Modulation-domain parametric multichannel kalman filtering for speech enhancement
File(s)
Accepted version
Author(s)
Xue, Wei
Moore, Alastair H
Brookes, Mike
Naylor, Patrick A
Type
Conference Paper
Abstract
The goal of speech enhancement is to reduce the noise signal while keeping the speech signal undistorted. Recently we developed the multichannel Kalman filtering (MKF) for speech enhancement, in which the temporal evolution of the speech signal and the spatial correlation between multichannel observations are jointly exploited to estimate the clean signal. In this paper, we extend the previous work to derive a parametric MKF (PMKF), which incorporates a controlling factor to achieve the trade-off between the speech distortion and noise reduction. The controlling factor weights between the speech distortion and noise reduction related terms in the cost function of PMKF, and based on the minimum mean squared error (MMSE) criterion, the optimal PMKF gain is derived. We analyse the performance of the proposed PMKF and show the differences with the speech distortion weighted multichannel Wiener filter (SDW-MWF). We conduct experiments in different noisy conditions to evaluate the impact of the controlling factor on the noise reduction performance, and the results demonstrate the effectiveness of the proposed method.
Date Issued
2018-12-03
Date Acceptance
2018-09-03
Citation
2018 26th European Signal Processing Conference (EUSIPCO), 2018, pp.2509-2513
ISBN
9789082797015
ISSN
2076-1465
Publisher
IEEE
Start Page
2509
End Page
2513
Journal / Book Title
2018 26th European Signal Processing Conference (EUSIPCO)
Copyright Statement
© 2018, IEEE.
Sponsor
Engineering & Physical Science Research Council (EPSRC)
Identifier
http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=LinksAMR&KeyUT=WOS:000455614900504&DestLinkType=FullRecord&DestApp=ALL_WOS&UsrCustomerID=1ba7043ffcc86c417c072aa74d649202
Grant Number
EP/M026698/1
Source
European Signal Processing Conference (EUSIPCO)
Subjects
Science & Technology
Technology
Engineering, Electrical & Electronic
Engineering
Speech enhancement
Microphone arrays
Kalman filtering
Modulation domain
MVDR BEAMFORMER
Publication Status
Published
Start Date
2018-09-03
Finish Date
2018-09-07
Coverage Spatial
Rome, Italy
Date Publish Online
2018-12-03