Noise reduction: the difference between ‘noise haters’ and ‘distortion haters’

Clinical Physicist Audiologist (i.o.) Rolph Houben of the Academical Medical Centre (AMC) Amsterdam (The Netherlands) investigates what noise reduction really does in the patient. Step by step he unravels the mystery of noise reduction.

At the Department of Clinical and Experimental Audiology of the AMC, Rolph Houben, together with Inge Brons Msc., performs clinical research into noise reduction on hearing aids. He explains: “In our clinical setting the users of noise reduction play the central role. How does the hard of hearing benefit from (single channel) noise reduction on a hearing aid? Does it work for him, is there a functional edge? These are the questions I’d like to answer since I started this research in 2009. I studied physics and we collaborate with electronics specialists at the Radboud Universiteit Nijmegen and Eindhoven Technical University (TU/e) , but we don’t develop algorithms for noise reduction ourselves.”

Every modern hearing aid uses two technologies: compression and noise reduction. “Remarkably, there is a lot of evaluating research into compression, but hardly into noise reduction. There are reasons for that. Firstly, noise reduction has no direct influence on understanding of speech. Secondly, noise reduction is a complex phenomenon, a ‘black box’. It is unknown what is actually in it. It is also remarkable that hearing aids have noise reduction, but that they often can hardly be fine tuned. That is remarkable, because noise reduction can theoretically be tuned in many ways. We know little of it and therefore can adjust little to it. Why is that? Isn’t it necessary to adjust it to individual needs and preferences?”

Less effort

Opening the black box proved to be far from easy. “You wish to know generally how manufacturers approach the subject and you have to find out through reverse engineering. You can start out by measuring in the device what it does and by studying the available literature to establish how it roughly works. What was established subsequently in international research, was that noise reduction does not enhance understanding of speech. It also became apparent that patients at the same time often do value the noise reduction on their hearing aid. This leads to new questions: why do people value something that doesn’t lead to enhanced understanding of speech? And is the noise reduction of one brand better than the noise reduction of another brand?”

The answer to the first question could be deduced relatively easily. “Complaint number one and important reason for purchase of a hearing aid is understanding of speech in background noise”, Houben establishes. “That might not be improved by noise reduction, but the necessary effort to reach a certain level of understanding of speech proves to be less, thanks to noise reduction. This offers a plausible explanation to why people value noise reduction: it is less tiresome.”


Then the second question: how good are the various noise reduction systems really? “To find the answer to this question we choose to work independently from sponsoring by manufacturers. We wish to avoid any suggestion that sponsoring would influence our scientific results. Given the fact that listening effort is a measure for the effectiveness of noise reduction, we can deduce the performance of noise reduction from a good listening effort test”, Houben tells.

Houben and Brons perform a promising test on a group level on the basis of the numbers test that was developed at the Vrije Universiteit (VU) . It proves that the response time of test persons is enhanced by noise reduction. Listening effort is traditionally a speciality of that other Amsterdam University, the VU. “Correct”, says Houben, “but their angle is different. They approach listening effort from the physiological and psychological angle, for instance by measuring the pupil. For us the most important thing is to establish what noise reduction technically does, with the goal to eventually enhance the user experience. That is a difficult task, because matter of factly there is not only one black box, but there are two black boxes. We already established that we don’t exactly know what is inside the devices. Apart from that, we don’t know how the patient processes it either. So first you meet even more questions; subsequently you can start solving them.”

Effect established

Houben and Brons took a number of hearing aids, fitted them and made recordings of the output to an artificial head. They performed this measurement twice: with and without noise reduction turned on. Houben: “This difference between noise reduction switched on and switched off enabled a comparison within the same device, but of course not a comparison between various devices. How can you then find out which noise reduction works the best? That is an all important question, as the noise reduction for the hard of hearing can be a factor of choice for a certain hearing aid.” Houben filtered the recordings in such a way that all devices tested without noise reduction sounded the same, fitted according to the standard of how they should sound. This was then clinically verified with the question whether tested persons could hear differences. After this, the noise reduction was switched on in all devices. “This enables a direct comparison of the noise reduction between various hearing aids”, says Houben. “It then appears that people prefer some systems over others. This sounds quite logical, but in fact this difference in evaluation enables an important conclusion. For the first time it is clinically established that the effect of noise reduction on listening effort does indeed differ between different hearing aids.”

Personal preferences

In the AMC research normal hearing as well as hard of hearing people participate. Analysis of the research data of normal hearing tested persons learned, that almost everyone had certain preferences. Generally, it is only possible to diminish noise at the cost of the quality of the speech signal. Some tested persons based their opinion on the amount of noise the system was able to reduce. Others rather focused more on the amount of distortion that was introduced by a certain noise reduction system. It appears you have ‘noise haters’ and ‘distortion haters’. Just like with for instance the choice of HI-FI systems, individual preferences play an important role in this. “We have not yet been able to establish whether these differences are clinically relevant. The outcomes do however explain why it is so difficult to perform this type of research”, says Houben, “The sound coloration of a hearing aid is a dominant factor. But when you isolate the noise reduction as a single factor, the differences between noise reduction algorithms appear.”

One step beyond from there: what are these differences and are these differences dependent on the situation? Houben: “For normal hearing people the diminishing of the listening effort proves to be dependent on the type of background noise. Whether these results are applicable for the hard of hearing and on the individual level is as yet unclear. We proceed in our search and expect that there is a lot to learn about the application of noise reduction in hearing aids.”

Leendert van der Ent, editor in chief of Audiology Infos North of Europe.

Photo: © Bureau Lorient Communicatie