Crafting Adversarial Examples For Speech Paralinguistics Applications

TL;DR

This paper introduces a method to generate adversarial audio examples that significantly degrade the performance of speech paralinguistic neural networks with minimal audio quality loss.

Contribution

It presents an end-to-end approach to create adversarial examples by perturbing raw audio waveforms, highlighting vulnerabilities in speech analysis systems.

Findings

Adversarial perturbations cause substantial performance drops in neural networks.

Minimal audio quality degradation is achieved despite effective attacks.

The method demonstrates potential security risks in speech-based applications.

Abstract

Computational paralinguistic analysis is increasingly being used in a wide range of cyber applications, including security-sensitive applications such as speaker verification, deceptive speech detection, and medical diagnostics. While state-of-the-art machine learning techniques, such as deep neural networks, can provide robust and accurate speech analysis, they are susceptible to adversarial attacks. In this work, we propose an end-to-end scheme to generate adversarial examples for computational paralinguistic applications by perturbing directly the raw waveform of an audio recording rather than specific acoustic features. Our experiments show that the proposed adversarial perturbation can lead to a significant performance drop of state-of-the-art deep neural networks, while only minimally impairing the audio quality.