Deep Spoken Keyword Spotting: An Overview

Spoken keyword spotting (KWS) deals with the identification of keywords in audio streams and has become a fast-growing technology thanks to paradigm shift introduced by deep learning few years ago. This allowed rapid embedding KWS myriad small electronic devices different purposes like activation voice assistants. Prospects suggest sustained growth terms social use this technology. Thus, it is not surprising that hot research topic among speech scientists, who constantly look for performance improvement computational complexity reduction. context motivates paper, which we conduct literature review into spoken assist practitioners researchers are interested Specifically, overview comprehensive nature covering thorough analysis systems (which includes features, acoustic modeling posterior handling), robustness methods, applications, datasets, evaluation metrics, audio-visual KWS. The performed paper allows us identify number directions future research, including adopted from automatic recognition unique problem