On single-channel noise reduction in the time domain

In this paper, we revisit the noise-reduction problem in the time domain and present a way to decompose the ltered speech into two uncorrelated (orthogonal) components: the desired speech and the interference. Based on this new decomposition, we discuss how to form different optimization cost functions and address the issue of how to design different noise-reduction lters by optimizing these new cost functions. Particularly, we cover the design of the maximum signal-to-noise-ratio (SNR), the Wiener, the minimum variance distortionless response (MVDR), and the tradeoff lters. It is interesting that with this new decomposition, we can now design the MVDR lter that can achieve noise reduction without adding speech distortion in the single-channel case, which has never been seen before. We also demonstrate that the maximum SNR, Wiener, and tradeoff lters are identical to the MVDR lter up to a scaling factor. From a theoretical point of view, this scaling factor is not signi cant and shoul...

• Type: Journal
• Year: 2011
• Where: ICASSP
• Authors: Jingdong Chen, Jacob Benesty, Yiteng Huang, Tomas Gänsler
• Keywords: ICASSP 2011, MVDR Lter, Scaling Factor, Signal Processing, Tradeoff Lters