Masking and Inpainting: A Two-Stage Speech Enhancement Approach for Low SNR and Non-Stationary Noise.
Xiang HaoXiangdong SuShixue WenZhiyu WangYiqian PanFeilong BaoWei ChenPublished in: ICASSP (2020)
Keyphrases
- non stationary
- low snr
- speech enhancement
- speech signal
- noisy environments
- noise model
- noise reduction
- low signal to noise ratio
- signal to noise ratio
- linear prediction
- autoregressive
- vocal tract
- image restoration
- background noise
- single channel
- random fields
- automatic speech recognition
- human visual system
- denoising
- wiener filter
- empirical mode decomposition
- smoothing algorithm
- pattern recognition
- gaussian noise
- noise level
- noisy images