mdctGAN: Taming transformer-based GAN for speech super-resolution with Modified DCT spectra.
Chenhao ShuaiChaohua ShiLu GanHongqing LiuPublished in: INTERSPEECH (2023)
Keyphrases
- single image super resolution
- super resolution
- low resolution
- high resolution
- image patches
- image reconstruction
- image restoration
- low resolution images
- high quality
- image super resolution
- discrete cosine transform
- motion estimation
- high resolution images
- video super resolution
- spatial domain
- resolution enhancement
- super resolution algorithm
- principal component analysis
- super resolved
- single frame
- light field
- super resolution reconstruction
- higher resolution
- motion blur
- structuring elements
- multi frame
- speech signal
- multiresolution
- multiscale
- dct coefficients
- computer vision
- image compression
- feature vectors
- feature extraction