A Multi-scale Subconvolutional U-Net with Time-Frequency Attention Mechanism for Single Channel Speech Enhancement.
Sivaramakrishna YechuriThirupathi Rao KomatiRama Krishna YellapragadaSunnydayal VanambathinaPublished in: Circuits Syst. Signal Process. (2024)
Keyphrases
- single channel
- speech enhancement
- multiscale
- wavelet decomposition
- wavelet transform
- multi channel
- frequency domain
- signal processing
- edge detection
- independent component analysis
- scale space
- wiener filter
- multiresolution
- wavelet coefficients
- subband
- image representation
- natural images
- noisy environments
- prior information
- image compression
- wavelet domain
- image segmentation
- brain activity
- signal to noise ratio
- high frequency
- multi modal
- image processing
- spatial domain
- denoising
- motion estimation
- prior knowledge
- sound source
- pattern recognition
- face recognition