Real-Time Spectrum Analysis:
A New FONIX 6500 Tool for Improving a Hearing Aid Fitting
by Larry Revit, hearing scientist
"My Voice Sounds Like It's in a Barrel!"
We Can Help-
What is REAL-TIME SPECTRUM ANALYSIS?
"Spectrum Analysis" means that a complex sound, such as speech or noise, is broken into its individual frequency elements and then measured. The result is an amplitude-versus-frequency graph that tells the observer how much energy is present at each frequency. "Real-time" means it happens on the spot. The Fonix 6500 has done this all along, primarily using the Speech-Weighted Composite signal. But now you can use any signal you want, and get a Real-Time Spectrum Analysis from either the test chamber or the probe microphone.
How can REAL-TIME SPECTRUM ANALYSIS help in a hearing-aid fitting?
Here are some of the ways a dispenser can use Real-Time Spectrum Analysis:
- Take the guesswork out of the question: "How will the hearing aid perform under real-use conditions?" -- by measuring real-ear gain using live or recorded environmental sounds.
- Use real speech sounds to see that a hearing aid is producing them correctly. For example, you can analyze the "[f]" and "[s]" sounds, through a hearing aid, to see why a person might be having trouble discriminating between the words "fifty" and "sixty".
- Analyze the client's own voice in the earcanal. This is one of the most interesting uses of Real-Time Spectrum Analysis -- for two reasons:
First, it's a way to involve the client in the real-ear measurement process. The client can speak various speech sounds and watch the frequencies change on the screen. Since the sounds are being picked up in the client's own earcanal, this is an excellent way to familiarize a person with real-ear measurements. Plus, it's fun, and impressive.
But there's another important professional use of this new measurement capability: solving the "occlusion effect" problem. The "occlusion effect" is the cause of one of the most common complaints from new hearing-aid users. Almost every dispenser has heard the words, "My voice sounds like it's in a barrel!" But the example below clearly illustrates how Real-Time Spectrum Analysis can help solve the occlusion effect problem, and perhaps even save a hearing aid fitting:
The graph in Figure 1 shows the spectrum of a sustained "ee" sound from the client's own voice, measured in the client's earcanal.
FIGURE 1 -- OCCLUSION EFFECT. Example spectrum of hearing-aid-wearer's own voice, sustained "ee" sound, in the earcanal, with a pin-hole vented earmold in place, with the hearing aid turned off.
The hearing aid, with only a pinhole vent, is in place, but the aid is turned off. The earmold itself has a considerable amplification effect on the low frequencies of the wearer's own voice. Even with hearing aid off, the low frequencies are cooking away at over 95 dB SPL, and the overall RMS OUTPUT is more than 100 dB SPL! No wonder there's a "barrel" problem.
Figure 2 shows the solution.
FIGURE 2 -- OCCLUSION EFFECT. Example spectrum of hearing-aid-wearer's own voice, sustained "ee" sound, in the earcanal. Thick curve in lower graph is with a pin-hole vented earmold; thin curve in lower graph is with a wide-open earmold , both with the hearing aid turned off. Upper graph shows the difference between the two lower curves.
The thick curve in the lower graph is the same curve as before. It has been saved as the "UN-AIDED" curve. The "AIDED" curve shows a similar measurement, but with the vent wide open. (Keep in mind that these curves are of the client's own voice, momentarily sustaining an "ee" sound while the real-time measurement is captured.) The curve in the upper graph, labeled "INSERTION GAIN", is the difference between the two lower curves. From either graph, it is clear that opening the vent in the earmold has drastically decreased the occlusion effect, by up to 20 dB!
The same set of measurements is possible with the hearing aid turned on, as illustrated in Figure 3.
FIGURE 3 -- OCCLUSION EFFECT. Example spectrum of hearing-aid-wearer's own voice, sustained "ee" sound, in the earcanal. Thick curve in lower graph is with a pin-hole vented earmold; thin curve in lower graph is with amedium vented earmold, both with the hearing aid turned on. Upper graph shows the difference between the two lower curves.
Here we see a 10-dB improvement in the low frequencies, this time in going from a pinhole vent to a medium vent. The wide open vent was not possible with the hearing aid turned on, because of feedback. But a 10-dB improvement means that the low frequencies of the wearer's own voice (the "barrel" sound) will be only half as loud as before (a 10-dB change means half the loudness). This kind of observation can give the dispenser, and the client, confidence of an improved hearing-aid fit.
Step-by-step, "cook-book" procedures for all the Real-Time Spectrum Analysis tests mentioned above are given in the revised Chapter 10 of the 6500 Operator's Manual, shipped with Software Version 2.66. I'm confident this new feature of the Fonix 6500 can be an important addition for our customers.