Single-channel speech enhancement using improved progressive deep neural network and masking-based harmonic regeneration.
HUANG PingYafeng WuPublished in: Speech Commun. (2022)
Keyphrases
- single channel
- speech enhancement
- neural network
- multi channel
- noise reduction
- independent component analysis
- noisy environments
- prior information
- signal to noise ratio
- pattern recognition
- brain activity
- wavelet decomposition
- frequency domain
- multiscale
- speech signal
- high frequency
- wiener filter
- linear prediction
- background noise
- additive noise
- optical flow
- principal component analysis
- image segmentation