CST-UNet: Cross Swin Transformer Enhanced U-Net with Masked Bottleneck for Single-Channel Speech Enhancement.
Zipeng ZhangWei ChenWeiwei GuoYiming LiuJianhua YangHouguang LiuPublished in: Circuits Syst. Signal Process. (2024)
Keyphrases
- speech enhancement
- single channel
- multi channel
- independent component analysis
- wiener filter
- frequency domain
- noisy environments
- wavelet decomposition
- signal to noise ratio
- noise reduction
- prior information
- speech signal
- linear prediction
- brain activity
- visual information
- computer vision
- signal processing
- vocal tract
- multiscale