Semi-supervised Multichannel Speech Enhancement with Variational Autoencoders and Non-negative Matrix Factorization.
Simon LeglaiveLaurent GirinRadu HoraudPublished in: ICASSP (2019)
Keyphrases
- negative matrix factorization
- speech enhancement
- single channel
- semi supervised
- multi channel
- linear prediction
- denoising
- wiener filter
- semi supervised learning
- matrix factorization
- sparse representation
- principal component analysis
- nonnegative matrix factorization
- independent component analysis
- frequency domain
- pairwise
- noise reduction
- document clustering
- active learning
- labeled data
- signal to noise ratio
- noisy environments
- image segmentation
- image coding
- prior information
- unsupervised learning
- spectral clustering
- optical flow
- speech signal
- prediction error
- wavelet decomposition
- machine learning
- lossless compression
- signal processing
- neural network
- missing data
- image restoration
- generative model
- markov random field
- co occurrence
- object recognition
- pattern recognition
- brain activity
- computer vision
- vocal tract