Roundtable Discussion: Directional Hearing Aids

Do directional microphones benefit individuals with normal hearing to 1500 Hz and severe loss in the high frequencies?

Todd: It is possible to provide good directionality at high frequencies with some hearing aids; thus if an individual is capable of using high frequency information, then he or she can get directional benefit. But if the hearing aid only has good low frequency directivity then clearly there would be no advantage. Data suggest the magnitude of average benefit may be predictable from Speech Intelligibility Index (SII) calculations.

Which is better: single- or dual-microphone directionality?

Ruth: The answer to this is highly dependent upon how well the microphones in a dual-microphone system are matched. If the microphones are not matched well, they will act like an omnidirectional microphone. However, another big issue affecting directionality is debris in the microphone port. In a multisite hearing aid study, we fitted people with a dual microphone system on one ear and a single microphone design on the other. When the data were examined at 6 months and 12 months postfitting, we found that debris plugging the ports had the biggest impact upon the effectiveness of the directional microphone systems. To combat this some companies have made caps that will allow for easy cleaning of the port; however, these have unfortunately removed most of the benefit of the directional microphone. Todd: We encountered the same issue with debris. But of course this will equally affect single- and dual-microphone systems. One advantage of some dual-microphone systems is that they can equalize the response of the two microphones, potentially offsetting minor problems with microphone mismatch (due to dirt, age, etc.). A second advantage of a dual-microphone system is that it makes adaptive directionality possible. But in a fixed directional mode, there does not currently seem to be a strong argument for selecting a dual microphone over a single microphone. I acknowledge that it is easier to tweak a dual-microphone system to maximize directivity on the head, but the theoretical maximum of first-order systems is the same as in the single-microphone design.

How do you explain the fact that sometimes patients show little or no directional benefit on a test of speech in noise and yet you know that the hearing aids and directional microphones are working?

Brian: Indeed we have encountered this in some of our studies. Some people don’t get a directional advantage in the test booth despite directional microphones that are working. Further, these patients can end up being successful users of directional microphones in everyday living! In this regard, you might want to look at a paper from our laboratory by Mary Cord and associates [Editor’s note: Journal of the American Academy of Audiology, 2004;15:353-364]. In that study, she compared the mean directional advantage obtained by patients who regularly used their directional microphone option and patients who never switched from the omnidirectional mode. There was no significant difference in the mean directional advantage obtained by the successful and unsuccessful users of directional technology. Some patients who obtained a large directional advantage in the test booth never used their directional microphones in everyday living. Further, some patients who failed to obtain a directional advantage in the test booth regularly used their directional microphones in everyday living. I can’t tell you exactly why some patients don’t seem to make very good use of the signal processing provided by directional microphones. Whether or not the directional microphones are working optimally, whether or not the patient obtains a directional advantage in the test booth, and whether or not the patient will be successful with directional microphones in everyday listening surely are not unrelated. However, they may not be as closely related as our intuition might lead us to expect. Todd: Some of it comes down to the fact that even when we spend a lot of time and energy trying to replicate a real-world situation in the laboratory, it still isn’t the same as being in the real world. There are two issues: First, does the patient ever actually encounter the listening situation you simulated in the laboratory? And second, do patients behave the same in both places? For instance, we had a patient who consistently showed amazing benefit in our simulated real-world condition, but he never uses his directional setting in the real world. He seems to be a passive listener who doesn’t think about what he is hearing and how best to set his hearing aids. Audience member: Have any of you tried to train a passive listener to be an active listener? Todd: That is something we, as clinicians, need to do. But it is not easy, since not all strategies can be applied by every individual. Audience member: The person also needs an opportunity to unlearn old habits and to relearn new ones over an extended period of time. Todd: Yes, and that requires additional follow-up and training sessions which are not always easy to schedule. Samantha: I would like to add that some people have such a poor ability to understand speech in noise that the advantage provided by directional microphones is sometimes not sufficient to overcome their problem. For these individuals, frequency modulation (FM) technology may be more beneficial in difficult listening environments, since it improves the signal-to-noise ratio (SNR) to a greater extent than directional microphones can. Brian: There is undoubtedly a range of SNRs within which the benefit of directional microphones is most noticeable to the patient. The functions relating intelligibility to changes in SNR for both microphone modes take the shape of an ogive. Recognition scores for the directional mode in the mid range will exceed those for the omnidirectional mode. However, at very unfavorable SNRs and very favorable SNRs the two functions come together; that is, performance is the same. In these flat parts of the functions, we do not observe a directional advantage in behavioral testing. So, if a patient is in a relatively quiet listening situation or is in a very noisy one, he or she may not notice any benefit from the directional processing. The directional processing is working. It’s just not in a range of SNRs that can result in a noticeable performance difference between the two microphone modes. It’s not clear to me why some people don’t use the directional microphones on their hearing aids when they ought to be helpful. Is it because they don’t know when or where to use them? If so, we need to find a way to educate them so that these individuals can be turned into successful users.

Is there value in clinicians comparing omnidirectional versus directional speech testing as routine clinical protocol?

Todd: It depends upon your goal. If you want to prove that directional microphones work, then testing both microphone settings is worthwhile, but if you want to predict how well the microphones will work in the real world, then no. Ruth: I would suggest that giving patients the opportunity to listen with both microphone settings in the clinic is very worthwhile. You should caution them, however, that things may not sound dramatically different in the real world. So, my answer is no. Brian: As Todd suggested and as I indicated earlier from the description of the work in our laboratory, speech testing in the clinic doesn’t enable one to predict real-world benefit with a high degree of certainty. If you could perfectly mimic an everyday listening situation in the clinic or laboratory, you might expect to get good prediction of performance in that real-world listening situation. But this is difficult to do. It is not that clinical speech testing is of no value, it is just that listening in the test booth is so different from everyday listening that the ability to generalize from the clinic to the real world is relatively poor. Ruth: So, the question is, what should clinicians do relative to assessment? Most booths are equipped with only two speakers. If you carry out testing with one speaker in front and one behind you’ll never make correct real-world predictions. Audience member: I am interested to know why some people don’t show directional benefit in the laboratory. I think that is something that should be looked at. Is it that the directional microphones in the hearing aids worn by some people do not work or is it that our tests of speech understanding are not sufficiently sensitive? Brian: I agree that this is an important question. I doubt if there is a simple answer. I wonder whether, in some cases, it might have to do with aging and/or an inability to use the new information, especially without a period of familiarization or training. Todd: Sometimes the reason has to do with the test procedures used. We’ve looked at speech testing using adaptive procedures and using a fixed SNR. Sometimes people who show no benefit with an adaptive test do show benefit when tested at a fixed SNR and vice versa. We don’t know why this should be the case, but it has made us aware that we often need to do both types of testing. Samantha: Another factor might be auditory processing abilities. In our work looking at directional microphones and FM technology, one of the things we noticed is that some of our subjects who performed poorly on an adaptive test of speech in noise, and who derived little directional benefit, also had auditory processing problems. Bear in mind, however, that these data are from a small study in which only four subjects had evidence of an auditory processing problem.

When should we use adaptive directionality?

Ruth: Over the past 3 years we have been studying various aspects of adaptive directionality. We have found that if there is more than one noise source present, adaptive directionality is no better than nonadaptive directionality. However, we are currently completing a study using a second-order microphone that has three ports rather than just two. It has a narrower beam than a two-port microphone and thus has more directivity. We have found that adaptive directionality with this type of microphone is statistically better than with a two-port microphone, but the difference is small. The bottom line is, if there are two or more noise sources that differ in level by less than 15 dB then adaptive directionality is no more effective than fixed directionality. Unfortunately, it is rare that you have just one source of noise in a real-world listening situation. Todd: I completely agree. There are situations in which there is just a single noise source in which adaptive directionality does work, such as in outdoor environments. In these cases the microphone null is located in the correct place relatively often. Ruth: Incidentally, research on adaptive directionality is difficult because the noise must move within the speaker array.

How well do automatic directional microphones work-that is, hearing aids that automatically switch between omnidirectional and directional settings?

Brian: If a hearing aid could make the right decisions all of the time then this type of system would be ideal. This is a lot to ask because it entails the hearing aid sampling the acoustic environment and having an algorithm make the correct decision regarding the patient’s preference for one microphone mode or the other based on some acoustic analysis. In our examination of these products, we have found that many patients prefer the manual mode probably because they don’t like the decisions made by the instrument; that is, the automatic algorithm does not consistently set the hearing aid in the preferred microphone mode. So they want to have control. I admire the manufacturers who are working hard on this problem. It is a formidable one, because you are asking the hearing aid to determine, among other things, intention; that is, to what does the patient wish to listen? That might not always be in front of the patient. Hopefully as we learn more about the types of listening environments in which each microphone mode is preferred, engineers can design algorithms that make better decisions than those currently available to us. Another huge problem that ultimately must be addressed is individual differences. For example, we saw a patient in my laboratory with a severe hearing loss who wore her hearing aids set in the directional mode 90% of the time. Compare this to the average patient who may use the directional mode only about a third of the time or less. How do you design that kind of individual variability into an algorithm? Perhaps we may ultimately need to develop trainable hearing aids that can use patient feedback in everyday listening to adapt the signal processing to the patient. People are fundamentally so different. This technology is new and we are currently using simple approaches. Things will get better and better. Ruth: It is my understanding that current automatic directional microphones use level-based algorithms for switching of microphones. The question is, do they examine the level of the noise or the speech? Todd: Based on my experience, my opinion is that automatic switching is not as good as manual switching.

How does today’s directional technology compare with FM technology?

Samantha: FM technology improves the SNR to a greater extent than does directional microphone technology. In one study, we found that individuals using FM technology realized a 15-dB improvement over hearing aids alone. So, if a patient presents with particular difficulties hearing in background noise, we may want to recommend the use of FM technology in addition to hearing aids equipped with directional microphones. Interestingly though, subjective reported benefit did not always correlate with measured benefit. Todd: We have just started a study with kids in the classroom to compare the effectiveness of FM technology and directional microphone technology. We have found that if there is just a single speaker, FM technology is far superior. But for listening in groups with multiple speakers, FM technology is effective, but it is inconvenient.

What experiences have the panel had with the LINK-IT™ from Etymotic?

Ruth: The LINK-IT™ has an array microphone so its directional index is high. However, it is very visible to others. Generally users want their hearing aids to be invisible, so I am not sure how well used it will be. Todd: With an array microphone like that of the LINK-IT™, people will do better in a lot of listening environments. However, the effectiveness of directional technology decreases as reverberation increases and as the distance between the signal source and the microphone increases.

Is there a situation in which listeners prefer onmnidirectional microphones over directional microphones?

Brian: In the majority of everyday listening situations, the typical patient will prefer omnidirectional processing to directional processing. This includes most quiet listening situations, but also includes noisy listening situations where the signal of interest is not in front of the listener and/or at a distance. Again, directional microphones generally will be preferred when noise is present and the signal of interest is in front of and relatively close to the listener. The apparent importance of distance between the signal source and listener in our studies undoubtedly reflects acoustic factors that vary with distance, such as intensity of the signal; SNR, which when there is spatial separation of the signal and noise sources will vary with the distance; and reverberation, which will increase with the distance. To be effective, directional microphones require some spatial separation between the signal location and the noise location. Reverberation has the effect of eliminating the acoustic separation of the signal and noise. So, the effect of reverberation is to muddy the waters for directional microphones. There may be other reasons why omnidirectional processing might be preferred in everyday listening situations, even when you might think that directional processing would be advantageous. It has been suggested, for example, that patients generally prefer omnidirectional processing because it is more like natural hearing, which obviously is largely omnidirectional. We may like the sense of connection with our acoustic environment that omnidirectional hearing provides. Todd: A few specific cases: When listening in quiet with high reverberation, reflected energy from the source reverberates for a longer time and this reflected energy acts to mask the source. In contrast, when there is low reverberation, the reflected source energy can add to the source energy enhancing performance. If you attenuate the reflections (with a directional microphone) in the environment with low reverberation, listeners may not like it and can’t perform as well because the overall source energy level is lower. Brian: I just want to emphasize again that there may be issues largely unrelated to intelligibility that may influence preferences for omnidirectional or directional processing. I mentioned a couple of issues already-the listener’s intention and connection with the acoustic environment. Another area that I think needs more work is the relative loudness of an acoustic environment coming through the two processing modes. Even when you completely compensate for the low-frequency roll-off in the directional mode, there can be loudness differences because the directional processing is introducing nulls in the polar response. Patients will sometimes complain that things sound too soft in the directional mode. This can be true, even with complete equalization. This reduction in loudness also might be a good thing in certain noisy environments. So, it’s possible for the directional processing to have little or no effect on intelligibility, but the patient might prefer-or not prefer-such processing due to loudness differences between the two processing modes. Now, having said all that, our preliminary data on loudness differences and preferences for omnidirectional and directional microphones suggest that these effects appear to be rather subtle. However, I think more work needs to be done here to determine the extent to which loudness differences might influence microphone preferences.

How does degree of hearing loss impact the benefit with directional microphones?

Todd: In the first study I talked about regarding SIIs, we looked at mild, moderate, and severe hearing losses. The absolute directional benefit was lower for those with severe hearing loss. But the directional benefit was equally predictable from SII across all three groups. We calculated SII assuming a common hearing loss desensitization factor. The bottom line is that clinically you will see less directional benefit with more severe hearing losses, but that is expected given articulation index theory. The SNR/performance function is known to be flatter with more severe hearing losses regardless if a directional hearing aid is used or not.

Do compression settings affect directional benefit?

Todd: Essentially, no. There are some instances in which there may be some minor interactions. However, in most situations, signal and noise are present simultaneously, and compression systems apply the same gain to both signal and noise when they occur at the same instant, so compression does not affect the SNR. A few years ago we looked at directional benefit, compression mode, and hearing aid style in a reverberant environment, and found that the compression setting never significantly impacted directional benefit.

Do digital noise reduction algorithms affect directionality?

Todd: No, they are independent.