Semi-Blind speech enhancement basedon recurrent neural network for source separation and dereverberation.
Masaya WakeYoshiaki BandoMasato MimuraKatsutoshi ItoyamaKazuyoshi YoshiiTatsuya KawaharaPublished in: MLSP (2017)
Keyphrases
- recurrent neural networks
- single channel
- source separation
- speech enhancement
- multi channel
- independent component analysis
- feed forward
- neural network
- artificial neural networks
- frequency domain
- wiener filter
- blind source separation
- prior information
- speech signal
- noise reduction
- brain activity
- wavelet decomposition
- signal processing
- machine learning
- multi modal
- multimedia