Gain versus Target Gain

image

I’m an aspiring DIYer using Phonak Target, but I think the following questions are not specific to that software, except possibly for the third one.

The thicker blue lines are the gain, for 80 dB speech, 65 dB speech, and 50 dB speech, respectively, from bottom to top. The thin blue lines, which for the most part coincide with and are thus obscured by the the thicker ones, are the target gains for each of the input levels. (The dashed magenta line is the estimated feedback threshold, and the gray line is the gain limit.)

(1) I don’t know exactly what “gain” means here. It’s not a target; it’s not a measured result. It’s the output of a fitting prescription, but what is it?

(2) I don’t understand why, in the region of approximately 5 kHz - 8kHz, the gain meanders about the target gain like a drunk driver.

(3) The 50 dB speech gain seems to be constrained by the feedback threshold in the 3 kHz - 6 kHz region. Is there any way to eliminate that constraint? No Feedback and real ear test was performed, and the feedback suppression feature (WhistleBlock) is turned off in all programs.

Thanks for any insight…

After posting, I had this amazing [cough] insight, or guess:

Target gain is that required to offset the audiogram hearing loss.

Gain is the gain prescribed by the fitting formula to meet its design goals, usually a tradeoff of natural sounding and (speech) clarity.

1 Like

@brec Target can measure the feedback threshold. You can set it to varying degrees of suppresion at the Finetuning program options under whistleblock. You can allow Target to exceed the threshold by 5dB, though, by selecting it at the bottom:

Thanks, but is there any way to overcome the threshold completely, whether measured or estimated? I’d like to be able to do that as an experiment in order to see how the benefit of “liberated” gain compares with the cost of more instances of feedback.

Turn whistle block off.

1 Like

Notice how the gains for 65 and 50 dB inputs adhere to the estimated feedback threshold despite the target gains being higher:

image
However, in some instances the gain is below target even where target is below the feedback threshhold, e.g., for the 80 dB input between about 3 kHz and 6 kHz. I don’t know why that is.

In initial settings.

Where, exactly? In Settings I see nothing about WhistleBlock or feedback.

Global settings I believe. At the bottom of the page.

That’s what I meant by “Settings.”

Don’t run the feedback test and turn off whistleblock, now that’s going to give you pretty"liberated" gain, don’t forget you can change the MPO as well if all you want is more gain, set up initially in a new client with moulds (no vent, closed) remember Target’s algorithm only allows for so many changes, so your experiments may not pan out the way you want them to.

@tenkan Could you elaborate on this? It seems to imply that every once in a while you should get a new profile? Or would changing the acoustics settings be enough?

1 Like

What I mean is for every action there is a reaction, so changing one thing (say 80dB at 3kHz automatically changes in the 65dB and 50dB ) so sometimes you simply can’t get the desired result you actually want because of Target’s “algorithm”
No you don’t have to change to a “new profile” I only suggested this for @brec to do for his “experiments”
and yes you can just change the acoustics or even the fitting formula to try and get the desired results, but with a few different clients set up you can keep the different settings and change between them to see which works best when “experimenting”

1 Like

I have been fooling with Target for a couple of years now. I agree that it does some things I don’t understand. The wiggle in gain curves seems to be an artifact. I have tried, but never was able to fully get rid of it. For some reason, my current G65 gain is higher than my G50 gain. Go figure. It has, if I remember correctly, either 6, 10, or 18 gain handles, or frequency bands that can be adjusted. If I change the gain of one band, Target changes the gains of the two adjacent ones. It does this regardless of how many gain handles I select. It has some other odd things it seems to want to have a certain way, and not let me change. I would have to have the program up and running to remember them, and I am on a different PC right now. I imagine professionals who use it every day learn to ignore the odd behavior. Actually, I have never had a professional seem to want to talk about the programming software in any detail at all. Even when I told them that I have it and use it.

I found a way.

In reviewing another topic I just noticed that in a @Sierra reply in which he posted a diagram of a DSL v5 fitting against my audiogram, the gains were not limited by the estimated feedback threshold. I hadn’t been able to achieve that with my experiments using the two Phonak forumulas. Aha - only the Phonak formulas honor the estimated feedback threshold! DSL and the NALs do not. However, all the formulas honor measured feedback, i.e., the result of a Feedback and real ear test.

A feedback suppression mechanism like Phonak’s WhistleBlock, which can be separately enabled, with a fitter-selected strength, on a per-program basis, can’t suppress feedback “noise” without also suppressing the desired signal, right?

Yes that’s one of the problems that all these so called feedback algorithms have, so contrary to what is spinning out from the manufacturers, hearing aids are still dumb,they can’t think for themselves like say AI.
I’m still not sure what exactly you are trying to achieve, just more gain without feedback or something else?

I’m trying to achieve more gain without feedback, hoping that will yield increased speech intelligibility. I’m also trying to understand what Target is doing.

Getting the proper size hearing aids (gain) for your hearing loss is how you get best speech intelligibility. Proper acoustics will eliminate feedback.
These are fundamental hearing aid procedures.

Have you got your aids yet?

I found if you performed the feedback test and then switched off the WhistleBlock via the feedback and real ear test section. The lines are higher and the dotted line disappears.

1 Like