Fast match acoustic models in large vocabulary continuous speech recognition | IEEE Conference Publication | IEEE Xplore