Keyword Spotting

We propose a novel speech keyword spotting mechanism which is robust against attacks exploiting the inherent vulnerabilities of an ML model in correctly classifying the spoken word.