Conditional MMSE-based single-channel speech enhancement using inter-frame and inter-band correlations.
Hajar MomeniHamid Reza AbutalebiEmanuël A. P. HabetsPublished in: ICASSP (2016)
Keyphrases
- inter frame
- single channel
- speech enhancement
- multi channel
- motion estimation
- video sequences
- bit rate
- frequency domain
- independent component analysis
- prior information
- video frames
- wiener filter
- quadtree
- noisy environments
- lossless compression
- wavelet decomposition
- noise reduction
- prediction error
- signal processing
- speech signal
- signal to noise ratio
- sound source
- computer vision
- synthetic and real images
- visual quality
- visual information
- correlation analysis
- image coding
- motion vectors
- image restoration
- image quality
- multiscale