ACL RD-TEC 1.0 Summarization of H91-1052
Paper Title:
SESSION 9: SPEECH III
SESSION 9: SPEECH III
Primarily assigned technology terms:
- acoustic modeling
- algorithm
- atis
- baurn-welch ml training
- baurn-welch ml training algorithm
- bayesian learning
- codebook adaptation
- cross-speaker recognition
- decision trees
- decoder
- error reduction
- hmms
- k-means
- k-means training
- language modeling
- learning
- lexical access
- ml training
- model interpolation
- modeling
- multi-layer perceptron
- normalization
- parameter smoothing
- perceptron
- recognition
- recognition search
- recognizer
- search
- si recognition
- smoothing
- speaker adaptation
- speaker normalization
- speech recognition
- stack decoder
- statistical language modeling
- training algorithm
- viterbi
- viterbi training
Other assigned terms:
- approach
- backoff
- backoff model
- bigram
- cache
- codebook
- darpa test set
- data sets
- duration
- error rate
- formalism
- grammar
- grammars
- implementation
- interpolation
- labeled training data
- language model
- language models
- lattice
- learning problem
- lexicon
- mapping
- n-gram
- n-gram models
- natural language
- perplexity
- phoneme
- procedure
- recognition rate
- speech data
- technique
- test set
- topics
- training
- training data
- trees
- trigram
- trigram language model
- unigram
- vocabulary
- word
- word error rate
- word sequence
- word-pair grammar
- words
