Semi-supervised multichannel speech enhancement with variational autoencoders and non-negative matrix factorization.
Simon LeglaiveLaurent GirinRadu HoraudPublished in: CoRR (2018)
Keyphrases
- negative matrix factorization
- speech enhancement
- single channel
- semi supervised
- multi channel
- linear prediction
- denoising
- semi supervised learning
- wiener filter
- principal component analysis
- independent component analysis
- matrix factorization
- frequency domain
- document clustering
- nonnegative matrix factorization
- labeled data
- sparse representation
- pairwise
- image segmentation
- prior information
- noisy environments
- noise reduction
- speech signal
- optical flow
- signal to noise ratio
- active learning
- signal processing
- wavelet decomposition
- image coding
- missing data
- unsupervised learning
- spectral clustering
- image classification
- lossless compression
- non stationary
- sound source
- pattern recognition
- feature extraction
- machine learning