ACL RD-TEC 1.0 Summarization of H89-2044
Paper Title:
ACOUSTICAL PRE-PROCESSING FOR ROBUST SPEECH RECOGNITION
ACOUSTICAL PRE-PROCESSING FOR ROBUST SPEECH RECOGNITION
Authors: Richard M. Stern and Alejandro Acero
Primarily assigned technology terms:
- algorithm
- blind deconvolution
- computing
- corrective training
- database
- database collection
- database collection and analysis
- databases
- deconvolution
- evaluation system
- feature extraction
- normalization
- parsing
- peripheral processing
- pre-processing
- preprocessing
- processing
- recognition
- recognition systems
- resource management
- robust speech recognition
- sampling
- scoring
- signal processing
- signal-processing
- spectral normalization
- spectral subtraction
- speech recognition
- speech recognition systems
- speech-recognition
- table lookup
- training process
- transducer
- transfer function
Other assigned terms:
- additive noise
- array
- background noise
- case
- cepstral domain
- cepstral vector
- cluster
- correlations
- degraded speech
- derivational equivalence
- duration
- error rate
- estimation
- evaluations
- events
- experimental results
- fact
- feature
- feature vectors
- frame
- implementation
- method
- names
- noise
- normalization factor
- pauses
- perplexity
- procedure
- process
- recognition accuracy
- recursion
- resource management task
- segments
- signal
- signal energy
- signal-to-noise ratio
- silence
- speech signal
- terms
- training
- training time
- triphone
- understanding
- vocabulary
- vocabulary size
- vocal tract
- word
- words
