ACL RD-TEC 1.0 Summarization of H92-1080
Paper Title:
APPLYING SPHINX-II TO THE DARPA WALL STREET JOURNAL CSR TASK
APPLYING SPHINX-II TO THE DARPA WALL STREET JOURNAL CSR TASK
Authors: F. Alleva and H. Hon and X. Huang and M. Hwang and R. Rosenfeld and R. Weide
Primarily assigned technology terms:
- acoustic modeling
- algorithm
- beam search
- beam search strategy
- beam-search
- clustering
- computing
- context modeling
- continuous speech recognition
- decoder
- decoding
- decomposition
- dynamic programming
- dynamic programming algorithm
- evaluation procedure
- forward-backward algorithm
- hidden markov
- hidden markov models
- hmms
- lpc analysis
- modeling
- normalization
- partitioning
- processing
- programming algorithm
- pruning
- recognition
- recognition system
- resource management
- scoring
- search
- secondary storage
- segmentation
- signal processing
- speaker normalization
- speech recognition
- speech recognition system
- sphinx-ii
- supervised training
- table lookup
- training procedure
- transcription
- viterbi
- viterbi beam
- viterbi beam search
- vocabulary adaptation
- word modeling
- wsj csr
Other assigned terms:
- acoustic models
- acoustic score
- approach
- backoff
- beam
- bigram
- bigram language model
- cache
- case
- clustering procedure
- co-articulation
- codebook
- context model
- contextual information
- continuous speech
- data set
- data sets
- data structure
- dictionary
- dictionary data
- distribution
- document
- error rate
- evaluation data
- evaluation test
- events
- feature
- frame
- grammar
- grammars
- implementation
- interpolation
- language model
- language models
- lexical coverage
- lexicon
- mapping
- mapping table
- markov models
- method
- nist
- noise
- orthographic transcription
- probabilities
- probability
- probability distribution
- procedure
- pronunciation
- pruning threshold
- punctuation
- recognition task
- resource management task
- search space
- search strategy
- sentence
- sentences
- signal
- speech recognition task
- speech signal
- standard deviation
- style
- technology
- terms
- test data
- test data set
- test set
- text
- topology
- training
- training corpus
- training data
- training data set
- transformation
- transition probabilities
- transition probability
- triphone
- uniform distribution
- utterance
- vocabulary
- word
- word boundaries
- word error rate
- word error rates
- word level
- word sequence
- word sequences
- words
- wsj corpus
