ACL RD-TEC 1.0 Summarization of H90-1035
Paper Title:
PHONEME-IN-CONTEXT MODELING FOR DRAGON'S CONTINUOUS SPEECH RECOGNIZER
PHONEME-IN-CONTEXT MODELING FOR DRAGON'S CONTINUOUS SPEECH RECOGNIZER
Authors: Paul Bamberg and Laurence Gillick
Primarily assigned technology terms:
- algorithm
- alignment algorithm
- approximation
- clustering
- continuous speech recognition
- continuous speech recognizer
- continuous-speech recognition
- database
- digit recognition
- dynamic programming
- hidden markov
- hidden markov modeling
- hidden markov models
- isolated-word recognition
- labeler
- large-vocabulary continuous speech recognition
- linear alignment
- markov model
- markov modeling
- modeling
- parallel processing
- parameter tying
- phonetic spelling
- processing
- real-time recognition
- recognition
- recognition system
- recognizer
- resource management
- search
- speaker adaptation
- spectral interpretation
- speech recognition
- speech recognizer
- speech-recognition
- spelling
- training algorithm
- training process
- triphone clustering
Other assigned terms:
- alphabet
- approach
- case
- clustering procedure
- continuous speech
- continuous-speech
- dictionary
- distribution
- duration
- duration information
- error rate
- fact
- feature
- frame
- grammar
- hypothesis
- implementation
- interpretation
- labeled training data
- labeling
- likelihood
- markov models
- perplexity
- phoneme
- phonemes
- phrase
- precision
- probabilistic model
- probabilities
- probability
- probability distribution
- procedure
- process
- pronunciation
- recognition task
- resource management task
- secondary stress
- segments
- sentence
- sentences
- silence
- spoken language
- stress
- suffix
- syllables
- term
- terms
- test data
- text
- tokens
- training
- training data
- training database
- training set
- transition probabilities
- triphone
- user
- utterance
- vocabulary
- vocabulary size
- vowel
- word
- word boundaries
- word level
- words
