H90-1004 be incorporated directly into a continuous speech recognizer . Because except by using an
H90-1004 information is processed by a continuous speech recognizer . The output of a continuous
C04-1168 an HMM-based , large vocabulary continuous speech recognizer . The acoustic HMMs were triphone
H05-1026 in a state-ofthe-art real-time continuous speech recognizer for French Broadcast News . Word
H92-1066 prefiltering ) models for Dragon 's continuous speech recognizer . The rapid match module is intended
H92-1082 and language model matching in a continuous speech recognizer . A previous paper described
H90-1045 . Although the performance of continuous speech recognizers has improved significantly in
H90-1004 top-N sentence hypotheses in a continuous speech recognizer . The search A deg uses an algorithm
H90-1004 interface was proposed to link a continuous speech recognizer and higher level language modules
H93-1023 makes use of a large vocabulary continuous speech recognizer . We present a theoretical framework
E09-1019 reserved for testing . We use a continuous speech recognizer optimized for low memory footprint
H93-1023 which uses a large vocabulary continuous speech recognizer as a preprocessor of the speech
C00-2158 utilized in a large-vocabulary , continuous speech recognizer to achieve low word-error rates
H91-1091 demonstrate robust , high-performance continuous speech recognizer ( CSR ) techniques and systems
H90-1089 demonstrate robust , high-performance continuous speech recognizer ( CSR ) techniques and systems
H90-1036 we have found that running our continuous speech recognizer with rapid match speeds up the
H89-2067 demonstrate robust , high-performance continuous speech recognizer ( CSR ) techniques and systems
H89-2028 specification for an interface between a continuous speech recognizer ( CSR ) and a natural language
H92-1066 rapid match module for Dragon 's continuous speech recognizer was first presented nearly two
H89-2028 a Hidden Markov Model ( HMM ) Continuous Speech Recognizer ( CSR ) , and a grammatical component
hide detail