Audio-replay attack detection countermeasures

TL;DR

This paper compares various spoofing detection methods for speaker verification, highlighting the effectiveness of deep learning approaches under changing acoustic conditions and the importance of feature-based classifiers.

Contribution

It introduces and evaluates multiple spoofing detection approaches, emphasizing the superior stability of deep learning methods and the role of high-level features with SVM classifiers.

Findings

Deep learning approaches show stable efficiency across acoustic conditions.

SVM classifiers with high-level features significantly improve detection performance.

Fusion systems benefit from combining different spoofing detection methods.

Abstract

This paper presents the Speech Technology Center (STC) replay attack detection systems proposed for Automatic Speaker Verification Spoofing and Countermeasures Challenge 2017. In this study we focused on comparison of different spoofing detection approaches. These were GMM based methods, high level features extraction with simple classifier and deep learning frameworks. Experiments performed on the development and evaluation parts of the challenge dataset demonstrated stable efficiency of deep learning approaches in case of changing acoustic conditions. At the same time SVM classifier with high level features provided a substantial input in the efficiency of the resulting STC systems according to the fusion systems results.