EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Mel frequency spectral domain defenses against adversarial attacks on speech recognition systems

Nicholas Mehlman, Anirudh Sreeram, Raghuveer Peri, Shrikanth Narayanan

JASA Express Letters · 2023

DOI: 10.1121/10.0017680 ↗

Kurz gefasst

Fachartikel in der Zeitschrift JASA Express Letters, 2023. Verfasst von Nicholas Mehlman und 3 weiteren Autor:innen. Beteiligte Einrichtung: Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, California 90089, USA nmehlman@usc.edu,…. Hauptthemen laut MEDLINE: Speech, Speech Perception. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 18 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Annenberg Fellowship.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automatic speech recognition (ASR) systems are vulnerable to adversarial attacks due to their reliance on machine learning models. Many of the defenses explored for defending ASR systems simply adapt defense approaches developed for the image domain. This paper explores speech-specific defenses in the feature domain and introduces a defense method called mel domain noise flooding (MDNF). MDNF injects additive noise to the mel spectrogram speech representation prior to re-synthesizing the audio signal input to ASR. The defense is evaluated against strong white-box threat models and shows competitive robustness.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

SpeechSpeech PerceptionSpeech Recognition SoftwareNoiseMachine Learning

Autor:innen und Institutionen

  • Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, California 90089, USA nmehlman@usc.edu, asreeram@usc.edu, rperi@usc.edu, shri@usc.edu.

Förderung

  • Annenberg Fellowship

Zitieren

APA

Mehlman, N., Sreeram, A., Peri, R., & Narayanan, S. (2023). Mel frequency spectral domain defenses against adversarial attacks on speech recognition systems. JASA Express Letters. https://doi.org/10.1121/10.0017680

Vancouver

Mehlman N, Sreeram A, Peri R, Narayanan S. Mel frequency spectral domain defenses against adversarial attacks on speech recognition systems. JASA Express Letters. 2023. doi:10.1121/10.0017680

BibTeX
@article{Mehlman2023,
  author = {Mehlman, Nicholas and Sreeram, Anirudh and Peri, Raghuveer and Narayanan, Shrikanth},
  title = {Mel frequency spectral domain defenses against adversarial attacks on speech recognition systems},
  journal = {JASA Express Letters},
  year = {2023},
  doi = {10.1121/10.0017680}
}
Kostenlos starten →