standard spoofing database

Abstract

It is well known that speaker verification systems are subject to spoofing attacks. The Automatic Speaker Verification Spoofing and Countermeasures Challenge -- ASVSpoof2015 -- provides a standard spoofing database, containing attacks based on synthetic speech, along with a protocol for experiments. This paper describes CPqD's systems submitted to the ASVSpoof2015 Challenge, based on deep neural networks, working both as a classifier and as a feature extraction module for a GMM and a SVM classifier. Results show the validity of this approach, achieving less than 0.5\% EER for known attacks.