I've long thought the problem with hearing aids is the calibration process. People (often old, confused, and less tech-familiar) try them on and spend half an hour answering questions about "does that sound better?"
What if, instead, the calibration process worked constantly? Give the user a button to press when they don't hear something well, and another when they do. Let them have this for a month, let them try it in their kitchen, their bedroom, the local shop or bar, outdoors. Run some reinforcement learning algorithm to optimize for getting more "good" presses and fewer "bad" ones. Optionally, adapt separately to each environment.
Is there a "smart" hearing aid calibrator that works like this? If not, but you think it's plausible, I'm interested in working on it.
There are actually a good reasons not to do this. A very large part of your hearing is brain training...your brain learning how to interpret the new stimulus it is getting. You need to give your brain time to adapt to new stimulus before evaluating whether you are hearing better or not. If you are constantly fiddling with your hearing aids, you never get the benefit of training.
This may not apply if your hearing loss was sudden, but many people get hearing aids only after years of decline. Your brain adapts to what it has available and you learn to hear as best you can with the limited stimulus. When you finally get hearing aids, it feels like your ears have been unblocked.
When it comes to interpreting speech though, especially in noisy circumstances, you have to give it time. Your brain isn't used to using the extra stimulus to interpret speech. In fact, it may have gotten used to hearing loss filtering out background noise and you may initially find all the stimulus makes it harder to interpret speech because of all the background noise you now hear that you previously didn't.
If you don't have patience and start immediately futzing with volume all the time, you never give your brain a chance to adapt, and will be very dissatisfied.
Yes! Modern hearing aids do some fancy processing to shift frequencies into the more audible range of the patient. Many people do not have a simple "not enough volume" hearing loss. Personally, it's like someone went and messed with all the settings on the audio equalizer, and then added some muffling to make sharp sounds less distinct.
When programmed properly, I find my hearing aids on their various speech modes to sound distinctly artificial and processed. I imagine it would be jarring to the uninitiated. Like a good speech codec or vocoder. It's unlike anything else.
But having grown up with DSP hearing aids doing this all my life, I find speech coming through them very comprehensible, often shockingly so compared to my normal hearing -- even with other amplification like regular headphones. I do dislike my hearing aids for music, though, even on the normal or music settings.
Couldn't this brain training effect be included in a progressive algorithm like the one suggested? Such that the brain is gradually "guided" to the best performance by ramping up the hearing aid?
My (Phonak) hearing aids had a multi-week ramp-up to full power, in order to give my brain time to slowly adjust. Dealing with things like the occlusion effect (https://en.wikipedia.org/wiki/Occlusion_effect) just takes time, and getting used to something being in your ear for a large number of hours per day.
...especially with the use of a smart phone app...
Based on my observations of my father's struggles with various hearing aids, hearing aid smartphone apps are just as terrible as any other apps developed by hardware manufacturers. Don't expect the "clinicians" to help with this aspect either, because they were trained on a previous version of a different app and how could they be expected to figure out the apps that are used now?
What's really needed is a decoupling of the hearing test from the calibration activity. A basic hearing test takes a few minutes in an isolated box with headphones (the 'ol click a button if you hear the tone test). From that profile it should be trivial to build out the profile of what frequencies to boost and what not to and program the aids there.
I appreciate the optimism (and do think there's some room for efficiency improvements!), but each step is a little more complicated than that. a true audiometric room/booth to do the testing (eg fully soundproof) in is very rarely found outside of audiologist offices or specialized academic spaces, for starters, and I do believe those are required for paperwork (whether it be insurance related part of the process to getting the right device fitted).
then there's also additional complications like types of hearing loss impacting calibration - some people react very negatively to higher (or lower) frequencies than others do, to the point facial stimulation can happen, and some people just have a frankly tricky profile that requires a trained audiologist on hand to work with them through several sessions, if not outright yearly on top of that, since the nerves and brain output can change by itself.
(source: have a cochlear implant, been doing the described experience since I was ~ 3yrs old.)
To add to this — I suffer from Ménière's disease, which manifests for me as variable hearing loss. Some days are bad, and I can hardly hear/understand speech even when someone is right in front of me and I can lip-read. Some days are good, and my hearing aids are uncomfortably loud or make things sound "weird". But most days, my hearing aids do exactly what I need them to :-)
I've had audiograms taken on moderate and bad days, and the difference in frequency response was significant (up to 30dBA difference across only the lower frequency bands).
my partners hearing aids has a smart phone app, but it seems to be built with the concept of someone else putting the calibration into it. They can then create profiles for the different situations, however it strikes me that the hearing test calibration could be done in the app as well. I would imagine for a lot of people, if you could just buy them online you wouldn't need the clinicians for many cases.
The app for my hearing aids allows my audiologist to make remote adjustments, so I don't even need to visit their office for straightforward changes. I really want to reverse-engineer it so that I can modify some of the details of the profiles myself.
All audio devices used to have an instant-feedback feature like this. You could dial it in with zero latency. Early technosapiens called it the Volume knob :-).
> What if, instead, the calibration process worked constantly? Give the user a button to press when they don't hear something well, and another when they do.
On TVs, they call those buttons volume buttons, it is revolutionary.
Hearing aids have volume controls, which some clinicians disable for new users - as it causes more problems than it fixes.
And just like the TV remote they increase all sounds - good (speech) and bad (music and explosions).
The issue is not just "is it loud enough" but rather,
"can you understand what you want to hear when you want to hear it?"
imagine yourself at a busy restaurant.
you want to hear your spouses speach better, and the wait staff- but turning their speech up also turns up the people you do not want to hear.
how do you fix that?
noise reduction, speech enhancement, directional microphones are just some of the tools a hearing aid uses.
But ultimately what needs to happen is your brain works harder at comprehending what you want to hear(yes, it is work, and like all work people do not want to do it).
So they think...hmmm, I will just pay $xxxx.xx and I wont have to work anymore.
Reminds me of the race to the bottom we're also seeing in orthodontics.
Whether like me, you never got the hang of a retainer after braces - or went off braces too soon, in my case - there has been a number of companies trying to get costs down and make the process less expensive and onerous.
Yet, talking to my regular dentist, who has used Invisalign with patients before, it's not the fact that there are more competitors in the field, essentially putting pressure on the DMD's and others to up their game, it's that after the initial consult, the entire process is (largely) unsupervised.
If you don't fit (or tune) something like this properly, it doesn't work.
A lot of the time it's not absolute volume. If you can't hear certain frequency ranges well, boosting the volume of all ranges is just going to drown out the ones you can't hear well.
Exactly - hearing is much more complicated than just noise sensitivity. For example, a somewhat underdiagnosed problems is issue with the frequency bins that the brain segments sound into for recognition deteriorating. Symptoms are - can hear just fine in a nice, uncluttered audio environment, say a test centre, but has issues if there is any significant background noise.
> Symptoms are - can hear just fine in a nice, uncluttered audio environment, say a test centre, but has issues if there is any significant background noise.
Also a problem if you're not a native speaker. I am fluent in English and have been able to watch movies without subtitles for 20 years at least, probably more.
But put me in a noisy bar and my error correction suffers a lot. Suddenly it's hard to follow a conversation that in my native language would be effortless.
The worst situation is listening to a loud video on bad speakers. Doesn't matter how much you crank the volume, people are hard to understand because the brain's error correction isn't good enough. Use a good speaker and turn down the volume, suddenly it's perfect.
Do any hearing aids frequency shift, or compress the frequency scale (e.g. just higher frequencies compressed down to lower frequencies so that voice tone stays constant and sibilants get shifted down into hearing range).
yes, I know Phonak hearing aids were doing that since about 2012.
For some of our more severe hearing impaired patients it was a noticeable improvement in speech comprehension.
That is exactly what the tech does,
automatically adjusts based on the amount of noise and speech in the environment.
And going a step further, applying directional microphone control to better pick up sounds where the speech is.
go further, and having 'hearing aids' that would let you rewind the last 5-10 seconds of a conversation to hear something again without asking people to repeat themselves...
I think this is more akin of the Android phone I have asking me if the call I just placed was: clear, had cutouts or dropouts, fuzzy background noise ... And they take that feedback and adjust my call quality in GoogleFi (in theory)... It's not just the concept of s volume button, but more the idea of reinforced (ai) learning over time, or algorithm adjustment with human feedback.
...which will simply make you more deaf. hearing aids do not simply amplify volume. you know, because they're not headphones and a microphone.
but your idea is great too. if you can't hear one frequency, let's amplify all frequencies till the one you need gets loud enough, and go deaf to the other ones in a year. when you get old and need a hearing aid, you should stick with your method.
What if, instead, the calibration process worked constantly? Give the user a button to press when they don't hear something well, and another when they do. Let them have this for a month, let them try it in their kitchen, their bedroom, the local shop or bar, outdoors. Run some reinforcement learning algorithm to optimize for getting more "good" presses and fewer "bad" ones. Optionally, adapt separately to each environment.
Is there a "smart" hearing aid calibrator that works like this? If not, but you think it's plausible, I'm interested in working on it.