On speech features fusion, α-integration Gaussian modeling and multi-style training for noise robust speaker classification.
A. VenturiniLeonardo ZãoRosângela CoelhoPublished in: IEEE ACM Trans. Audio Speech Lang. Process. (2014)
Keyphrases
- classification accuracy
- feature vectors
- noisy environments
- feature space
- feature set
- restricted boltzmann machine
- speech recognition
- training set
- rotational invariant
- classification method
- spectral features
- feature extraction
- classification performances
- mel frequency cepstral coefficients
- training samples
- emotion classification
- classification models
- audio visual
- svm classifier
- speaker verification
- data fusion
- support vector
- training phase
- multiple features
- decision trees
- visual speech
- discriminatory power
- speaker identification
- extracting features
- supervised learning
- extracted features
- highly discriminative
- image features
- speaker recognition
- training examples
- speech signal
- training and testing data
- lexical features
- pattern recognition
- noise reduction
- feature subset
- training process
- automatic speech recognition
- discriminative classifiers
- support vector machine
- image classification
- probabilistic neural network
- hidden markov models
- feature selection
- linear svm
- acoustic models
- support vector machine svm
- gaussian noise