ACL RD-TEC 1.0 Summarization of H89-2039
Paper Title:
ACOUSTIC MODELING OF SUBWORD UNITS FOR LARGE VOCABULARY SPEAKER INDEPENDENT SPEECH RECOGNITION
ACOUSTIC MODELING OF SUBWORD UNITS FOR LARGE VOCABULARY SPEAKER INDEPENDENT SPEECH RECOGNITION
Authors: Chin-Hui Lee and Lawrence R. Rabiner and Roberto Pieraccinit and Jay G. Wilpon
Primarily assigned technology terms:
- acoustic modeling
- algorithm
- backtracking
- beam search
- beam search algorithm
- beam search recognition
- clustering
- corrective training
- database
- database management
- decoding
- discrimination procedure
- feature representation
- finite state
- forward-backward algorithm
- frame-synchronous beam search
- hidden markov
- hidden markov model
- hidden markov models
- k-means
- k-means training
- large vocabulary speech recognition
- lexical modeling
- lpc analysis
- markov model
- matching
- model creation
- modeling
- parsing
- pattern recognition
- recognition
- recognition algorithm
- recognition system
- recognition systems
- recognizer
- reporting
- resource management
- sampling
- search
- search algorithm
- segmentation
- smoothing
- speech recognition
- training procedure
- viterbi
- viterbi decoding
- vocabulary recognition
- vq design
- word bigram
- word deletion
- word discrimination
- word insertion
- word recognition
- word substitution
Other assigned terms:
- acoustic models
- approach
- beam
- bigram
- byblos system
- case
- cluster
- clusters
- codebook
- composition
- content words
- context dependency
- continuous speech
- convergence
- darpa naval resource management
- derivational equivalence
- duration
- estimation
- evaluations
- fact
- feature
- feature vector
- finite state network
- frame
- function word
- function words
- gaussian mixture
- grammar
- grammar \
- grammar case
- implementation
- interpolation
- interpolation scheme
- khz rate
- knowledge
- labeling
- language model
- large vocabulary speech
- lexical entries
- lexical entry
- lexical representation
- lexicon
- likelihood
- linear combination
- mapping
- markov models
- message
- method
- methodology
- model parameters
- names
- natural language
- open test
- pairs of words
- perplexity
- phoneme
- phonemes
- phrase
- priori
- probabilities
- procedure
- recognition accuracy
- recognition task
- representations
- resource management task
- segments
- sentence
- sentences
- silence
- speech database
- speech input
- statistics
- symbol
- symbols
- technique
- terms
- test set
- text
- tokens
- training
- training data
- training material
- training set
- utterance
- vocabulary
- word
- word boundaries
- word model
- word pair
- word pair grammar
- word sequence
- words
