Publications-Detail

Near-End Listening Enhancement by Noise-Inverse Speech Shaping

Authors:
Niermann, M.Jax, P.Vary, P.
Book Title:
Proceedings of European Signal Processing Conference (EUSIPCO)
Organization:
EURASIP
Pages:
p.p. 2390-2394
Date:
Aug. 2016
DOI:
10.1109/EUSIPCO.2016.7760677
Language:
English

Abstract

In communication systems, clean speech is often reproduced by loudspeakers and disturbed by local acoustical noise. Near-end listening enhancement (NELE) is a technique to enhance the speech intelligibility in environmental noise by adaptively preprocessing the speech based on a noise estimate. Conventional NELE-algorithms adaptively filter the speech by applying spectral gains which are determined by maximizing intelligibility measures. Usually, this leads to speech amplifications at highly disturbed frequencies to overcome masking. In this paper, a new approach is presented which shapes the speech spectrum according to the inverse of the noise power spectrum. It is based on a simple gain rule. Its advantages are a predictable spectral behavior and a fixed computational complexity, since no optimization problem with an unknown number of iterations needs to be solved. Simulations have shown that it copes with a wide range of noise types and provides a similar performance compared to conventional algorithms.

Download

BibTeX

Copyright © by IKS
niermann16b.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.