ACL RD-TEC 1.0 Summarization of H89-2034
Paper Title:
SPEAKER ADAPTATION USING MULTIPLE REFERENCE SPEAKERS
SPEAKER ADAPTATION USING MULTIPLE REFERENCE SPEAKERS
Authors: Francis Kubala and Richard Schwartz and Chris Barry
Primarily assigned technology terms:
- approximation
- computing
- continuous speech recognition
- cross-speaker recognition
- data collection
- database
- dynamic time warping
- feature normalization
- hidden markov
- hidden markov model
- hidden markov models
- large vocabulary continuous speech recognition
- markov model
- matching
- normalization
- normalization procedure
- partitioning
- recognition
- reference speaker selection
- resource management
- smoothing
- speaker adaptation
- speaker-adaptation
- speaker-adaptation system
- speaker-dependent recognition
- speaker-dependent training
- speaker-independent recognition
- speaker-independent training
- speech normalization
- speech recognition
- speech recognition evaluation
- time warping
- transcription
- weighting
Other assigned terms:
- approach
- baseline performance
- case
- class membership
- codebook
- continuous speech
- cross-speaker model
- error rate
- estimation
- evaluation test
- evaluations
- experimental results
- fact
- feature
- feature set
- feature space
- feature vectors
- frequency counts
- grammar
- index
- labeling
- large training
- lexicon
- mapping
- markov models
- method
- model parameters
- norm
- phonemes
- procedure
- recognition error rate
- recognition evaluation
- speaker identity
- speaking style
- speech database
- style
- system description
- technique
- term
- terms
- test data
- test material
- test set
- tokens
- training
- training and test data
- training data
- training material
- training set
- transformation
- vocabulary
- word
- word error rate
- word-pair grammar
