Adjust-free adversarial example generation in speech recognition using evolutionary multi-objective optimization under black-box condition
This work addresses the problem of generating robust adversarial examples for speech recognition systems, specifically for attackers who need to play adversarial audio without precise timing.
This paper proposes a black-box adversarial attack method for automatic speech recognition systems that generates adjust-free adversarial examples. The method uses Evolutionary Multi-objective Optimization (EMO) to create examples robust against timing lag, eliminating the need for precise timing by an attacker.
This paper proposes a black-box adversarial attack method to automatic speech recognition systems. Some studies have attempted to attack neural networks for speech recognition; however, these methods did not consider the robustness of generated adversarial examples against timing lag with a target speech. The proposed method in this paper adopts Evolutionary Multi-objective Optimization (EMO)that allows it generating robust adversarial examples under black-box scenario. Experimental results showed that the proposed method successfully generated adjust-free adversarial examples, which are sufficiently robust against timing lag so that an attacker does not need to take the timing of playing it against the target speech.