Speaking style compensation on synthetic audio for robust keyword spotting | IEEE Conference Publication | IEEE Xplore