Audio Adversarial Examples: Attacks Using Vocal Masks
Sound
2021-02-09 v2 Artificial Intelligence
Audio and Speech Processing
Abstract
We construct audio adversarial examples on automatic Speech-To-Text systems . Given any audio waveform, we produce an another by overlaying an audio vocal mask generated from the original audio. We apply our audio adversarial attack to five SOTA STT systems: DeepSpeech, Julius, Kaldi, wav2letter@anywhere and CMUSphinx. In addition, we engaged human annotators to transcribe the adversarial audio. Our experiments show that these adversarial examples fool State-Of-The-Art Speech-To-Text systems, yet humans are able to consistently pick out the speech. The feasibility of this attack introduces a new domain to study machine and human perception of speech.
Cite
@article{arxiv.2102.02417,
title = {Audio Adversarial Examples: Attacks Using Vocal Masks},
author = {Kai Yuan Tay and Lynnette Ng and Wei Han Chua and Lucerne Loke and Danqi Ye and Melissa Chua},
journal= {arXiv preprint arXiv:2102.02417},
year = {2021}
}
Comments
9 pages, 1 figure, 2 tables. Submitted to COLING2020