Minimum Phoneme Error Based Heteroscedastic Linear Discriminant Analysis for Speech Recognition.
Bing ZhangSpyros MatsoukasPublished in: ICASSP (1) (2005)
Keyphrases
- linear discriminant analysis
- speech recognition
- discriminant analysis
- face recognition
- speech synthesis
- dimensionality reduction
- feature extraction
- hidden markov models
- principal component analysis
- language model
- pattern recognition
- automatic speech recognition
- support vector
- speaker dependent
- dimension reduction
- high dimensional data
- support vector machine svm
- feature space
- speech signal
- null space
- error rate
- speech recognizer
- speaker identification
- speech recognition systems
- scatter matrix
- similarity measure
- low dimensional
- machine learning
- data sets
- speaker independent
- noisy environments
- probabilistic model
- maximum likelihood