Wavelet Perceptual Feature Extraction and Quotient Itakura-Saito Distance Measure for Polish Speech Recognition.
Jakub GalkaTomasz JadczykMariusz ZiólkoPublished in: Bio Algorithms Med Syst. (2010)
Keyphrases
- distance measure
- speech recognition
- feature extraction
- relative entropy
- wavelet transform
- bregman divergences
- pattern recognition
- kullback leibler divergence
- texture features
- euclidean distance
- speaker identification
- cepstral coefficients
- mahalanobis distance
- similarity measure
- speech signal
- hidden markov models
- information theoretic
- speech processing
- automatic speech recognition
- language model
- speech synthesis
- noisy environments
- frequency domain
- speech recognizer
- image processing
- multiscale
- wavelet coefficients
- feature vectors
- information theory
- face recognition
- speech recognition technology
- speech recognition systems
- cost sensitive
- feature space
- multiresolution
- denoising
- maximum entropy
- wavelet decomposition
- subband
- feature set
- dimensionality reduction
- principal component analysis
- low level
- visual features
- speaker dependent
- power spectrum
- exponential family
- covariance matrix
- extracting features
- speaker recognition
- speaker independent
- log likelihood