Kenji Kita
Generalized LR Parsing in Hidden
Markov Model
Abstract:This report describes a continuous speech recognition method called HMM-LR.
This method uses an efficient parsing mechanism, a generalized LR parser,
driving an HMM-based speech recognizer directly without any intervening
structures such as a phoneme lattice. Very accurate, efficient speech
recognition/parsing is achieved through the integrated processes of speech
recognition and language analysis.