Cyclic Defense GAN Against Speech Adversarial Attacks

TL;DR

This paper introduces a cyclic GAN-based defense mechanism that reconstructs speech signals to counteract adversarial attacks on speech-to-text systems, showing effectiveness against various attack algorithms.

Contribution

It presents a novel cyclic GAN framework for implicit reactive defense in speech recognition, avoiding direct manipulation of malicious inputs.

Findings

Effective against white-box and black-box attacks

Improves robustness of speech-to-text models

Validated on DeepSpeech, Kaldi, and Lingvo

Abstract

This paper proposes a new defense approach for counteracting state-of-the-art white and black-box adversarial attack algorithms. Our approach fits into the implicit reactive defense algorithm category since it does not directly manipulate the potentially malicious input signals. Instead, it reconstructs a similar signal with a synthesized spectrogram using a cyclic generative adversarial network. This cyclic framework helps to yield a stable generative model. Finally, we feed the reconstructed signal into the speech-to-text model for transcription. The conducted experiments on targeted and non-targeted adversarial attacks developed for attacking DeepSpeech, Kaldi, and Lingvo models demonstrate the proposed defense's effectiveness in adverse scenarios.