Single-channel speech separation with memory-enhanced recurrent neural networks.
Felix WeningerFlorian EybenBjörn W. SchullerPublished in: ICASSP (2014)
Keyphrases
- recurrent neural networks
- single channel
- speech enhancement
- multi channel
- sound source
- speech signal
- neural network
- independent component analysis
- feed forward
- recurrent networks
- audio visual
- frequency domain
- echo state networks
- speech recognition
- noisy environments
- artificial neural networks
- signal to noise ratio
- wavelet decomposition
- noise reduction
- brain activity
- prior information
- wiener filter
- vocal tract
- signal processing
- automatic speech recognition
- non stationary
- image segmentation