Rapid speaker adaptation using speaker-mixture allophone models applied to speaker-independent speech recognition | IEEE Conference Publication | IEEE Xplore