Speech-in-Noise, Frequency Lowering, and More: Interview with Joshua M. Alexander, PhD
Douglas L. Beck, AuD, spoke with Dr. Alexander about frequency transposition, frequency compression, and more.
Douglas L. Beck (DLB): Good morning, Josh. It’s great to speak with you. I’ve been reading your publications for a few years and I’m surprised we haven’t met.
Alexander: Hi, Doug. Thanks…good to speak with you, too.
DLB: For the folks less familiar with your work, I should note you worked at Boys Town National Research Hospital for a number of years before joining Purdue, and you’ve been at Purdue for some five years.
Alexander: Exactly. I was at Boys Town from 2007 to 2009 and at that time, I had just finished my post-doc in speech perception at Wisconsin. My post-doc focused on questions and issues related to what does the auditory system need to maximize speech understanding.
DLB: And that brings us to the topic of the day, which is frequency lowering. In your article in Seminars in Hearing (May 2013) you pointed out there are a number of frequency lowering (FL) technologies, and as of yet, we have no outcomes-based, scientifically proven protocols to identify who to fit with which frequency-lowering technology. Is that still true today?
Alexander: Yes. We’ve got more FL technology than ever, but we don’t have the research to show who to apply which technology to and we have no direct “head-to-head” comparisons to indicate which option is best for a given patient.
Ideally, we should have replicable studies, based on control and experimental groups and/or random controlled trials (RCTs) and other scientifically proven results before we fit anyone. However, we’re really in a difficult position because hearing aid technology often proceeds by leaps and bounds before the actual research and validation studies. For example, we have had multichannel wide dynamic range compression available in hearing aids for about 20 years, yet questions remain as to which compression speed and number of channels is best for different individuals. Furthermore, as a field, we are still looking to find the true benefits of digital noise reduction, which patients seem to love, and data supporting conventional adaptive directionality remains elusive.
Nonetheless, to be clear, I feel there may be a place for FL technology, but we need to understand what the pros and cons are because unlike the examples I just mentioned, FL is not just another hearing aid feature. With FL, we have the ability to fundamentally change the cues people use to understand speech. That is, we have to be mindful that we don’t know what the outcome is going to be for a particular child or adult, and to just use the FL default programs as if they were a proven protocol, or to apply these technologies to every child or adult, is not a solid or rational plan!
DLB: Yes, it seems we need more outcomes-based evidence before we accept FL as the “go-to” or the default program.
Alexander: That would be great, but I am doubtful that just one study would answer all of our questions. One major reason for this viewpoint is that there are not standardized methods or even agreed upon goals for fitting this technology. Therefore, any evidence-based review of the literature should consider that subjects between studies and even within the same study likely did not receive the same “treatment.” This makes it difficult to know which technology or settings may or may not work for different configurations of hearing loss and which factors (e.g., etiology, duration of loss, cognitive processing) mediate the effects. On the other hand, positive or negative findings arising from an ideal study that controls for all of these factors may not be generalizable to other sub-populations based on configuration of loss, age, etc. and they will certainly not be generalizable to other methods of FL.
DLB So to be clear, is it true that each of the manufacturers implements their FL technology in a unique way?
Alexander: Yes, that’s a fair statement. Of the six manufacturers who supply commercially available FL technologies, there are vast differences in how they approach FL. Some use transposition, some use compression, and multiple versions and protocols of each are used, and we simply need to understand them before we apply them, as frequency compression by company A is not the same as frequency compression by company B…and so we should expect the outcomes would be different, too. Furthermore, while the terminology is sometimes similar, you cannot take the settings from one manufacturer and apply it to another.
Given these differences, the burden is placed on the clinician to understand a particular FL technology before applying it. I recognize, too, that it is sometimes difficult to get accurate information from the manufacturers or hearing aid analyzers about what the FL techniques are doing to the signal. To help the clinician understand these differences and to understand how they may influence the sounds an individual patient may hear, I have created a set of very popular online tools that demonstrate how the frequencies of sounds are remapped when changing the FL settings for the different manufacturers. These tools can be accessed on your desktop computer or mobile device by visiting the Pardue Ear Lab page on Frequency Lowering Fitting Assistants.
DLB: One of the major questions associated with FL technology, is the issue of what happens to the speech sounds that previously resided in the spectral area, where the new FL information is placed? Let’s call that the FL receptor zone, and define it as the spectral zone, which will receive the new FL information.
Alexander: That’s a great question, and it’s really an important issue. First, if we think of FL as a treatment, then for each patient we should have a clear idea about what it is that we are trying to treat (e.g., detection and discrimination of high-frequency speech cues for the fricatives and maybe some of the stop consonants). Second, like a good physician, we should have an idea about the treatment mechanism and about how it is supposed to help the symptoms. Third, and most importantly, we should have clear ideas about the potential side effects related to the mechanism. This is where my online tools can help.
Side effects from FL arise because at least one of two things may happen to the information in the receptor zone. With transposition/translation techniques, information that is already in the receptor zone may be masked by the overlapping of re-coded high-frequency energy. Of course, generally speaking, lows mask highs, but the relative loudness levels matter! With compression techniques, the information in the receptor zone has to be “squeezed” to make room for the re-coded high-frequency information. If the frequency relationships of the original information in the receptor zone matter to perception (e.g., primary formants of vowels and consonant transitions), then distortion of important speech cues may result. However, if the receptor zone is beyond the primary formants of speech and the information contained therein is broaDLBand and/or temporally based, then side effects should be much less.
Another concern is what happens when background noise undergoes FL, too. That is, if the listener is in a noisy environment, and the noise is overpowering the primary speech signal, wouldn’t the noise also undergo FL, which would eliminate the potential benefit? And is it the same situation across all background noises, like multi-talker babble versus steady state noise? Of course, this concern could be partially alleviated by having a detector that prevents FL from activating if the signal-to-noise ratio isn’t very good. That would be interesting, but I’ve not read or heard anything about that. Therefore, I wonder if the overall goal of amplification is to obtain the best possible signal-to-noise ratio (SNR), are we shooting ourselves in the foot via FL, when we have other technologies such as FM and remote microphones and more, which readily achieve that same goal?
DLB: Right…good point. I recall the study by Harvey Abrams where he and his colleagues in Florida found when veteran candidates for cochlear implants were offered a reasonably good FM system, they were quite happy with the FM system as the signal-to-noise ratio improved enough such that they were able to make sense of the sounds they wanted to attend to.
Alexander: Exactly…I believe that Abrams et al study was some 8 to 10 years ago, and it’s worth noting the FM systems available today are much improved – but yes, that’s the point. FL might offer an advantage if the patient can learn to use the FL information, but we have other solutions which we know work, and it might be worth exploring all those options, given the exact needs of a specific patient.
DLB: These are all good questions…and my question, which I’ll ask, too, is what happens to the spatial cues which were originally available from perhaps 3000 to 8000 Hz? Are the intensity differences still there, or are they eliminated via WDRC and FL?
Alexander: That’s a good point, and that’s a primary psychoacoustics concern, and I suspect the answer is different for each of the FL techniques—but I really don’t know. However, I see a bigger point to your question, which has to do with the effects of FL on other metrics, besides speech recognition. Spatial processing, as you mentioned, is one such metric, and sound quality and listening effort/fatigue are others. There is laboratory data on hearing-impaired and/or normal-hearing adults that indicates that all of these other metrics improve to some extent as high-frequency audibility is increased by simply extending the bandwidth. It is somewhat a matter of speculation at this point whether the same benefits might be had by using FL, or as you rightfully mention, whether certain FL techniques or settings make them worse. Again, this is the type of basic research questions we need to ask and try to answer before we indiscriminately apply these technologies!
DLB: This really is a fascinating topic, and I’m very interested to see where all this will go. It’s been great speaking with you and I am very appreciative of your time and knowledge!
Alexander: Thanks, Doug. I appreciate your interest in the work we’re doing!
Joshua M. Alexander, PhD, is an assistant professor, Department of Speech, Language and Hearing Sciences, at Purdue University.
Douglas L. Beck, AuD, Board Certified in Audiology, is the Web content editor for the American Academy of Audiology and the director of public relations with Oticon, Inc.