Providing Adequate Gain In A Hearing Aid Fitting

Several weeks ago our longtime friend Dr Bob Martin published the very helpful How to Design a “Maximum Hearing” Program for Sharon the schoolteacher who has a moderately severe hearing loss; but it’s worth deconstructing just why Sharon is having problems to begin with.

Dr Bob wants to discuss ways to modify listening programs to help patients hear well from a greater distance; and he presents a schoolteacher, named Sharon:

Sharon has a moderate hearing loss bilaterally. Her pure-tone thresholds are the same in both ears:
40 dB at 250 Hz, 50 dB at 500 Hz, 60 dB at 1000 Hz, 60 dB at 2000 Hz, 60 dB at 4000 Hz, and 65 dB at 8000 Hz.

OK, let’s plot this out on Mead Killion’s Count-The-Dots audiogram1 (more on this in a few moments):

Count-The-Dots composite unaided audiogram-thumb

Unaided audiogram, plotted with blue X’s for convenience. The Spech Intelligibility Index SII (or Articulation Index AI) is zero, which means conversational speech at 60dB SPL (re 20µBar) is unintelligible. What’s more, even if the speech is presented 10dB louder at 70dB SPL (re 20µBar) which is 50dB HL, it is still unintelligible
Click to enlarge in a new window


Dr Bob then writes:

When I enter these values into the manufacturer’s software and select the Universal program, the computer recommends gain values for soft, medium, and loud input signals. These values are:

 Gain values for the Universal program

               250  500    1000    2000     4000 Hz

 Soft            0       8       20       26          18 dB

 Medium     0       5       12        20         14 dB

Loud          0       2         6        12          0 dB

What you show demonstrates the well-documented problem of underfitting high frequencies with manufacturers’ “first fit” methods, especially in the high frequencies. Assuming the soft, medium & loud gain settings are for 40, 60 and 80dB SPL, and also reminding that Sharon’s 4kHz threshold of 60dB HL is actually 69.5dB SPL, we need to ask “at what input level will she hear sound?

The 4kHz gain at 40dB SPL input is 18dB, at 60dB SPL input is 14dB, and at 80dB SPL it’s a big fat zero. Interpolating, we have the gain with a 50dB SPL input of just 16dB, for an output of 66dB SPL. But! Her threshold is 69.5dB SPL, so basically speech sounds below about 53dB SPL at 4kHz Will Be Inaudible.

No wonder she has poor speech discrim: The critical high frequencies have been rolled off by the crapware’s “first fit” algorithm.

Going further into this, let’s see what happens at 2 & 3 kHz:

Assuming her threshold is 60dB HL at 3kHz as well as 2k & 4k, her 2 & 3k thresholds are 69 and 70dB SPL; and interpolating at 3k we have G40 = 22dB, G60 = 17dB; and again interpolating that would yield a G50 = 19.5dB. Let’s put these interpolated values into a similar table to the one above:

Input-Output table in dB SPL & dB HL

Let’s plot these (green dB HL) aided response on the Count-The-Dots Audiogram, bearing in mind that all values below 60dB HL are Not Audible to Sharon:

Count-The-Dots composite aided audiogram-thumb

Sharon has a Major League Problem: From below 1.3 kHz & above 2.5 kHz, she needs over 50dB SPL just for those critical unvoiced consonants such as /s/, /sh/, /t/, /θ/, /p/, & /φ/ to be audible — And guess what? At 3 feet with 70dB SPL speech those consonants are about 30dB down at 40dB SPL, while her aided thresholds are 10 dB & more worse.

How we would fix Sharon’s problem:

This is a classic case where the combination of measuring frequency-specific UCL’s, speech mapping, and the Best Practice of Real Ear Measurement (REM; probe mic) can — and should — be used to troubleshoot this problem fitting, in order to carefully squeeze more of those Red Dots into Sharon’s audible range. What’s more, consider switching the fitting prescriptive method to DSL5-I/O or even Brian CJ Moore’sCAM22 (short for CAMEQ2-HF, the methodology we at The Hearing Blog use), along with counseling Sharon that by boosting the highs we’re giving her “the good stuff” so she can hear even better — Patient counseling goes a Long Way when the clinician can explain the benefits of extended high frequency amplification if you phrase it properly.

Starkey Inspire MPO fine tuningClick to enlarge in a new window

Starkey Inspire MPO fine tuning
Click to enlarge in a new window

If “Sharon’s” UCLs represent a problem, one fitting quality control feature we at The Hearing Blog really like is in Starkey’s Inspire software, also used for Audibel and NuEar hearing aids: Instead of just guessing at what the UCL is, based on (often-dodgy) speech audiometry, Inspire has a rapid “beep UCL check” across the audible spectrum so that the clinician can quickly trim MPO’s to prevent Sharon’s discomfort; and just as importantly, sniff out a few extra dB in the UCL to leverage for better dynamic range, since “Sharon” just may have limited dynamic range at certain frequencies.


  1. If you haven’t yet acquainted yourself with Mead Killion’s updated Count-The-Dots audiogram, we highly recommend you download the article here and consider printing out and using the updated version on page 3, as the copyright has been released;
  2. By the same token, although recently retired, Cambridge Professor Brian CJ Moore still has his thumbs in their world-class Auditory Perception Lab — Check out their great work~
← The Hearing Blog Investigates: Is Hearing Loss from Susac Syndrome Sensorineural or ANSD? Three new technologies from AB, Phonak and Med-El →

About the author

Dan Schwartz

Electrical Engineer, via Georgia Tech

Leave A Reply

Time limit is exhausted. Please reload CAPTCHA.

%d bloggers like this: