Low-Power Convolutional Recurrent Neural Network For Monaural Speech Enhancement.
Fei GaoHaixin GuanPublished in: APSIPA ASC (2021)
Keyphrases
- low power
- recurrent neural networks
- speech enhancement
- noise reduction
- noisy environments
- signal to noise ratio
- low cost
- power consumption
- high speed
- single channel
- speech signal
- linear prediction
- neural network
- recurrent networks
- feed forward
- complex valued
- smoothing algorithm
- artificial neural networks
- network architecture
- vocal tract
- wiener filter
- multi channel
- hidden layer
- background noise
- speech recognition
- edge detection
- real time
- power dissipation
- back propagation
- logic circuits
- additive noise
- compression algorithm
- image coding
- markov random field
- denoising
- mixed signal
- genetic algorithm
- ultra low power