<< Adversarial audio attacks can be considered as a small perturbation unperceptive to human ears that is intentionally added to the audio signal and causes a machine learning model to make mistakes. >>
Mohammad Esmaeilpour, Patrick Cardinal, Alessandro Lameiras Koerich. A Robust Approach for Securing Audio Classification Against Adversarial Attacks. arXiv:1904.10990 [cs.LG] Apr 24, 2019.
https://arxiv.org/abs/1904.10990
Ingrid Fadelli. An approach for securing audio classification against adversarial attacks. May 7, 2019.
https://m.techxplore.com/news/2019-05-approach-audio-classification-adversarial.html