Probe-mic assessment of digital hearing aids?

Yes, you can!

By H. Gustav MuellerReprinted from The Hearing Journal, Vol. 54 No.1, January 2001 © 2001 Lippincot Williams and Wilkins


1. I’m a little surprised we’re still talking about probe-mic measurements. Since digital hearing aids are rapidly becoming the standard fitting, isn’t probe-mic testing just gradually fading away?


2. Well, I’ve been told by a couple hearing aid “reps,” and I think I also heard it in a talk somewhere, that you can’t do probe-mic testing with digital hearing aids. Don’t you agree?

No, that simply isn’t true. In fact, I’d start to wonder what it is that someone doesn’t want you to know.

Probe-mic testing is an important part of the hearing aid verification protocol—a tool that assists us with the verification of appropriate gain and output and also the features of the hearing aid. Digital hearing aids tend to have more features, so that is all the more reason why probe-mic testing is necessary. With infants and children, in addition to verification measures, probe-mic testing is an essential part of the selection procedure.

3. I just fitted two young children with digital hearing aids, but I’m not sure how probe-mic testing fits in.

Fits in? I don’t think you would want to fit digital hearing aids—or any other hearing aids on an infant or young child—without using probe-mic test results. As you know, because of our increased efforts to identify babies with hearing loss, we are fitting many more infants with amplification. And, yes, many of these children are being fitted with digital instruments.

However, unless we know the infant’s real-ear coupler difference (RECD)—hopefully measured through the child’s own earmold-selecting input-specific gain and output for these children is mostly guesswork. But if you know the RECD, you can essentially adjust and verify the hearing aid parameters for an infant in a 2-cc coupler.

4. I’ve always used aided soundfield testing as a verification procedure for young children. Isn’t that okay with digital hearing aids?

No more or less so than with analog hearing aids. I don’t believe this procedure will tell you what you need to know.

First, aided soundfield testing, even with adults, is fraught with measurement error. And, even if you assume that you have reasonably valid thresholds, threshold measures tell you nothing about the real-ear maximum output of the hearing aid, which, for the pediatric patient, might be more important to know than the aided thresholds. Moreover, if you’re fitting a hearing aid with WDRC (low compression kneepoints), which is quite likely when you’re fitting a digital instrument, the aided soundfield thresholds don’t represent the gain that will be present for average speech signals.

And, by the way, it’s been a few years since I’ve attempted frequency-specific soundfield measures on a 4-month-old, but I wasn’t very successful. Probe-mic determination of the RECD, and the subsequent adjustment of the hearing aid characteristics, can be conducted reliably at any age. It’s interesting to observe that, even with early identification, many of these children are not being fitted with hearing aids until they are a year old. It’s tempting to speculate that audiologists are waiting for the child to “grow into” the fitting procedure that they are most familiar with.

5. Okay, you’ve convinced me that probe-mic testing is essential with children. But what about adults? Isn’t there some problem with the digital noise-reduction feature?

Many digital hearing aids do have a noise reduction feature and, yes, this is something that you would want to test with your probe-mic equipment at the time of the fitting to make sure it’s working the way that you (and your patient) want it to work. But I don’t consider that a problem.

6. You don’t understand. I’ve been told that the noise reduction feature will cause me to record incorrect gain values, which is why I stopped doing probe-mic testing with digitals. Isn’t that true?

Only in isolated cases. Many of the digital hearing aids on the market don’t even have digital noise reduction, so there’s nothing to worry about with these. Of the ones that do, you can turn off the noise-reduction feature in the fitting software for nearly all of them. When verifying your desired targets, I would suggest that you turn off this feature before you measure input-specific gain and output. Then turn it back on and do probe-mic tests to see how the noise reduction works in the real ear.

If you don’t want to take the time to turn the noise reduction off or you’re fitting a hearing aid where it can’t be turned off, you can still obtain valid gain values by using input signals designed for testing digital hearing aids. So, I still don’t understand the problem.

7. Maybe I don’t either. I have fitted one of those products where you can’t turn off the noise reduction. What are those special signals you’re referring to?

All seven of the manufacturers of probe-mic equipment that I’m familiar with (and I don’t think there are many others) have some type of signal that can be used to obtain valid hearing aid gain values even when the digital noise-reduction feature is on. The key is to use a signal that has random modulations so that the hearing aid does not think that the input is noise and therefore reduces the gain (swept pure tones can be interpreted as “noise” by the hearing aid). Depending on the manufacturer, you’ll see terms like Dynamic Roving Tone (DRT), Digital Speech in Noise (DSIN), or Modulated Speech Noise (MSN). Several manufacturers of probe-mic equipment also have incorporated the recently developed ICRA noise signals.

8. Okay, you got me! What are ICRA signals?

ICRA is an acronym for International Collegium of Rehabilitative Audiology, the organization for which the signals were prepared by a working group dubbed Hearing Aid Clinical Test Environment Standardization (HACTES). There are different types of babble signals available (e.g., male, female, six-person) for different voice levels (normal, raised, loud), and these 11 noise” signals have modulation characteristics similar to those of natural speech. The signals are available on CD and, as I mentioned, one or more of them have been incorporated into several of the new probe microphone units (and are available as upgrades to existing units).

9. So, it sounds as if I’m in good shape to start conducting probe-mic testing with my digital fittings tomorrow, right?

I would think so. The worst case would be either that you’re stuck with a probe-mic system that hasn’t been upgraded to the new speech-like signals or that you have a hearing aid with digital noise reduction that you can’t turn off. But even then there is hope.

As you know, the digital noise reduction has an attack time. It varies from manufacturer to manufacturer, but if it’s longer than a second or two for the product you’re fitting, just make sure to use only a short “burst” when you present your test signal. That way you can capture the gain of the instrument before the noise reduction takes effect.

10. Assuming that my gain values are now valid, what sort of results should I be looking for?

Basically the same things you look for in testing an analog instrument. Digital hearing aids do tend to have more channels, so you might want to think a little about channel summation. Also, most digital hearing aids employ WDRC, so you’ll want to examine gain (or output) for varying inputs. Most people use inputs of 50 dB SPL, 65 dB SPL, and 80 dB SPL to estimate gain for soft, average, and loud signals. Presumably you have an idea of what you consider to be a “good fit” for this patient based on the resulting real-ear insertion gain (REIG) or real-ear aided response (REAR) findings, and you can now determine if your gold standard has been met.

If you’re using generic prescriptive fitting methods such as DSL4. 1, NAL-NL 1, or FIG6, it’s possible that you can simply display these targets on your probe-mic monitor. If you’re using a “special” fitting method from a particular manufacturer, then you’ll probably have to print out the real-ear targets and keep them by your side while you’re doing your probe-mic measures. If you don’t have any real-ear targets, then you have nothing specifically to verify, but you still could reach some general conclusions regarding the change in gain and output as a function of input intensity.

11. Earlier you mentioned something about measuring the real-ear effects of the digital noise reduction. How do I do that?

It’s really quite easy (and fun) to do, especially when you’re doing it intentionally and not by accident. Here’s a protocol that I’ve used with several different products that seems to work quite well:

  • Position the patient so he or she can observe the probe-mic display monitor
  • Begin testing with the noise-reduction feature turned off (if possible).
  • Present a continuous broad-band noise” signal to the patient at 65 dB SPL to 70 dB SPL. This signal remains on throughout the testing.
  • Observe the measured REAR (if you want to display the results in REIG, you would first have to store the patient’s real ear unaided gain-REUG).
  • Turn on noise reduction for all channels. Observe the drop in real-ear output (gain). Note the time that it takes for maximum reduction to occur. Discussion of the attack time of this feature is important in patient counseling, and it’s easier to explain if the patient has already observed the change while the hearing aid was in his or her ear.
  • Now talk to the patient at a level in excess of the continuous noise. Observe that output (or gain) will return to previous levels (at least for some channels) when speech becomes the dominant signal. Note how long it takes for the noise reduction to “release.” Again, the release time is something that needs to be explained to the patient.

Based on these results, you might decide to change the strength of the noise reduction or try using speech and/or noise signals at different input levels to determine how this feature will work in different listening conditions.

12. I think I understand what you’re doing, but doesn’t all this testing add a lot of time to the fitting procedure?

Assuming you have the person set up for probe-mic testing anyway, it probably adds another 5 minutes or so per ear (and I would recommend testing each ear separately). If digital noise reduction is one of the main features that you are “selling,” spending 10 or 15 minutes to verify that it is working appropriately does not seem unreasonable to me. And, my guess is that a good demonstration of this feature will improve the patient’s understanding, which likely will save time on post-fitting visits. Additionally, most of us learn a few things when we do these measures, which improves our tweaking and counseling skills.

13. You could be right. Speaking of noise reduction, is there also a way I can test the hearing aid’s expansion circuit in the real ear?

Maybe. It depends on the noise in the room where you do your probe-mic testing and the expansion kneepoints of the instrument you’re testing.

As you probably know, the expansion feature is designed to reduce gain for low-level sounds like ambient room noise and microphone noise. However, when the input to the hearing aid is above the expansion kneepoint, there is no effect. In some fitting rooms, ambient noise levels are above the expansion kneepoints, which may vary from channel to channel, but usually are around 35 dB SPL to 45 dB SPL. If your room is noisy, you can always check out the expansion feature using a 2cc coupler in a test box.

14. My probe-mic equipment is in a test booth, so the ambient noise level is quite low. Can you give me an idea of a test protocol that I could try?

Sure. It’s always more interesting to test hearing aids in the real ear, so try the following to test expansion with your next digital hearing aid fitting:

  • Begin testing with the expansion feature turned off.
  • Adjust your probe-mic equipment so that it is in a “spectral analysis” mode. The probe microphone in the ear canal will be functioning as a sound level meter; you will not be delivering any signals from the loudspeaker.
  • Observe the SPL values in the ear canal with the expansion circuit turned off.
  • Turn the expansion circuit on and observe (hopefully) a reduction in the ear canal SPL.
  • A quick conversion of your patient’s HL thresholds to ear canal SPL will help determine the potential benefit of this circuit and assist in making adjustments if this is a programmable option.

15. My verification protocol is getting longer all the time, and you probably still have some other things for me to test in the real ear, right?

Of course. We certainly can’t overlook directional-microphone technology. While directional mics aren’t unique to digital hearing aids, most digital aids do have them, and they are indeed something you want to test in the real ear. It is possible, for a variety of reasons, that a hearing aid considered to be “directional” could be functioning omnidirectionally. That’s something you want to know before you send the patient out the door.

16. I normally just do a “listening check.”— What’s wrong with that?

If it works, I guess it’s okay. My ears just aren’t that good at accurately measuring front-to-back ratios, so I tend to rely on my probe-mic equipment. Again, assuming you already have the patient set up for probe-mic testing, this measurement adds only a few extra minutes.

Here’s how I obtain a reasonably reliable estimate of the real-ear front-to-back ratio:

  • Place the patient in a swivel chair.
  • Set the WDRC to “linear” (if you don’t, the WDRC processing will counteract the effects of the directional microphone, producing results that underestimate directionality in the real world).
  • Use an input signal of 65 dB SPL to 70 dB SPL. Any one of the signals that we discussed earlier should work just fine.
  • Turn your patient around so that you conduct the first run with the loudspeaker of the probe system located behind the patient, at a distance of around 3 feet. Tell your probe system that this run is the REUR.
  • Now turn your patient back so that he or she is facing forward and conduct the second run with the loudspeaker located directly in front of the patient. Tell your probe system that this run is the REAR.

The probe system will automatically subtract the first run from the second. What is then presented on the monitor is the frequency-specific front-to-back ratio of the digital instrument.

17. Wait a minute. I’m fairly certain that the front-to-back ratio can be influenced by such things as the speaker-to-listener distance, room reverberation, and the polar plot design of the instrument. How do you account for these variables?

I don’t. I’m not suggesting that this method be used to compare one product to another, predict real-world benefit, or collect data for a JASA paper. If you’re a typical dispenser, 80%-90% of the digital products that you dispense are from one or maybe two manufacturers. After real-ear testing of a few patients, you will soon recognize the “expected” front-to-back ratio of a given product. Let’s say you commonly see a front-to-back difference of around 20 dB at 1500 Hz to 2000 Hz. Now, when you test that same product on a patient and see a front-to-back difference of only a couple of dB at 2000 Hz, you know that something is wrong.

18. Makes sense. I just might start doing that testing with my directional fittings. Have we covered all the digital features that need to be tested?

Well, since you asked, there is one more that I think deserves mention. That’s the feedback-control feature that’s available on most digital hearing aids.

These systems seem to work pretty well, as we can now detect where feedback is occurring (or will occur) and conduct band-specific gain reductions to reduce it. But, rather than looking at simulated results on the computer monitor, I’d like to see what is happening in the real ear. If my “digital feedback-suppression algorithm” eliminated all the real-ear gain above 2000 Hz, I might want to think of a different way of treating the feedback problem.

19. I agree, but what about systems that have “adaptive feedback.” How do you measure that?

Well, again, I think you want to know what happens to the real-ear frequency response when this feature is activated whether it’s accomplished through narrow-band notch filters or through phase alterations.

What I usually do first is conduct an REAR without any feedback present, and leave this on the screen for reference. Then I have the person perform a task that will cause acoustic feedback (e.g., opening their mouth widely, holding a telephone receiver to their ear, etc.). I allow the adaptive control to cancel the feedback, and then run another REAR.

It’s easy to observe the effect that the adaptive suppression has on the frequency response, and you then can determine if this degree of change is acceptable (it usually is). As you can tell, rather than thinking of any of these digital features as limitations to probe-mic testing, I prefer to view them simply as additional “listening situations” that require real-ear assessment.

20. So tell me, what is the limiting factor when I conduct probe-mic measures of digital hearing aids?

That’s easy—your imagination!

Leave a Reply

Your email address will not be published. Required fields are marked *