ACL RD-TEC 1.0 Summarization of H93-1014
Paper Title:
EFFICIENT CEPSTRAL NORMALIZATION FOR ROBUST SPEECH RECOGNITION
EFFICIENT CEPSTRAL NORMALIZATION FOR ROBUST SPEECH RECOGNITION
Authors: Fu-Hua Liu and Richard M. Stern and Xuedong Huang and Alejandro Acero
Primarily assigned technology terms:
- algorithm
- atis
- automatic speech recognition
- cart decision tree
- cdcn algorithm
- cepstral normalization
- cepstral-filtedng
- classification
- classification process
- continuous speech recognition
- database
- decision tree
- em algorithm
- environmental compensation
- incremental adaptation
- iterative algorithm
- log-spectral representation
- mean normalization
- normalization
- recognition
- recognition process
- recognition system
- recognition system \
- recognition systems
- robust speech recognition
- sdcn
- snr-dependent cepstral normalization
- spectral subtraction
- speech recognition
- speech recognition system
- speech recognition systems
- speech system
- sphinx-ii
- spoken language systems
Other assigned terms:
- additive noise
- approach
- benchmark
- bigram
- bigram language model
- case
- cepstral domain
- clusters
- codebook
- computational complexity
- continuous speech
- convergence
- density function
- dictionary
- document
- duration
- error rate
- evaluation data
- evaluations
- experimental results
- frame
- grammar
- knowledge
- language model
- likelihood
- measures
- method
- noise
- posteriori probability
- priori
- probabilities
- probability
- probability density
- probability density function
- procedure
- process
- punctuation
- recognition accuracy
- sentence-by-sentence basis
- sentences
- signal
- speech signal
- spoken language
- stress
- test data
- test set
- text
- training
- training data
- training database
- training set
- tree
- trigram
- utterance
- word
- word error rate
- words
