Unsupervised noise removal technique based on constrained NMF.
IET Signal Processing(2017)
摘要
This paper presents a novel approach to mitigate background interference noise in a noisy audio environment based on unsupervised speech enhancement without the prior knowledge of speech as well as noise signal. In this scheme, a noisy speech signal is represented by a non-negative matrix factorisation (NMF) with sparseness constraints. Using NMF with constraints, noise density spectrum is estimat...
更多查看译文
关键词
correlation methods,interference suppression,matrix decomposition,signal denoising,spectral analysis,speech enhancement,speech intelligibility
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要