mlsanityData to 14 Oct 2026

There is more than one kind of robustness: Fooling Whisper with adversarial examples

arXiv 2210.17316 ↗First posted 26 Oct 2022eess.AS cs.AI cs.CLcited by 11

Not on the map. Why?

Abstract (arXiv)

Whisper is a recent Automatic Speech Recognition (ASR) model displaying impressive robustness to both out-of-distribution inputs and random noise. In this work, we show that this robustness does not carry over to adversarial noise. We show that we can degrade Whisper performance dramatically, or even transcribe a target sentence of our choice, by generating very small input perturbations with Signal Noise Ratio of 35-45dB. We also show that by fooling the Whisper language detector we can very easily degrade the performance of multilingual models. These vulnerabilities of a widely popular open-source model have practical security implications and emphasize the need for adversarially robust ASR.

Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Builds on 21 of its 26 references that are in the map. First posted on arXiv 26 Oct 2022.

Builds on

21 of 26 references.

References · 21

  1. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  2. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  3. 2013Intriguing properties of neural networkscited by 5,155
  4. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
  5. 2019Common Voice: A Massively-Multilingual Speech Corpuscited by 1,384
  6. 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,340
  7. 2018Certified Robustness to Adversarial Examples with Differential Privacycited by 468
  8. 2021SpeechBrain: A General-Purpose Speech Toolkitcited by 406
  9. 2018Audio Adversarial Examples: Targeted Attacks on Speech-to-Textcited by 274
  10. 2019Imperceptible, Robust, and Targeted Adversarial Examples for Automatic Speech Recognitioncited by 124
  11. 2016Robustness of classifiers: from adversarial to random noisecited by 88
  12. 2020SoK: The Faults in our ASRs: An Overview of Attacks against Automatic Speech Recognition and Speaker Identification Systemscited by 46
9 more
  1. 2020Enabling Fast and Universal Audio Adversarial Attack Using Generative Modelcited by 32
  2. 2021Adversarial Attacks and Defenses for Speech Recognition Systemscited by 20
  3. 2019Imperio: Robust Over-the-Air Adversarial Examples for Automatic Speech Recognition Systemscited by 16
  4. 2021Sequential Randomized Smoothing for Adversarially Robust Speech Recognitioncited by 13
  5. 2022Recent improvements of ASR models in the face of adversarial attackscited by 13
  6. 2021Exploring Targeted Universal Adversarial Perturbations to End-to-end ASR Modelscited by 10
  7. 2021Are Multilingual BERT models robust? A Case Study on Adversarial Attacks for Multilingual Question Answeringcited by 6
  8. 2021Characterizing the adversarial vulnerability of speech self-supervised learningcited by 6
  9. 2022Language Dependencies in Adversarial Attacks on Speech Recognition Systemscited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 3; 2024: 4; 2025: 0; 2026: 4.

11 papers.

  1. 2023Reproducing Whisper-Style Training Using an Open-Source Toolkit and Publicly Available Datacited by 42
  2. 2023N-Shot Benchmarking of Whisper on Diverse Arabic Speech Recognitioncited by 20
  3. 2024Muting Whisper: A Universal Acoustic Adversarial Attack on Speech Foundation Modelscited by 12
  4. 2024Transferable Adversarial Attacks against ASRcited by 4
  5. 2023VoxArabica: A Robust Dialect-Aware Arabic Speech Recognition Systemcited by 3
  6. 2024Controlling Whisper: Universal Acoustic Adversarial Attacks to Control Speech Foundation Modelscited by 2
  7. 2026SyncBreaker:Stage-Aware Multimodal Adversarial Attacks on Audio-Driven Talking Head Generationcited by 2
  8. 2024Comparative Study on Noise-Augmented Training and its Effect on Adversarial Robustness in ASR Systemscited by 1
  9. 2026MORE: Multi-Objective Adversarial Attacks on Speech Recognitioncited by 0
  10. 2026Beyond Waveform Robustness: Robust Feature-Vocoder Adversarial Attacks on Automatic Speech Recognitioncited by 0
  11. 2026Room for Error: Large-Scale Simulation of Over-the-Air Acoustic Attackscited by 0