Statistical Speech Enhancement Based on Probabilistic Integration of Variational Autoencoder and Non-Negative Matrix Factorization.
Yoshiaki BandoMasato MimuraKatsutoshi ItoyamaKazuyoshi YoshiiTatsuya KawaharaPublished in: CoRR (2017)
Keyphrases
- negative matrix factorization
- speech enhancement
- matrix factorization
- principal component analysis
- noisy environments
- document clustering
- sparse representation
- noise reduction
- signal to noise ratio
- speech signal
- image segmentation
- single channel
- probabilistic model
- bayesian networks
- spectral clustering
- linear prediction
- pattern recognition
- wiener filter
- image processing
- data mining
- generative model
- dimensionality reduction