mlsanityData to 7 Oct 2026

WaveGuard: Understanding and Mitigating Audio Adversarial Examples

arXiv 2103.03344 ↗First posted 4 Mar 2021cs.CR cs.LG cs.SDcited by 39

Not on the map. Why?

Abstract (arXiv)

There has been a recent surge in adversarial attacks on deep learning based automatic speech recognition (ASR) systems. These attacks pose new challenges to deep learning security and have raised significant concerns in deploying ASR systems in safety-critical applications. In this work, we introduce WaveGuard: a framework for detecting adversarial inputs that are crafted to attack ASR systems. Our framework incorporates audio transformation functions and analyses the ASR transcriptions of the original and transformed audio to detect adversarial inputs. We demonstrate that our defense framework is able to reliably detect adversarial examples constructed by four recent audio adversarial attacks, with a variety of audio transformation functions. With careful regard for best practices in defense evaluations, we analyze our proposed defense and its strength to withstand adaptive and robust attacks in the audio domain. We empirically demonstrate that audio transformations that recover audio from perceptually informed representations can lead to a strong defense that is robust against an adaptive adversary even in a complete white-box setting. Furthermore, WaveGuard can be used out-of-the box and integrated directly with any ASR model to efficiently detect audio adversarial examples, without the need for model retraining.

Key facts. Cited by 39 papers in the map (data through 7 Oct 2026). Builds on 33 of its 58 references that are in the map. First posted on arXiv 4 Mar 2021.

Builds on

33 of 58 references.

References · 33

  1. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  2. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  3. 2018Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examplescited by 1,259
  4. 2017Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictionscited by 800
  5. 2015The Limitations of Deep Learning in Adversarial Settingscited by 784
  6. 2017Evasion Attacks against Machine Learning at Test Timecited by 686
  7. 2020On Adaptive Attacks to Adversarial Example Defensescited by 547
  8. 2017Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methodscited by 542
  9. 2017Countering Adversarial Images using Input Transformationscited by 517
  10. 2017Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networkscited by 484
  11. 2015Deep Speech 2: End-to-End Speech Recognition in English and Mandarincited by 449
  12. 2019On Evaluating Adversarial Robustnesscited by 442
21 more
  1. 2017Mitigating Adversarial Effects Through Randomizationcited by 382
  2. 2014Deep Speech: Scaling up end-to-end speech recognitioncited by 354
  3. 2017MagNet: a Two-Pronged Defense against Adversarial Examplescited by 268
  4. 2018Audio Adversarial Examples: Targeted Attacks on Speech-to-Textcited by 268
  5. 2019Imperceptible, Robust, and Targeted Adversarial Examples for Automatic Speech Recognitioncited by 123
  6. 2019Defensive Quantization: When Efficiency Meets Robustnesscited by 95
  7. 2019Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modelingcited by 86
  8. 2018Adversarial Attacks Against Automatic Speech Recognition Systems via Psychoacoustic Hidingcited by 83
  9. 2017SafetyNet: Detecting and Rejecting Adversarial Examples Robustlycited by 76
  10. 2018CommanderSong: A Systematic Approach for Practical Adversarial Voice Recognitioncited by 63
  11. 2017Detecting Adversarial Image Examples in Deep Networks with Adaptive Noise Reductioncited by 49
  12. 2018Did you hear that? Adversarial Examples Against Automatic Speech Recognitioncited by 48
  13. 2018Robust Audio Adversarial Example for a Physical Attackcited by 42
  14. 2019Universal Adversarial Perturbations for Speech Recognition Systemscited by 41
  15. 2017Synthesizing Robust Adversarial Examplescited by 39
  16. 2018The PyTorch-Kaldi Speech Recognition Toolkitcited by 38
  17. 2018Characterizing Audio Adversarial Examples Using Temporal Dependencycited by 34
  18. 2019Detecting and Diagnosing Adversarial Images with Class-Conditional Capsule Reconstructionscited by 30
  19. 2018QuSecNets: Quantization-based Defense Mechanism for Securing Deep Neural Network against Adversarial Attackscited by 21
  20. 2019Expediting TTS Synthesis with Adversarial Vocodingcited by 10
  21. 2018Isolated and Ensemble Audio Preprocessing Methods for Detecting Adversarial Examples against Automatic Speech Recognitioncited by 8

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 4; 2022: 11; 2023: 6; 2024: 3; 2025: 7; 2026: 8.

39 papers.

Sort:

By area:

  1. 2022"Real Attackers Don't Compute Gradients": Bridging the Gap Between Adversarial ML Research and Practicecited by 55
  2. 2022SoK: On the Semantic AI Security in Autonomous Drivingcited by 22
  3. 2022A Review of Speech-centric Trustworthy Machine Learning: Privacy, Safety, and Fairnesscited by 13
  4. 2023MASTERKEY: Practical Backdoor Attack Against Speaker Verification Systemscited by 8
  5. 2022A Tale of Two Models: Constructing Evasive Attacks on Edge Modelscited by 6
  6. 2025SafeSpeech: Robust and Universal Voice Protection Against Malicious Speech Synthesiscited by 5
  7. 2021SoK: A Modularized Approach to Study the Security of Automatic Speech Recognition Systemscited by 4
  8. 2022RoVISQ: Reduction of Video Service Quality via Adversarial Attacks on Deep Learning-based Video Compressioncited by 4
  9. 2021Robustifying automatic speech recognition by extracting slowly varying featurescited by 3
  10. 2023Improving Robustness Against Adversarial Attacks with Deeply Quantized Neural Networkscited by 3
  11. 2022Privacy against Real-Time Speech Emotion Detection via Acoustic Adversarial Evasion of Machine Learningcited by 2
  12. 2023Security and Privacy Problems in Voice Assistant Applications: A Surveycited by 2
27 more
  1. 2024Towards Evaluating the Robustness of Automatic Speech Recognition Systems via Audio Style Transfercited by 2
  2. 2024Detecting and Defending Against Adversarial Attacks on Automatic Speech Recognition via Diffusion Modelscited by 2
  3. 2024Transferable Adversarial Attacks against ASRcited by 2
  4. 2025De-AntiFake: Rethinking the Protective Perturbations Against Voice Cloning Attackscited by 2
  5. 2026Hijacking Large Audio-Language Models via Context-Agnostic and Imperceptible Auditory Prompt Injectioncited by 2
  6. 2021On the Exploitability of Audio Machine Learning Pipelines to Surreptitious Adversarial Examplescited by 1
  7. 2021Catch Me If You Can: Blackbox Adversarial Attacks on Automatic Speech Recognition using Frequency Maskingcited by 1
  8. 2022Attacks as Defenses: Designing Robust Audio CAPTCHAs Using Attacks on Automatic Speech Recognition Systemscited by 1
  9. 2022Mel Frequency Spectral Domain Defenses against Adversarial Attacks on Speech Recognition Systemscited by 1
  10. 2022Universal Fourier Attack for Time Seriescited by 1
  11. 2023Intrinsic Biologically Plausible Adversarial Robustnesscited by 1
  12. 2023Is Certifying ℓp Robustness Still Worthwhile?cited by 1
  13. 2025Investigating Vulnerabilities and Defenses Against Audio-Visual Attacks: A Comprehensive Survey Emphasizing Multimodal Modelscited by 1
  14. 2026Off-The-Shelf Image-to-Image Models Are All You Need To Defeat Image Protection Schemescited by 1
  15. 2022Hear No Evil: Towards Adversarial Robustness of Automatic Speech Recognition via Multi-Task Learningcited by 0
  16. 2022Privacy-Utility Balanced Voice De-Identification Using Adversarial Examplescited by 0
  17. 2023RobustPdM: Designing Robust Predictive Maintenance against Adversarial Attackscited by 0
  18. 2025Quantitative Analysis of Deeply Quantized Tiny Neural Networks Robust to Adversarial Attackscited by 0
  19. 2025Are Deep Speech Denoising Models Robust to Adversarial Noise?cited by 0
  20. 2025Adversarial Attacks in Multimodal Systems: A Practitioner's Surveycited by 0
  21. 2025E2E-VGuard: Adversarial Prevention for Production LLM-based End-To-End Speech Synthesiscited by 0
  22. 2026Precision-Varying Prediction (PVP): Robustifying ASR systems against adversarial attackscited by 0
  23. 2026Codec-Robust Attacks on Audio LLMscited by 0
  24. 2026Beyond Waveform Robustness: Robust Feature-Vocoder Adversarial Attacks on Automatic Speech Recognitioncited by 0
  25. 2026Hearing the Unspoken: Language Model Priors for Acoustic Adversarial Attackscited by 0
  26. 2026What Was That Again? Certified Robustness for Automatic Speech Recognitioncited by 0
  27. 2026Psychoacoustically Aligned Latent Smoothing for Adversarial Robustness of Full-Duplex Speech-to-Speech Dialogue Modelscited by 0