Publications-Detail

Wind Noise Detection: Signal Processing Concepts for Speech Communication

Authors:
Nelke, C. M.Jax, P.Vary, P.
Book Title:
Proceedings of German Annual Conference on Acoustics (DAGA)
Organization:
Deutsche Gesellschaft für Akustik (DEGA)
Date:
Mar. 2016
ISBN:
978-3-93929-610-2
Language:
English

Abstract

The acoustic signal induced by wind during speech recordings can be a severe problem, e.g., for mobile phones, video recordings or hearing aids. Due to the dimension and design constraints, many devices for these applications do not offer space for the use of mechanical windscreens. Therefore, it is necessary to combat the acoustic noise in the captured signal by digital signal processing techniques. The first step towards the reduction of an undesired noise component in a speech signal is a detection of segments with wind noise activity. The detector must be capable of adapting quickly to the non-stationary signal characteristics of wind noise.
In this paper three new wind detection concepts are presented. The first algorithm is based on the short-term mean introduced by wind noise in a recorded signal while the characteristic spectral shape and energy distribution of wind noise are exploited in the second and third approach. All three proposed methods are compared with known approaches from literature in terms of their accuracy using real wind noise recordings. All considered algorithms are implemented as real-time processing schemes working on the short-term spectrum of signal frames of 20,ms. This is realized by an overlap-add structure, which is widely used for digital speech processing procedures.

Download

BibTeX

Copyright © by IKS
nelke16.pdf
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.