ACL RD-TEC 1.0 Summarization of H91-1031
Paper Title:
SIGNAL REPRESENTATION ATTRIBUTE EXTRACTION AND THE USE DISTINCTIVE FEATURES FOR PHONETIC CLASSIFICATION
SIGNAL REPRESENTATION ATTRIBUTE EXTRACTION AND THE USE DISTINCTIVE FEATURES FOR PHONETIC CLASSIFICATION
Authors: Helen M. Meng and Victor W. Zue and Hong C. Leung
Primarily assigned technology terms:
- acoustic attribute extraction
- attribute extraction
- automatic speech recognition
- classification
- classifier
- collapsing
- computing
- data-driven approach
- feature representation
- identification
- input normalization
- learning
- linear discriminant
- modelling
- multi-layer perceptron
- neural network
- normalization
- optimization
- perceptron
- phoneme classification
- phoneme\/feature classification
- phonetic classification
- preprocessing
- principal component analysis
- recognition
- search
- speaker normalization
- speech recognition
- table look-up
- vowel classification
Other assigned terms:
- acoustic attribute
- acoustic information
- acoustic-phonetic information
- american english
- approach
- auditory model
- binary feature
- case
- classification accuracy
- classification performance
- classification task
- coefficient
- community
- comparative study
- context dependency
- corpora
- duration
- error rate
- feature
- feature specification
- feature vector
- feature vectors
- lexicon
- linear frequency
- linguistic
- linguists
- maps
- measure
- measures
- noise
- noisy speech
- optimization criterion
- phoneme
- phonemes
- pitch
- probability
- procedure
- process
- relation
- representations
- run-time
- segment duration
- signal
- signal-to-noise ratio
- speech signal
- terms
- test set
- testing data
- testing set
- timit corpus
- tokens
- training
- training and testing data
- training data
- training set
- understanding
- vowel
- vowel articulation
- white noise
