Publications – Details
Speech Dereverberation for Hearing Aids with a Binaural Data Link
- Authors:
- Heinrich W. Löllmann, Marco Jeub, and Peter Vary
- Book Title:
- Diskussionssitzung "Binaurales Hören mit Hörgeräten und Cochlearimplantaten" im Rahmen der 3 Ländertagung der ÖGMP, DGMP und SGSMP
- Event Date:
- 28.-29.9.2011
- Organization:
- DGMP
- Location:
- Wien
- Date:
- Sept. 2011
- Note:
- http://www.medphyswien2011.org/
- Language:
- English
Abstract
Persons with hearing aids often experience a substantial loss of listening comfort or speech intelligibility, if they communicate with other persons in noisy or reverberant environments.
Therefore, speech enhancement algorithms are a common feature of modern hearing aids. A
recent trend is thereby to establish a binaural data link between both hearing aid devices in
order to improve the enhancement of acoustically distorted speech signals. While most
algorithms for binaural speech enhancement aim for the enhancement of noisy speech, the
enhancement of reverberant speech is still in an early stage.
In this contribution, we present a two-stage approach for binaural speech dereverberation in
hearing aids, which achieves a suppression of early and late reverberant speech. An important
feature is that the overall binaural processing does not affect the most important binaural cues,
i.e., the interaural time difference (ITD) and the interaural level difference (ILD). This is
important for speech enhancement in hearing aids to preserve the ability for source
localization in the azimuth plane.
The first stage of the algorithm is based on a spectral subtraction where the weights depend on
the spectral variance of the late reverberant speech. The computation of this spectral variance
in turn requires an estimate of the reverberation time. This is accomplished by an efficient
algorithm, which is based on a maximum likelihood (ML) estimation. The output of the first
stage is further enhanced by a two-channel Wiener filter in a second step. This filter is derived
by a coherence model which takes the shadowing effects of the head into account.
The presented algorithms works blindly, i.e., all needed quantities are estimated from the
reverberant speech so that no a priori information about the acoustical environment is
required. Experiments have shown that the proposed system achieves a significant reduction
of early and late reverberation.
Download of Publication
Copyright Notice
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.
The following notice applies to all IEEE publications:
© IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.