Perceptual Based Adversarial Audio Attacks
This addresses a security vulnerability in ASR systems for real-world applications, representing a novel extension beyond digital-only attacks.
The authors tackled the problem of creating physically realizable adversarial attacks on automatic speech recognition systems by developing a method based on psychoacoustic properties and automated room impulse responses, resulting in attacks that are robust across multiple environments and virtually imperceptible to listeners.
Recent work has shown the possibility of adversarial attacks on automatic speechrecognition (ASR) systems. However, in the vast majority of work in this area, theattacks have been executed only in the digital space, or have involved short phrasesand static room settings. In this paper, we demonstrate a physically realizableaudio adversarial attack. We base our approach specifically on a psychoacoustic-property-based loss function, and automated generation of room impulse responses, to create adversarial attacks that are robust when played over a speaker in multiple environments. We show that such attacks are possible even while being virtually imperceptible to listeners.