An Effective Energy Mask-based Adversarial Evasion Attacks against Misclassification in Speaker Recognition Systems

TL;DR

This paper introduces a novel energy masking-based adversarial attack method called MEP that effectively fools speaker recognition systems while maintaining high audio quality and minimal perceptual distortion.

Contribution

The study presents MEP, a new energy masking technique in the frequency domain that enhances adversarial attack effectiveness against speaker recognition models.

Findings

MEP outperforms FGSM in evasion success.

Audio quality remains high with minimal perceptual distortion.

Effective against advanced speaker recognition models.

Abstract

Evasion attacks pose significant threats to AI systems, exploiting vulnerabilities in machine learning models to bypass detection mechanisms. The widespread use of voice data, including deepfakes, in promising future industries is currently hindered by insufficient legal frameworks. Adversarial attack methods have emerged as the most effective countermeasure against the indiscriminate use of such data. This research introduces masked energy perturbation (MEP), a novel approach using power spectrum for energy masking of original voice data. MEP applies masking to small energy regions in the frequency domain before generating adversarial perturbations, targeting areas less noticeable to the human auditory model. The study primarily employs advanced speaker recognition models, including ECAPA-TDNN and ResNet34, which have shown remarkable performance in speaker verification tasks. The proposed MEP method demonstrated strong performance in both audio quality and evasion effectiveness. The energy masking approach effectively minimizes the perceptual evaluation of speech quality (PESQ) degradation, indicating that minimal perceptual distortion occurs to the human listener despite the adversarial perturbations. Specifically, in the PESQ evaluation, the relative performance of the MEP method was 26.68% when compared to the fast gradient sign method (FGSM) and iterative FGSM.