Bayesian Separation With Sparsity Promotion in Perceptual Wavelet Domain for Speech Enhancement and Hybrid Speech Recognition.
Yu ShaoChip-Hong ChangPublished in: IEEE Trans. Syst. Man Cybern. Part A (2011)
Keyphrases
- wavelet domain
- speech recognition
- speech enhancement
- speech signal
- noisy environments
- human visual system
- noisy speech
- spatial domain
- wavelet coefficients
- wavelet transform
- wavelet decomposition
- denoising
- hidden markov models
- vocal tract
- multiresolution
- automatic speech recognition
- multiscale
- speech synthesis
- subband
- single channel
- sound source
- additive noise
- language model
- pattern recognition
- image quality
- linear prediction
- high frequency
- noise reduction
- image denoising
- signal to noise ratio
- sar images
- visual quality
- visual information
- maximum likelihood
- background noise
- frequency domain
- speaker identification
- image compression
- noisy images
- neural network
- low level
- bayesian networks
- non stationary
- bit rate
- higher order
- motion estimation
- smoothing algorithm