A simplified Subspace Gaussian Mixture to compact acoustic models for speech recognition.
Mohamed BouallegueDriss MatroufGeorges LinarèsPublished in: ICASSP (2011)
Keyphrases
- speech recognition
- gaussian mixture
- acoustic models
- speech recognizer
- speaker independent
- gaussian mixture model
- em algorithm
- hidden markov models
- probability density function
- expectation maximization
- automatic speech recognition
- density estimation
- covariance matrix
- low dimensional
- mixture model
- language model
- pattern recognition
- speech signal
- speech recognition systems
- feature space
- noisy environments
- speaker identification
- principal component analysis
- high dimensional
- high dimensional data
- dimensionality reduction
- generative model
- model selection
- data points