ACL RD-TEC 1.0 Summarization of H89-2047
Paper Title:
IMPROVEMENTS IN THE STOCHASTIC SEGMENT MODEL FOR PHONEME RECOGNITION
IMPROVEMENTS IN THE STOCHASTIC SEGMENT MODEL FOR PHONEME RECOGNITION
Authors: V. Digalakis and M. Ostendorf and J.R. Rohlicek
Primarily assigned technology terms:
- acoustic modelling
- algorithm
- algorithm development
- approximation
- automatic recognition
- automatic segmentation
- automatic training
- classification
- classifier
- continuous speech recognition
- database
- em algorithm
- hidden markov
- hidden markov model
- hidden markov modelling
- hidden markov models
- hmms
- iterative algorithm
- linear discriminant
- markov model
- markov modelling
- matrix inversion
- maximum likelihood
- modelling
- paraameter estimation
- parameter estimation
- parameter reduction
- parameterization
- partitioning
- phoneme classification
- phoneme insertion
- phoneme recognition
- recognition
- recognition algorithm
- recognition system
- reestimation
- search
- segmentation
- speaker-independent phoneme classification
- speaker-independent phoneme recognition
- speech recognition
- speech recognition system
- time warping
- viterbi
- viterbi search
- word recognition
Other assigned terms:
- acoustic model
- approach
- case
- classification performance
- classification task
- classification tasks
- continuous speech
- correlation
- data set
- distribution
- duration
- duration information
- estimation
- experimental results
- fact
- feature
- feature vectors
- frame
- gaussian distribution
- hypothesis
- independence assumption
- interpolation
- lemma
- likelihood
- likelihood parameter
- linear time
- mapping
- markov chain
- markov models
- mechanisms
- method
- model performance
- phoneme
- phonemes
- posteriori probability
- probability
- process
- segments
- sentences
- statistics
- stochastic segment model
- technologies
- terms
- test set
- timit database
- training
- training data
- training set
- transformation
- transposition
- triphone
- understanding
- word
