2021/09/05 by Guochen Yu, Yutian Wang, Yu, Guochen +7 · 5 citations
Computer Science · Engineering · Neuroscience · #Algorithm #Artificial Intelligence (cs.AI) #Artificial intelligence #Artificial neural network #Audio and Speech Processing (eess.AS) #Complex system #Computer science #Convolution (computer science) #Deconvolution #FOS: Computer and information sciences #FOS: Electrical engineering #Generative grammar #Hearing Loss and Rehabilitation #Magnitude (astronomy) #Music and Audio Processing #Noise (video) #Noise reduction #Pattern recognition (psychology) #Physics #Residual #Sound (cs.SD) #Speech and Audio Processing #Speech recognition #cs.AI #cs.SD #eess.AS #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2109.02011
published in arXiv (Cornell University) (Cornell University) · Accepted by Speech Communication
arxiv created 2021/09/05 · openalex publication_date 2021/09/05 · arxiv updated 2021/09/07 · openalex created_date 2021/09/13 · openalex updated_date 2026/07/28
Cycle-consistent generative adversarial networks (CycleGAN) have shown their promising performance for speech enhancement (SE), while one intractable shortcoming of these CycleGAN-based SE systems is that the noise components propagate throughout the cycle and cannot be completely eliminated. Additionally, conventional CycleGAN-based SE systems only estimate the spectral magnitude, while the phase is unaltered. Motivated by the multi-stage learning concept, we propose a novel two-stage denoising system that combines a CycleGAN-based magnitude enhancing network and a subsequent complex spectral refining network in this paper. Specifically, in the first stage, a CycleGAN-based model is responsible for only estimating magnitude, which is subsequently coupled with the original noisy phase to obtain a coarsely enhanced complex spectrum. After that, the second stage is applied to further suppress the residual noise components and estimate the clean phase by a complex spectral mapping network, which is a pure complex-valued network composed of complex 2D convolution/deconvolution and complex temporal-frequency attention blocks. Experimental results on two public datasets demonstrate that the proposed approach consistently surpasses previous one-stage CycleGANs and other state-of-the-art SE systems in terms of various evaluation metrics, especially in background noise suppression.