Mitigating wind noise in outdoor microphone signals using a singular spectral subspace method

Eldwaik, O and Li, FF 2017, Mitigating wind noise in outdoor microphone signals using a singular spectral subspace method , in: IEEE, Seventh International Conference on Innovative Computing Technology (INTECH 2107), 16-18 August 2017, Luton, UK.

[img] PDF - Accepted Version
Restricted to Repository staff only

Download (658kB) | Request a copy

Abstract

Wind noise is one of the major concerns of outdoor microphone signal acquisition. Filtering and removal of wind noise are known to be difficult due to its broadband and time varying nature. This paper proposes the use of singular spectrum analysis to address the problem of microphone wind noise removal and/or separation. The paper is presented from the context of reducing microphone wind noise when deploying outdoor acoustic sensing in smart city applications and soundscapes monitoring. But concepts and methods can be generalized beyond its original scope. The method includes two complementary stages, namely decomposition and reconstruction. The first stage decomposes mixed signals in eigen-subspaces, selects and groups the principal components according to their contributions to wind noise and wanted signals in the singular spectrum domain. The second stage is used to reconstruct the signals back to the time domain, resulting in the separation of wind noise and wanted signals. Following a brief review of the wind noise removal problem, this paper presents the algorithm and some experimental results, and discusses the potentials of the singular spectrum analysis for microphone wind noise reduction.

Item Type: Conference or Workshop Item (Paper)
Schools: Schools > School of Computing, Science and Engineering
Publisher: IEEE
Depositing User: Omar Eldwaik
Date Deposited: 14 Aug 2017 13:02
Last Modified: 14 Aug 2017 13:12
URI: http://usir.salford.ac.uk/id/eprint/43527

Actions (login required)

Edit record (repository staff only) Edit record (repository staff only)

Downloads

Downloads per month over past year