A frequency-weighted HMM based on minimum error classification for noisy speech recognition.
Hiroshi MatsumotoMasanori OnoPublished in: ICASSP (1997)
Keyphrases
- speech recognition
- minimum error
- hidden markov models
- noisy environments
- pattern recognition
- language model
- classification accuracy
- speech synthesis
- speech recognition technology
- cepstral coefficients
- speech signal
- speech recognizer
- speech recognition systems
- automatic speech recognition
- feature vectors
- support vector
- speaker independent
- machine learning
- feature extraction
- decision trees
- text classification
- cross validation
- feature selection
- support vector machine svm
- discriminant function
- support vector machine
- computer vision
- speaker identification
- handwriting recognition
- decision rules
- training samples
- digit recognition
- image classification
- feature set
- image processing