PhD defence by Niels Overby

On Tuesday 16 January, Niels Overby will defend his PhD thesis "Scene-aware compensation strategies for hearing aids in adverse conditions".

Time: 14:00

Place: Building 303A, auditorium 41 & zoom: https://dtudk.zoom.us/j/69885577630?pwd=WmF4eHk3WUVVbXVLcW1CSmpKQ1NoZz09

Please be aware that the PhD defense may be recorded - This will also be informed at the beginning of the PhD defense.

Supervisor: Associate Professor Tobias May

Co-supervisor: Professor Torsten Dau

Assessment committee:
Associate Professor Bastian Epp, DTU Health Tech

Professor Pamela E. Souza, Department of Communication Sciences and Disorders, Northwestern University

Professor Volker Hohmann, Carl von Ossietzky Universität Oldenburg

Chairperson:
Associate Professor Jeremy Marozeau, DTU Health Tech

Abstract:
People with hearing impairment can experience quiet sounds as inaudible, while loud sounds might still appear loud. Because of this, hearing aids adapts their amplification such that quiet sounds are increasingly amplified without amplifying loud sounds to an uncomfortable level. The adaptive type of amplification is known as dynamic range compression and is considered a fundamental building block of modern hearing aids.

Dynamic range compression can operate as either fast-acting or slow-acting depending on the rate at which it reacts to changes in signal level.  Fast-acting compression can amplify individual components of a speech signal, thus providing improved audibility, however, this might also amplify background noise and room reverberation, which can appear disturbing to the listener. Slow-acting compression ensures the overall listening level is pleasant but cannot provide the same short-term audibility as fast-acting compression.

As an alternative, scene-aware compression is an adaptive compression method that applies fast-acting compression in the presence of speech and switches to slow-acting compression in the absence of speech.

Speech enhancement is another building block commonly used in hearing aids that aims to attenuate background noise while preserving the target speech signal. However, speech enhancement can unintendingly attenuate the soft speech components that the compression aimed to amplify, and compression might also amplify the noise which the speech enhancement system attenuated. To prevent some of these opposing interactions, the scene-aware compressor can be configured to be controlled by the speech enhancement system.

In this thesis, several combinations of compression and speech enhancement systems are evaluated. In the first part, only physical metrics are used which can describe the overall effects of the hearing aid signal processing such that each system can be compared. The second part focuses on the perceived effects of the signal processing, where two experiments are presented. The first evaluates how each signal processing method affects the overall preference, and the second experiments aims to analyze how the physical metrics relates to how a hearing aid is perceived.

The overall findings of this thesis explore the possibilities to develop and evaluate hearing aid signal processing using physical metrics and be less reliant on time-consuming listening tests.