Rethinking Complex-Valued Deep Neural Networks for Monaural Speech Enhancement.
Haibin WuKe TanBuye XuAnurag KumarDaniel WongPublished in: INTERSPEECH (2023)
Keyphrases
- complex valued
- speech enhancement
- neural network
- activation function
- recurrent neural networks
- real valued
- noisy environments
- noise reduction
- signal to noise ratio
- artificial neural networks
- feed forward
- speech signal
- single channel
- neural nets
- linear prediction
- back propagation
- hidden layer
- linear equations
- cellular neural networks
- background noise
- network architecture
- learning rate
- vocal tract
- wiener filter
- multi layer perceptron
- radial basis function
- genetic algorithm
- multi valued
- multi channel
- independent component analysis
- speech recognition
- image restoration
- signal processing
- multiresolution
- pattern recognition
- smoothing algorithm