A new combined modeling of continuous speech recognition | IEEE Conference Publication | IEEE Xplore