Assessing the Vulnerability of Self-Supervised Speech Representations for Keyword Spotting Under White-Box Adversarial Attacks | IEEE Conference Publication | IEEE Xplore