Oticon More "deep neural network" processing: Does it WORK?

The More aids may not be able continue to learn but it is your brain that is relearning all the time because of what has been programmed in to the More processor.

To whom are you addressing your question? The @SpudGunner is definitely NOT an audiologist!

[The @SpudGunner is a retired … Gunner!]

@KiwiJohn said:

@SpudGunner Thanks for confirming my doubt that Mores can dynamically learn. My first post was simply in response to gkumar who wondered if they could.

. O-kaaay …

@KiwiJohn said:

And I must say that my audiologist, who was willing to sell me Mores if I wanted them, is also sceptical and just regards the claim as further evidence of Oticon’s skilful marketing

… but this accusation of false claims on Oticon’s part didn’t/doesn’t sound to me like a response to @gkumar 's query, that’s all.

I know I’m repeating myself, however, I have yet to read anything from Oticon that could be construed as an assertion that More’s on-board DNN “noise filter” is capable of dynamic learning.

@Volusiano has also debunked this notion several times.

@Volusiano said:

Just a clarification that the More DNN is not really preloaded with 12 million different sounds. It’s actually 12 million different sound SCENES. Sound scenes are various sound “environments” which can be simple or complex, indoors or outdoors, with maybe one or two sounds to dozens of sounds going on.

And the sound scenes are not preloaded into the More. There’d be no room to load that kind of data anyway. What they did was they took a special “globe” device that has microphones on it in all different directions in 3 dimensions. Then they record a sound scene (not sure for how long), then another, up to 12 million sound scenes.

Then in the lab, they created an initially very crude deep neural network model. This is the secret sauce here. We don’t know what the neurons in this deep network represent, and how wide (how many neurons) each layer has, and how many layers there are in the network. And each neuron is assigned weights and biases that are basically can be mathematically manipulated.

The DNN is basically a network where the sound scenes are input into it one at a time, cranked through it through the initially assigned weights and biases associated with each neuron (which represents “something”, probably some kind of detector and/or analyzer and/or sprectral representations that break down of the input sound scenes at the various levels of the network → just my guess here). The output is a crude “recreation” of the sound scene that was input, but its outcome is probably very different initially from what it should be (the reference sound scene). So they feed this difference back into the network and do mathematical manipulation of the weights and biases of each neuron to minimize the difference between this crude outcome and the reference (good) outcome. Through this cycle, the network is trained to be improved.

Then millions of various sound scenes get sequentially fed in, and through each of these cycle, the network gets trained through the process described above, to be more and more accurate over time, such that eventually it gets good enough, or maybe pretty darn good.

Oticon then loads this final “trained” network with all the smart weights and biases for each neuron that have been highly fine tuned through each cycle, into the More chip. Now you can feed through it any new real live sound scenes (other than the 12 million that were captured), and this smart network knows how to manage this sound scene accurately. In the process, the network can also break down all the sounds of any sound scenes into various sound components, and this allows the network to manipulate them much more easily to achieve a “balanced” sound scene in accordance to the user settings, and in accordance to the classification of that sound scene by the network (simple or moderate or difficult, etc).

Sorry to be long winded, but hopefully this gives a glimpse into how the nuts and bolts are handled in the More DNN, albeit still at a very high and abstract level at that.

I’m pretty sure that nobody who’s actually wearing Oticon More hearing instruments believes that they’re “learning on the job” (please correct me if I’m wrong!).

I hope this puts it to bed …

In that case I choose to take my audiology advice from my audiologist who always treats me with dignity and respect and does not attack me. Maybe I’m just funny that way, but I don’t like being attacked on a forum I come to for help and, hopefully, a degree of compassion.

I haven’t seen where you have been attacking either, all any of us can do is explain our individual experiences. We are all different and that being we all have different experiences, likes and dislikes.

While all of the back and forth in this conversation has been about whether the Oticon More lives up to its billing, I think it’s worth noting that Oticon isn’t the only company using aArtificial Intelligence in its hearing aids. Widex, Starkey, Philips, Audibel, and Whisper have all come out with hearing aids using AI in the past year. And Whisper does in fact say that it’s “brain” does in fact learn your listening environment and improve the longer you wear it.

Yeah Chuck, tis all open to conjecture…… No one will know if they like a hearing aid, until they try it! I have yet to see 2 audiograms that are identical, you get similar, but I have never seen one that is the same, they are unique to the individual. This uniqueness is perhaps exacerbated by our own little foibles, preferences or perhaps biased opinions. For instance, you probably wouldn’t ask a “Punk Rock” lover if he liked “Country & Western”, you would most likely get a mouthful of abuse? Similarly if you asked a “Country & Western”, if they liked “Punk”, you might want to switch off your aids So all types of sound, tone, bass, music, voice, are also a unique preference to the individual, coupled with your unique audiogram, then it is little wonder why opinions might be polarized…. This is perhaps “Why” some folks are downright obstinate, in their fellow HOH opinions, the sound I love from my hearing aid, you may hate or dislike intensely! And also perhaps why some AuD’s have a hard time fitting certain clients, no matter how well a hearing aid is fitted, you might not adapt to the way it sounds…… My 2 cents worth, cheers Kev

If my remarks were misinterpreted I apologize. The comment struck me as accusatory. My Audi by the way specializes in hyperacusis and I have the utmost trust and confidence in her.

@billgem said:

While all of the back and forth in this conversation has been about whether the Oticon More lives up to its billing, I think it’s worth noting that Oticon isn’t the only company using aArtificial Intelligence in its hearing aids. Widex, Starkey, Philips, Audibel, and Whisper have all come out with hearing aids using AI in the past year. And Whisper does in fact say that it’s “brain” does in fact learn your listening environment and improve the longer you wear it.

I never got the impression that anyone in this thread made any claim that Oticon is the only AI HA in town anyway.

With regard to your last statement about the Whisper brain actually doing the self learning, I would take issue with it. Below are a couple of screenshots from their web site that has “marketing-speak” about continuous improvement. I think the language they use is very ambivalent here.

In the first screenshot, my interpretation of their continuous improvements is via the regular software upgrades (as part of their subscription model). They never explicitly said that their brain unit does self learning.

Same with the second screenshot, they only said that they’re constantly refining …(blah blah blah)… This can easily be interpreted as refining in their labs. They never explicit say that their brain unit continuously learns and improves on its own.

Those are the only 2 marketing items I could find on their website about continuous improvement. But if you have some other info from them that actually said that their brain is self learning, I’d be interested to read up on it some more if you can share it.

@billgem said:

I think it’s worth noting that Oticon isn’t the only company using aArtificial Intelligence in its hearing aids. Widex, Starkey, Philips, Audibel, and Whisper have all come out with hearing aids using AI in the past year.

@billgem: There are just SO many digital devices today that AI and Machine Learning are ubiquitous. It would be difficult to imagine how any product as sophisticated as a digital hearing instrument could be developed without some form of AI being involved in the process.

Furthermore, to the best of my knowledge, nobody on this Forum has intimated that Oticon was alone in using those technologies in their HAs.

I would argue that it’s HOW Oticon is using the technology that diferentiates it from the competitors you’ve mentioned.

Sorry you having the delays. I have been discussing my ear molds with my audiologist, I am still having minor slipping of my left ear mold and now my right ear mold is slipping. At May convenience I will go in and meet with the clinic hearing aid technician and get new impressions and then new skeleton ear molds for both ears. My Audiologist says they are still getting a ten day or less turnaround on ear molds and aids repair from the New Jersey Oticon lab.

Thanks, Chuck, @cvkemp. I’m sorry too, because these 85dB receivers I’ve been wearing for the past 3 weeks have the correct wire length, but they don’t give me great comprehension because the HAs are fitted/tuned for 100dB receivers. Close, but no cigar!

The USVA is probably the single largest volume dispenser of Oticon HAs in the world, so you can be sure that Demant allocates a lot of resources to keeping them happy.

Our Canadian market is only 1/10 of the size, at best. My biggest issue is that it would cost very little for Oticon Canada to say "Due to higher-than-normal demand, lab turnaround times for custom moulds, electronic repairs, testing and calibration may be longer than is usually the norm. Please bear with us, and call 1-888-1234 if you require further information."

Yes, to do this would cost something, but probably just a pittance as a percentage of the grossly inflated prices we pay for top-drawer hearing instruments.

My VAC agent doesn’t get involved in stuff like this. It’s assumed that my audiologist of choice will be my advocate for the resolution of customer service issues like mine.

That’s why I’m so pleased with my audiologist - the entire staff has picked up the ball without my having to ask

Yes my Audiologist with the Oticon reps help made the decision to go the the 105db receivers for me even though the 100s would have worked. As you know the 105s are molded into the ear molds, and Oticon has two versions of the 100s one can be used with non acrylic ear molds and the other requires the acrylic ear molds. My Audiologist said for me to come by the clinic and have the impressions done then he will have my ear molds mailed to me. Replacing them is easy. I did it on my left one already.

@SpudGunner said:

I’m not happy with the service Oticon Canada is giving me

Isn’t it insane that the New Jersey office is really in close proximity to Ontario, and yet you are screwed by International Politics!
That is a sad state of affairs, Spudster, and we may be losing our Chief Oticonian in the process, as well!

I also find it hard to believe the Oticon facilility does not stock the 100dB receivers. They are the same ones as the OPN, OPN S series.
I mentioned the VAMC in Long Beach, CA stocks receivers locally for walk-in repairs.

@SpudGunner said:

Our Canadian market is only 1/10 of the size, at best

I can’t see why a mfg would not have a proportional stock of common parts on hand.
Receivers are probably the most common failure point.