Natural Backdoor Attacks on Speech Recognition Models

arXiv:2607.15724 · cs.CR, cs.LG, cs.SD · Submitted 2026-07-17 · Read on arXiv

Jinwen Xin, Xixiang Lyu, Jing Ma

cs.CR, cs.LG, cs.SD

Submitted: 2026-07-17

Comments: This is the authors' manuscript of a chapter published in Machine Learning for Cyber Security, Lecture Notes in Computer Science, vol. 13655, pp. 597-610 (2023)

Journal ref: Machine Learning for Cyber Security, Lecture Notes in Computer Science, vol. 13655, pp. 597-610 (2023)

DOI: 10.1007/978-3-031-20096-0_45

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Accuracy — 3× in this paper · explained in 1 episode
• Adversarial Attacks — 3× in this paper · explained in 1 episode
• Eating Sound Collection — 2× in this paper
• Transfer Learning — 2× in this paper · explained in 10 episodes
• Verification — 2× in this paper · explained in 2 episodes
• Adversarial Examples — 1× in this paper · explained in 2 episodes
• Benchmark — 1× in this paper · explained in 2 episodes
• Benign Accuracy — 1× in this paper
• Clustering — 1× in this paper · explained in 2 episodes
• Feature Extraction — 1× in this paper · explained in 2 episodes
• Federated Learning — 1× in this paper · explained in 14 episodes
• Loss Function — 1× in this paper · explained in 2 episodes

Sources

• Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
• Towards Deep Learning Models Resistant to Adversarial Attacks
• Can You Hear It? Backdoor Attacks via Ultrasonic Triggers
• Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning
• Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer
• Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger
• Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition
• A neural attention model for speech command recognition

Related papers

• SoK: AI-Augmented Binary Reversing
• Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
• Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
• Efficient Fuzzy PSI under One-Sided Assumptions
• Sealing the Audit-Runtime Gap for LLM Skills
• Token Composition: A Graph Based on EVM Logs