ACL RD-TEC 1.0 Summarization of H89-2037
Paper Title:
TOWARDS SPEECH RECOGNITION WITHOUT VOCABULARY-SPECIFIC TRAINING
TOWARDS SPEECH RECOGNITION WITHOUT VOCABULARY-SPECIFIC TRAINING
Authors: Hsiao-Wuen Hon and Kai-Fu Lee and Robert Weide
Primarily assigned technology terms:
- algorithm
- bottom-up clustering
- clustering
- clustering technique
- context generalization
- continuous speech recognition
- corrective training
- data collection
- database
- databases
- dictionary generation
- error rate reduction
- forward-backward algorithm
- hmms
- interpolation algorithm
- large-vocabulary continuous speech recognition
- maximum likelihood
- maximum-likelihood
- modeling
- parsing
- processing
- rate reduction
- recognition
- recognizer
- resource management
- search
- speaker-dependent training
- speaker-independent recognition
- speaker-independent speech recognition
- speech recognition
- speech recognizer
- subword clustering
- subword modeling
- top-down approach
- top-down clustering
- training process
Other assigned terms:
- accent
- alphabet
- approach
- cluster
- clustering procedure
- clusters
- continuous speech
- derivational equivalence
- dictionary
- error rate
- events
- experimental results
- fact
- generation
- grammar
- interpolation
- intonation
- knowledge
- large training
- likelihood
- model parameters
- perplexity
- probability
- procedure
- process
- prosody
- resource management task
- sentences
- speech research
- statistics
- stress
- technique
- timit database
- trained model
- training
- training corpus
- training data
- training database
- training set
- training time
- triphone
- usability
- vocabulary
- word
- word-pair grammar
- words
