ACL RD-TEC 1.0 Summarization of H93-1015
Paper Title:
COMPARATIVE EXPERIMENTS ON LARGE VOCABULARY SPEECH RECOGNITION
COMPARATIVE EXPERIMENTS ON LARGE VOCABULARY SPEECH RECOGNITION
Authors: Richard Schwartz and Tasos Anastasakos and Francis Kubala and John Makhoul and Long Nguyen and George Zavaliagkos
Primarily assigned technology terms:
- algorithm
- atis
- blind deconvolution
- blind deeonvolution
- byblos speech recognition
- classifier
- corpus collection
- deconvolution
- feature preprocessing
- forward pass
- hidden markov
- hidden markov models
- hmms
- language modeling
- language processing
- large vocabulary speech recognition
- microphone adaptation
- microphone selection
- model rescoring
- modeling
- multi-pass search
- n-best paradigm
- n-best rescoring
- natural language processing
- optimization
- phoneme insertion
- phrasing
- preprocessing
- processing
- reading
- recognition
- recognition system
- recognizer
- rescoring
- resource management
- scoring
- search
- selection algorithm
- speaker adaptation
- speaker-dependent recognition
- speaker-independent recognition
- speaker-independent training
- speech recognition
- speech recognition system
- training process
- transcription
- vocabulary recognition
Other assigned terms:
- approach
- bigram
- bigram language model
- break
- byblos system
- case
- cepstral vector
- codebook
- context models
- continuous speech
- csr corpus
- dictionaries
- dictionary
- error rate
- fact
- feature
- finite-state language model
- frame
- grammar
- grammars
- hypotheses
- hypothesis
- index
- knowledge
- language model
- language models
- large vocabulary speech
- likelihood
- markov models
- method
- n-best list
- n-gram
- n-gram models
- natural language
- noise
- ordered list
- orthographic transcription
- pauses
- phoneme
- phonemes
- probability
- process
- punctuation
- recognition accuracy
- recognition error rate
- rm corpus
- search space
- search strategy
- sentence
- sentences
- signal
- silence
- speech corpus
- stress
- symbols
- technique
- test set
- text
- training
- training data
- training set
- transcriptions
- transformation
- trigram
- trigram language model
- trigram model
- triphone
- user
- utterance
- vocabulary
- wall street journal corpus
- word
- word error rate
- word error rates
- word sequence
- word sequences
- words
- wsj corpus
