Research areas Binaural Hearing / Acoustic Virtual Reality / Medizinische Akustik

Today’s hearing aids offer a wide variance of signal processing to improve hearing comfort and speech intelligibility. Examples are beam forming, noise reduction and dynamic compression. Further research and improvements are done with every new generation of hearing aids. Despite these efforts, there are still manifold situations where users of hearing aids are not satisfied with the results. One of the biggest factors in user satisfaction is the number of environments in which the hearing aid performs well. A survey shows a significant relationship between user satisfaction and the number of environments where consumers derive utility from their hearing aid. Accordingly, it is important to improve the behavior of hearing aids, especially in environments often encountered by hearing aid users.

Many hearing aid users complain on performance in classrooms, workplaces and restaurants. These environments have some acoustic similarities, mostly a large number of unwanted sources, like noise or talking people, in combination with a medium reverberation time, leading to a subjective loud, noisy environment which makes concentration on one respondent hard, even for healthy hearing persons. One of the difficulties in these environments, reducing speech intelligibility, is reverberation. Reverberation reduces speech intelligibility for everyone, based on the coloration of the received sound and the addition of various late reflections from the same source, but hearing aid wearers seem to suffer a lot more than people without hearing loss. One part of this problem is that many of the actually used signal processing algorithms tend to bad performance in reverberant environments or even start degenerating the perceived audio signal, making the situation even worse for the hearing aid wearer. Some promising enhancements, like some sorts of adaptive beam formers, are not even usable in hearing aids due to instabilities within reverberant situations.

The normal auditory system of a healthy listener automatically decomposes the incoming signal into groups of components representing the original sound sources. With hearing loss, however, this decomposition ability is often severely limited. With the advent of digital hearing aids and the use of multiple microphones, it is possible, in principle, to enhance the signal to noise ratio before the sound reaches a damaged ear, and a number of procedures has been proposed to reduce noise and suppress reverberation. Unfortunately most of these algorithms only perform satisfying in certain, often theoretical, environments, while showing no performance, or even degenerating the perceived signal, under reverberated and noisy circumstances.

One of the biggest problems when dealing with reverberant situations is, that the acoustic behavior is not well-known. Most algorithms make some assumptions on the acoustic field they work with. Usually that is free field, diffuse field or even a perfectly incoherent noise field. The quality of this assumption directly determines how good the algorithm works in realistic situations. The main drawback is that everyday situations can seldom be described by one of those ideal models, but are mostly a combination of different types of sound fields, varying from free-field conditions, outside, to diffuse field conditions, in reverberant rooms, and even encountering standing wave like conditions in small enclosures like cars. A good working noise reduction, de-reverberation or beam forming algorithm has thereby to be able to deal with all types of noise fields, either by knowing how the actual sound field is composed, or by an assumption that is general enough to deal with all situations.

For the development of such algorithms further knowledge about the sound field in everyday environments has to be gained. Therefore it is necessary to identify situations hearing aid users encounter, examine the sound field and find parameters that describe this sound field. Based on those sound field identifiers, a sound field classification can be developed, which gives a guideline which signal processing methods will perform as expected and which will cause problems. With this sound field classification a hearing aid has a tool for determining which parts of the signal processing chain to switch on or off, or even decide between several algorithms, choosing the one that performs best in the current situation.