Inside Starkey Omega AI+ Hearing Aids: Quad DNN, Better Speech and New Health Features

Hello everyone, and welcome to This Week in Hearing, live from Starkey Day in Minneapolis for the launch of their latest device, the Omega AI+. The similarity of the name with their previous device belies the amount of new innovation actually packed inside. Since this is Starkey, that includes new health metrics in addition to audiological improvements. First, I’ll get all the background in the morning sessions, then go deeper in conversations with several of Starkey’s key people. Let’s begin with Sarah Burdack, Chief Audiology Officer, and learn how the Plus version improves the patient experience versus the first-gen Omega. Well, thanks a lot, Sarah. I appreciate you joining me on the show. This is our first time, and I was really looking forward to it when we set this up.

Yeah, thanks for having me. There’s always a first time for everything. And we were just saying how many conferences we see each other in and engage quite a bit, but we’ve never formally sat down and had the conversation.

Yeah, so it’s my pleasure. And congratulations on the launch of the new Omega AI+.

Thank you. We’re just coming off of Starkey Day and it’s been really incredible and we could feel the energy and all the momentum in the room. I always talk about clappable moments, and I felt like there were so many. It was going on and on with all the great things we have with Omega AI+.

It was a great session this morning. It was. So let’s talk about the audiological benefits, right? Especially because it’s called Omega AI+, it’s obviously the next generation Omega. What are the key audiological benefits in the Plus version Compared to the original?

I think we’re always looking at intelligent hearing and how do you evolve that to the next level. And when you think about speech preservation, that’s always critically important, but then also speech enhancement. And we’re seeing better signal-to-noise ratios than we’ve seen in our previous technology. But we also look at all of the environments and the acoustics of everyday life. That all the patients are experiencing, and we’re seeing better comfort results, we’re seeing better loudness level results, and lower listening effort. So that’s so significant because again, we want people to really be able to focus on what it is that they want to hear. And the other benefit of what we’re doing with DNN is we want it to be always on. We want it to be always automatic. And from an audiological perspective, one of the things that I get the most excited about then with all of that is the patients aren’t having to engage with the technology as much. I’ve been around a long time, and I remember having to set up multiple programs in the hearing aids, and now that just really isn’t needed anymore. The technology is adapting, it works, and it’s a lot more seamless where patients don’t have to remember how to use the technology.

And I think that’s a huge benefit. Even me, who’s, you know, comfortable doing anything, when I first got fitted in 2018, I was constantly having to change. I had to customize programs and change them to get them just so. I don’t have to do that anymore, and that’s really nice. So how would the typical patient know what’s the optimum setting?

Exactly. They, they don’t. So with Edge Mode, we have that function in their hands based on what their intention is. Because again, I keep dating myself, there’s been times I have gone to locations with a patient to try to figure out how can I improve this fitting to know what their intention is. With what we’re doing with Edge Mode, they decide. They’re choosing to say, am I going to be focused on speech right now, or is this an environment where I just want to be comfortable? So they’re able to tell our AI systems what their interest in is in that moment of time, which I think is so fantastic because it’s an easy control. They’re not having to say, I’m in this specific environment. Is this restaurant or am I theater? You know, what do I want to do? They just choose, what do I want to hear better right now? And it’s going to use that AI and DNN system to do the snapshot for that moment in time and make adjustments.

Okay, so wait a minute then. I thought I heard that you have the capability at least of automating Edge Mode or leaving it on all the time.

We do.

Okay.

And I’m so glad that you asked that question. In our clinical studies— so my team manages all of our clinical validation— in our clinical studies, what we saw is there were individuals that had a preference for it to be always on, and there were some who didn’t. And so we’ve put that conversation into the hands of the audiologist. Most of the time at Starkey, we are not going to turn a feature on automatically without a conversation. So the default is like I was telling you, where the patient will be able to choose. Yeah, choose the device. Yeah, they can choose the device, they can set it in the app. But we did know some people liked it so much, and the question is obvious. It says, oh, if this is so automatic, why isn’t it all on the You know, that was exactly my question. Everybody says that, right? Why isn’t it on all the time? It is still because it shows you how difficult it is to know exactly what the user wants. But for those who really liked Edge Mode, a large number of them did say, I’d rather just have this on all the time. So it’s exciting that it is an option to have.

I want to just— and of course we had this conversation with the Omega also, the original Omega. But just to clarify, Edge Mode puts additional emphasis on speech than over running automatic without it. Is that correct?

Again, it depends on what they’ve selected. But yes, with Always On, we’re always going to be looking at speech preservation and enhancement.

So I mean, so I don’t have Edge Mode automatically, and I’m in a situation and I’m not using Edge Mode, but then if I put it in Edge Mode, there’ll be additional speech emphasis.

Yes.

Okay. And so then I guess a scenario I could see where it’d be useful to have it manual is, is let’s say I’m in a jazz club. I’m there primarily to listen to the music. I would probably choose to leave Edge Mode off, wouldn’t I? But if I’m having a conversation with people and the music is strictly background, I might be better off putting Edge Mode on. Is that like a good scenario for manual Edge Mode?

It is, but here’s what I will say. When you’re in the the app, in my Starkey app, you would be able to choose best sound, best speech, or comfort. And I think that’s an important distinction because you would quickly be able to try the different Edge Mode settings.

Mm-hmm.

Because you’re telling it what you want it to focus on, and I think that’s a key difference because it will have a different setting, if you tell it right there, you know, this place is just so loud, I don’t really care what anyone is talking about right now, you could certainly set it to focus on comfort and it’ll do that. It’ll engage a more aggressive noise management than the traditional default setting.

Okay.

Yep.

So it’s, yeah, I think I actually like the way you’re approaching it. Some people want it on automatic all the time, And some people wanna be able to choose and play.

To choose what their focus is. But what we’re seeing, you know, back to this fewer adjustments, we’re seeing a strong preference for Edge Mode even over their personal main memory. And that tells me it’s working. And that’s back to some of this automatic nature of not having to play around with the technology. You are someone who totally knows how to use all of the technology at hand, but a lot of people don’t want the technology in their way. And I think that’s so nice from an audiology perspective that people are able to walk out with a fitting and don’t have to engage unless they want to.

And that includes me, actually. I like just going from scenario to scenario, not having to do a thing. Just not have to pull out the phone every time.

Exactly. And I love that too. A lot of times, you know, I’m glad our app is rated so high with my Starkey, but I’m also okay if people aren’t using it because it means they don’t have to.

Right.

They’re just hearing better.

And I’ll give you props because I love the audible feedback.

Yes, yes.

You don’t get lost in the volume control, for example. It tells you when you’re at the default setting in words, for example. Or if I, if I put it into restaurant mode, it says restaurant mode, right? So that’s, I think that’s a really good part of the interface.

We’ve come a long way with that. And thank you for bringing that up Because it’s harder to do than you think in multiple languages all over the world and having these very recognizable, meaningful inputs in the day-to-day interactions with the hearing aids.

Yeah, no, I like that a lot. Another big emphasis on was the feedback management. Let’s talk about the feedback management. What have you done differently in this one that makes feedback management better?

Yeah, first of all, we’re talking about intelligence overall with Omega AI+. We’ve done a lot with our feedback management system. And one of the ways we’ve been able to think about feedback is looking at what we’re doing with our dual compressor as well. Because if you think about this compression system that has to be understanding the envelope of speech, why can’t we use it for something like feedback that can be as fast, as dynamic, and make sure that we’re tracking it across a broader range of the frequencies? And one of the biggest areas that we’ve changed is looking at how we manage mid-frequencies specifically. We tend to think feedback is only in the high frequencies. But with this, we started seeing with all the dynamic nature of these features, a fairly significant shift where we do have feedback more in the mid-frequency. So we’ve added a lot of logic there where we’re able to say, okay, what do we need to do there? How do we need to fully understand the dynamics of the feedback path? In the patient’s ear, and this is where the audiologist knows this better than anyone. What is the style? How deep is it in the canal? What’s the vent initialization? What are all of these features to then make sure we’re collectively understanding what’s going on with that unique patient? I always think it’s kind of like a fingerprint. So are the acoustics and the feedback path for that patient. So we’re applying all of that and we’ve received great results. We’re showing an overall 11 dB improvement in our gain margin.

Compared to what?

Compared to our legacy systems.

You mean the previous generation Omega units?

Yes.

So you get 11 dB more headroom?

Yes, yes, which is great. So that combined system, this mid-frequency management is a 21 dB difference peak to peak, and that was with audible feedback. So that’s really big in that frequency range to have that big of a change.

So you can actually fit people with more severe losses with a smaller receiver in the RIC style then?

We hope, we hope, yeah. You want— we’re getting a lot less feedback and artifact. I mean, that’s where I would look at sometimes with more severe hearing losses, It is the feedback artifact, the chirp.

Yep.

You know, where the feedback canceler sometimes wanders off a little bit, as I will say. We saw significant improvement there with no audible chirping feedback, and that difference was 32 dB. I mean, these are really big numbers. And again, this was our own comparison to our legacy devices Yeah, so you’ve improved it. Yes.

Yeah, you’ve just given the hearing care professional more headroom.

Yes, which is great. And what I like to see too, I always talk about being really good at getting to the best, best fit as quickly as you possibly can. And this was showing really such a decrease and an improvement of lowering feedback at the initial fit.

Okay.

Which is great. That just saves all of us time.

Okay. And then you’ve done a lot of other ancillary things too. Like there was some mention of the new charger. What, to me, charger seems like old tech. What’s new about the charger?

The charger is always going to be something that we work on, and here’s why. Everything is the product. And what do I mean by that? The hearing aid is the product, the software is the product, the mobile app’s the product, the charger is the product. And one of the things that we’re seeing and recognizing is that the patients using our technology want a little more information from the charger. So we’re making the LEDs on the charger smarter so that they’re really indicating when it’s charged, when it’s fully charged, All of those different meaningful inputs that are helping a patient. The other thing that we’ve worked on is the case design. People tend to like a lid on the charger to protect the devices. With the new design that we have, we put some ridging in place to really make sure that the devices stay strongly seated in the charger.

Okay, so in other words, because you had a lid before.

We did have a lid before.

But if I’m— it’s in my suitcase or something, the hearing aids— might pop out of it and not be charging.

Exactly.

And now they won’t.

How do you make it portable? We don’t want to assume people always just keep this on their nightstand. How do you make it portable? How do you keep these devices so that they more consistently charge? As you know, we’ve really led on 51 hours of battery performance.

Which you’re still maintaining, correct?

Which we’re still maintaining, yeah, with everything that we’re doing with Omega AI+. But it is an evolution of the charger. When we hear something from a patient, that they want, we’re going to do what we can to get it implemented across really the entire scope of the products.

Okay. Okay. And another thing that, that was interesting was hearing aid left behind. Now, there are other devices, earbuds and hearing aids both, who will do Find My. But what’s the difference between Find My and hearing aid left behind?

Yeah, it’s going to notify you before you have to do Find My.

So it’s proactive. If I walk out of range of the hearing aid, Yeah, so it’s going to say, hey, wait a minute. Okay. And I have one more question for you, and that is sleep metrics. Now I’m going to talk with Achin about the sleep metrics in more detail, but I’m just curious from your point of view, and I should precede the question by saying what form factors are actually available in this new platform? And then I want to ask you which form factors you actually expect people to sleep in.

I think this is another area that’s very, very individual. We’ve had a lot of conversations about this too, but what we were seeing is there’s so many devices then that people are wanting. You’ve heard us talk about the, the ear is the new wrist. Can we think differently and have one device for somebody that’s accomplishing all of these things that they’re interested in? So we recognize this isn’t going to be for everybody, But we also were hearing some people just want one device that does all of these things, and they are willing to sleep, you know, in that device. Dave talked about the turbocharging, and, you know, having the longer battery life matters with this because if somebody is wearing their devices longer to get that sleep information, it would be helpful. And maybe it’s when they’re traveling, or, you know, maybe they’re looking at a more specific use case for them.

Right.

But you’ve heard us talk a lot about this evolution of hearing health and obviously all of the connected comorbidities. And everywhere I go, everything I read about sleep is it’s as important and you really can’t function well without it. And that’s why we said this is the next evolution of, you know, let’s—

Sleep.

what this looks like. Let’s see the use cases for this. And similar to what we’ve been doing with fall alerts and detection, we started with that. Then we said, okay, now let’s add in a balance assessment. Now let’s add in exercises like Balance Builder. And I feel like sleep for us will have a similar evolution.

Yeah. Okay.

Gotta start somewhere. And, you know, and then—

So you have to get people to use it though, right? Because you have both, you both have in-the-ear and RICs now.

Yes, yes, we do.

And I just, in my own imagination, I can’t imagine sleeping in a RIC routinely, but I could imagine sleeping in an ITE if I wanted to use the sleep metrics.

And I think it will be very, very specific to the user. My team, as I said, does our clinical research. We had several of those individuals sleeping with the devices on, you know, just again, making sure how does that feel? You know, what does that look like? You maybe heard that we’re able to track which side, so we needed to make sure we were accurately doing that and, you know, how that dynamic affected the person’s sleep.

Well, I’ll have to try it. I mean, these are still the Gen 1 Omegas, but just for the heck of it, I’m curious what it would be like to sleep in a RIC. So maybe tonight I’ll just leave it in.

I encourage, you know, of course we’re like, everybody needs to be trying Omega AI Plus. That’s why we’re here. And it’s fun to just, again, see all of that progression because we care so deeply about hearing. We care so deeply about all the health technology and the health monitoring that we’re doing. And then everything that we’re looking at from a personal—

Well, you guys have taken the health monitoring to like whole new levels.

No one else is doing this in our technology.

It’s amazing what you guys are doing.

And the assistant technology too. That we have is really meaningful.

Yeah, no, you guys are doing some really interesting things.

I agree. I love it.

Is there anything else people should know about the Omega AI+?

I think we’ve covered everything, but I really believe this is a leap forward. All of the, the data science we have proves it out. And I will say it’s so hard in brief meetings to go through all of this data, but we do on our Pro site have all of our white papers, all of our peer-reviewed publications. They are all posted. Since you can order the devices now. So if anyone wants to find out more, I know a lot of people want to dig deeper, get into all of the numbers and how we did things. It’s all available on our website.

Well, congratulations again.

Well, thank you.

And thanks a lot for spending some time with me.

Yes, it’s been wonderful. Thank you.

You’re welcome. I always enjoy talking to Achin Bhomik, CTO, because he is so passionate about how neuroscience can be applied to hearing devices. Let’s hear from him how the 4 DNNs work together and talk a bit more about our health ecosystem overall and the new sleep tracking in particular. Well, thanks a lot, Achin I really appreciate you spending some time with me. Congratulations on the launch of the Omega AI+.

Well, thank you, Andy, to sit down and talk to you again. Thanks for coming to our launch event.

I think this is the 3rd time we and I— you and I have talked on launches, and it’s always a great conversation.

And I enjoy it as well. You have very insightful questions and such insight on patients and the industry. So it’s always a pleasure to talk to you.

Well, thank you. So you said actually something in the sessions this morning, which are really good. The brain is an ensemble of DNNs. I really love that. That’s such a great description, which kind of leads into what you’re trying to do in the hearing devices is create more than one DNN that acts on different parts of the speech scene. To give an overall better hearing experience. Now, the name itself implies this is actually not a new hardware platform, correct?

Well, it’s a new processing platform. So we had to make improvements in order to accommodate the new Quad DNN architecture. So we are not going into the details of exactly, you know, what elements are new, but it’s a new processing platform.

Okay. And that’s because I think this is a new element of hearing devices today with the ability to create DNNs.

Yes.

You know, in the old days, if you want to do something different, you created new hardware. And now you don’t necessarily have to do that. We’re learning so much more about how to create better, more efficient models that act on speech in different ways. And that’s why I was kind of curious about what you’re doing there.

You do have to tweak and optimize And it doesn’t mean you have to tape out necessarily. So that’s what you mean? That’s what I mean.

You haven’t taped out anything in here.

It’s a flexible platform that allows you to vastly expand the processing capabilities, speed up. You saw the results, 2.7 times faster. And you can accommodate many more DNNs running in parallel. So compared to Omega AI, it’s a jump up in competition.

You had mentioned 2.5 times faster. 2.5 times faster than what?

Than co-processor.

Okay, so having the DNNs running on the main chip is faster than running in a coprocessor?

If we attach a coprocessor, you move the data back and forth. You decide at what conditions do you want to fire up that extra chip, bring the signal back, combine with your main audio path and send out, there’s going to be a delay between the two.

So you mean— okay, so when you say faster, you’re talking in terms of latency?

Latency, right. So we actually published a pretty detailed white paper on this, on how this is measured relative to the main audio path, to the AI-processed path, the delays that are measured between integrated platform like ours. And in fact, we have compared with all the other hearing aids in the market. Let you read the white paper.

In terms of latency. So you’ve measured your latency and you’ve measured other people’s latency. What’s your latency?

You’ll see that in the white paper, it’s less than 5 milliseconds overall from sound collected to sound output.

It’s pretty fast.

Right. And note, there is— because we are in the main audio path, we literally have 0 milliseconds delay between the main audio path and DNN processed because the way we architected our sound processing architecture, Every sound is processed through deep neural network as it goes out. It’s not handed off to some other place, brought it back, and combined with the amplified path. So there is 0 milliseconds delay between multiple processing elements. They are in the main audio path.

Okay.

That’s a unique approach that we took in our signal processing.

Okay. And then let’s talk about the 4 DNNs and what they’re doing because you had 3 in the previous one, but it looks like this is a completely different architecture.

Right.

You have 1 for directionality.

Yeah.

And does that mean tracking speech as it moves? Is that what you’re talking about?

Great. So before I get to that, you mentioned the brain and how we explain it. For your audience, I want to go back because if you quickly go into like, is 1 DNN better than 2 DNNs, better than 2 or 3 or 4? But The approach is this, you know, the way I look at it, as you know, I’m still kind of relatively new in the hearing aid industry, but not quite new. This is my 9th year, right? But I worked on AI systems before, and the way I look at it is, and you appreciate this, computer vision area has always led the adoption of AI. Hearing aids were not— I mean, you had some classifications of environment with machine learning, right? That’s not There’s no deep neural network. But borrowing from the page of how computer vision systems allow you to sense and understand with imaging devices, navigate, interact, you have to go back to the human sensory perceptual system. What functions do the sensors perform? And then what computations are happening in your brain? So the way I explained this morning is, I mean, it’s all not new. This is very well known in neuroscience. Yes, you have to make up for the losses in the ear. Of course, we do that. Connecting back to biology, you’re making up for lost outer hair cell amplifications. To do that, you need wide dynamic range in the chip. So we make sure that we have— we start off with a very wide dynamic range. Next, as the signal goes to the brain and and things get processed. What do we really do? Well, parts in your brain is helping you suppress noise and enhance speech. It’s pretty clear. If you sit in a cafeteria and you’re just paying attention to everything, the environment is noisy. Yet, with healthy hearing system, you can pay attention to somebody and you literally can ignore things that— everything else that’s going on. So brain’s performing an amazing computation.

Yes, I remember the pre-hearing loss days when I could pay attention to the conversation behind me, right?

Yes.

focus on that versus everything else, which is harder and harder to do once you have hearing loss. It all sounds like noise to you.

So the good reason that you do that with, with the deep neural network, the task of speech enhancement and noise reduction, which hearing aid industry has done for many, many, many years, but what we found collectively now, everyone in the industry is doing it because the neural network approach is just inherently more robust. You don’t have to handcraft for certain specific environments, but neural network has the emergent property where if you train it with a large enough dataset, it does really well out in the wild when it’s encountering new info. So suddenly that was the first wave of deep neural network adoption in hearing aids. Now many companies are doing it. But then if you look at, again going back to human auditory system, its job is not just suppressing noise and enhancing speech. Well, its job is to localize sound and help you with directional system. You may want to be aware of everything that’s going on despite noise reduction and speech enhancement. So that was the reason why we invested in a deep neural network approach for directionality. That’s the way I explained it. It’s almost, in our testings as well, it’s almost got a predictive nature to it. And that’s the beauty of neural networks because of the ways that it analyzes the multi- streams of data, acoustics and motions and everything else, it almost feels like it can predict what will happen in 5 to 10 milliseconds from now. It can detect an onset of something, so it’s able to move much faster than statistical processes that we used to have in directional systems. So we are having this amazing benefit of DNN directionality. DNN spatial awareness? Yeah, it’s a 3rd DNN system that’s enhancing important sounds around me. So, you know, these people who are using Omega AI and now Omega AI Plus, they tell me that you hear more but not annoyed. That’s because of the spatial awareness DNN that’s presenting to you, which otherwise would have been suppressed, important sounds. What’s an important sound? Well, that’s the beauty of neural networks. You train a good AI system, it’s able to determine that without you having to handcraft something. So the 4th DNN— first of all, why do we care about adding another DNN?

Why?

Again, going back to the— this is not— we have been working on it for years now, but human brain has this amazing ability, as I mentioned this morning, about loudness mapping.

Mm-hmm.

In hearing aid parlour, we call that compression. So the way that compression is done in hearing aids is also you’re having classic signal processing schemes in figuring out when you invoke the slow compressor, when you invoke the fast compressor, and then you make a compromise. And what we found in our work is, again, if you invoke a neural network, again, it’s not surprising to me because that’s exactly how brain does it. The way brain does loudness mapping, you know, I was pointing to the circuitry in the brain where it happens. Those are all neural networks, right? So when we experimented with invoking a neural network-based compressor, we found great results. So that’s what informed and inspired the fourth—

Well, in that Dual compression system, right? Is that the 4th DNN you were about to say?

Yes, that’s the DNN dual compressor.

Yes.

Because it’s not just a dual compressor. I mean, many hearing aids also have dual compressor. You have a fast one and a slow one, and then you look at the signal statistics and decide which one you want to use. But it goes— it makes mistakes. And the neural network approach minimizes those mistakes because it’s able to more accurately analyze the acoustic environment and dynamically allocate resources between slowness and fastness. You don’t even have to know what’s going on because it’s trained with acoustic environments.

Well, and when you showed this this morning, like you had both compressors running, it looked like almost dynamically in real time.

Right.

So how is that actually working? How is the slow and the fast compression used together?

So back to the brain again. So brain doesn’t choose to switch from one to another. It works in parallel because you have multiple systems. First of all, if this environment or, you know, there is always a need for compressing the sound to my biological dynamic range. But you don’t want to do that with a slowly varying loudness mapping system.

Mm-hmm.

It’s always happening in our brain. But right now you’re silent, I’m speaking. The moment there is fast-acting signals and my brain is saying, look, I spot a speech, and in order to help you understand this the best I can with the lowest amount of cognitive effort or listening effort, I need to dynamically turn on a fast loudness mapping. For speech, it’s a very different strategy than slowly varying environmental sound because the task here is consonants, as you know, are Smaller energies packed into those consonants, but they’re fast moving in timescale. So the loudness mapping function, the biologists call it, equivalently compression technology in hearing aid, needs to be respectful of that and then not mix the two. Because then if you invoke fast compression when there is no speech, you’re going to have artifact. I give an example of keyboard clicks. And then you will have a little— I call it pumping, that’s how I explain it— like a little bit of a pumping sound that is not there, but you hear it. We used to say that’s inevitable, these are hearing aids. But it’s amazing that with a neural network approach, it’s copying the way brain does it with the neural network and we don’t have those artifacts.

Then the 2 compressors are acting differently on the different parts of speech, correct?

Slow compressor is always on.

Okay.

Yeah, because even if there is speech, there’s also other sounds around the world. So I want this— what do you think about it? The slow compressor accounts for your audiogram. So when I said your soft sounds are audible, you have to amplify the soft sounds to make sure—

So slow compressor is always on, this is typical.

Yes.

But then you’re actually acting dynamically on the different parts of speech as I’m speaking. Do I understand that right?

Yes.

So as I transition between vowels and consonants and so on, the fast compressor is actually acting differently to help enhance the speech.

Yes, and it’s able to quickly disappear and come back on between the snippets of speeches. That’s the reduction of the artifacts that we notice with 100% people testing it saying, yeah, it’s like, you know, I don’t notice the typical hearing aid artifacts.

Okay, okay. And then the other thing that was really interesting to is, I mean, every generation you’ve added to the health metrics.

Right.

And now you’ve added sleep.

Right.

So how are you actually deriving sleep? Obviously, you’ve got the accelerometers in the hearing aids, and of course on the head, so you have head positioning. Is that what you’re using, or are you doing anything else too?

Thank you for asking that. So this year now, with this product, we’re adding sleep tracking. So Hearing aid has already got multiple sensors in there. You have IMUs, you have microphones, right? So I will— we have a white paper, you can read up. But we found interestingly, not only that it’s highly accurate sleep tracking system, again, head, you’ve got benefits, you’ve got 2 of them, right? But a unique advantage to doing sleep tracking with hearing devices is that I can tell you what side you slept on. You cannot do that with your ring or watch because wrist could be anywhere and I have no way to tell you, are you a back sleeper, are you a side sleeper? And it turns out, you know, my friends at Stanford Sleep Labs, they will often want the patients to come and sleep in the lab with instrumented settings. So then they want to know which side you’re sleeping because for certain type of conditions, the recommendations of what side you should sleep on is different. Whether you have sleep apnea or you have gastrointestinal issues. So it turns out that it matters which side you sleep on, and we don’t even know what side I’m sleeping on. So we are going to provide that insight in the hands of the patient. They will know. We’re not saying everybody should now start sleeping with their hearing aids, right? Well, first thing is like, when do you charge the devices? You heard Dave talk about that. It’s like 10 minutes. of charge, 3 hours of operation.

Yeah, right. So you can easily top them back up again because they flash charge. I’m actually curious how many people will sleep in their hearing aids. And I think it depends between the in-the-ear model versus the RIC too, whether people will actually be comfortable. Yes.

So by the way, we have been— so for a couple of years, as you know, it takes time to develop this technology. In our conversations, Mr. Austin himself, some of the audiologists just say, yeah, we often see patients who sleep with their hearing aids. Why so? Because they are afraid of missing out on important auditory events. You know, there might be an alarm or somebody might be knocking at the door. People sleep with their hearing aids, apparently. Not everyone, but some do. But more important, you know, this is what Mr. Austin would often say, when we were developing fall detection, not everybody needs it. I’m not worried about fall, but my mother should be because, you know, she’s frail and she needs that. Even if 2% people, 1% people benefit in terms of a feature, we will develop it. Not everybody will benefit from sleep tracking, but that individual who’s got sleep apnea or sleep disorder, and they can only know if they’re sleeping right by going to a sleep lab— yeah, happens once in like 3 years. Now every night you could test, and then when you want to test it, use it. When you don’t want to test it, don’t use it. But you got the technology. So we are not always asking, is it 100% use? If it’s valuable for health, we want to do it so that if someone who needs help or can use it, they can benefit from it.

I’m really intrigued by it. And although these are still the original Omegas, I’m just going to give it a reality check by sleeping in them tonight.

You will see if it’s comfortable sleeping. You got upgraded to Omega i+?

These are the original ones, not the AI+, but they’re the same size.

Yes. So I’m just curious if it’s comfortable sleeping in a RIC hearing aid, which I’ve never I would love to hear from you because some would say, I sleep with them, they don’t fall off. But some others will say, I don’t have good retention, it falls off. For me, I have no problem because I also have these amazing custom molds.

There you go.

To remind your audience, I have no hearing loss. For me, I’m one of those individuals who say it gives me superhuman hearing. I prefer to have them on than not, even though I have normal audiograms.

You use the voices to say this.

that every time I’m getting a phone call, I don’t have to, oh wait, let me put my devices on. Or I’m just walking or sitting at the airport listening to a podcast or your interviews, and it’s always there. So I definitely benefit from it. And also just the signal-to-noise ratio improvement. It makes the world of hearing less effort, even for somebody with normal hearing. It’s an amazing thing. Why not, right?

Yeah, no, I could see that. Even if you’re really running no gain in them, you’re still curating the sound.

Yeah, because you’re expanding the SNR even if you’re not providing any gain.

Right, right. So you have more listening comfort over the course of the day. So anything else I should know about the device?

Right. And to end, you also heard me on stage saying I would rather stand here today and talk about The next product, because as you know, for research and development, for engineers building the product, you finish and then you take quite a bit of time for clinical validations, manufacturing ramp before you launch something, right?

So what’s next?

So the next, I hope that we get to sit down together again soon because I’m so excited. You know, you saw me show that exponential improving curve. That requires— in order to be on that curve, it requires significant improvement in computational capabilities, new algorithms, new architecture, more and more copying of the human brain. So let’s sit down soon again and we’ll talk about the next product.

Looking forward to being here a year from now.

Yes, indeed.

Well, thanks a lot. Appreciate you spending some time with me.

Always a pleasure. Thank you.

And congratulations again.

Thank you.

I appreciate that Brandon Sawalich has always made time in his busy day to share with me his vision for the future. This time was no different. And our conversation was as interesting as ever. Well, thanks a lot for joining me, Brandon. I know it’s a very busy day. I appreciate you spending a few minutes with me, and congratulations on the launch of the Omega AI+

Well, thank you. We’re, we’re excited, and you know, to be here, and you feel the energy in the room. We have like 600 people in there, and, and we had a lot more on the live stream than we’ve had before. I mean, thousands from around the world just wanting to know what’s up with Starkey. I think, you know, over the years our credibility and our proof of innovation is there. And every time there’s a Starkey Day or something new from Starkey, everybody wants to know what’s going on.

Well, there was a terrific program this morning. Very well done.

Yeah, it was fun. Thank you. Thank you.

You know, it leaves me, and of course a generation or two, I said the same thing, right? If I think about how AI has evolved, it started out being straightforward, reduce the noise. And now we’ve evolved to where you’re running multiple DNNs, curating the entire audio experience with an emphasis on making sure people are environmentally aware and yet able to understand speech with minimal effort. Yes. Where do you actually go from here? Is it— are there things left undone, or is it now a matter of improving what you’ve already created?

Oh, it’s a matter of— it’s going to be a rocket ride from where we are now. Our advanced research team and Achin and our engineers is— I know what’s next. And it’s really good right now. It’s going to make this look like a toy. And it’s really for the— because job one at Starkey is better hearing. I said that this morning. It always has been. Because people, that’s why they go and see the professional. And that’s our job. That’s our focus. But I think what’s next, because to me, and again, speaking historically, is our technology roadmap and what we’re doing, that’s set. I know what it’s going to be in the next 5 to 7 years. I think what’s next is another area that’s kind of underserved, and that’s the care. And that’s the tools and the resources for the patient, the patient experience, the patient journey. Because you’ve got the technology. And you and I both know that people aren’t going in just to transact or buy something off the shelf, or they want to understand what it is, but they also want it easy. And, you know, from the moment the person walks into a clinic, anybody’s office, what’s that experience? What are the tools and resources that we need to be thinking of to help them along that journey and continue to lessen the stigma about hearing aids? You know, that’s the downfall of OTC. That’s, you know, there’s You know, some good qualities there, of course, we all know it. But I think that it’s the confusion it causes because with the great technology now, you know, you don’t want people to be swayed away from it or have a bad attitude because, you know, they saw something that doesn’t represent what we’re doing and where we’re going.

Just, I love— I loved going into your museum where you have all the variations of beige banana, right? I actually have a picture of myself wearing one of those.

Oh, every time I see, you know, there could be a national— we all see it, national article or and they have a hearing aid that’s like from 25 years ago, and they represent that as today’s technology. It is so far from the truth. We know that. But at Starkey, you’re going to continue to see continued evolution of our Starkey AI for hearing in noise, better hearing. Of course. I mean, it’s going to be very, very good. You’re looking at simultaneous translation. You’re looking at some other health and wellness. And, you know, the personal assistant where The product continues to help you in other ways outside of just being a hearing aid.

Well, and you mentioned the health metrics. I mean, what you guys are doing in the health metrics is absolutely amazing, how much you’re able to leverage with the sensor package, right? Which is basically inertial units and microphones, and you’re getting so much health data out of that. And you’ve now added sleep.

Right. So think of all what we— from Our Quad DNN and Starkey AI, you’ve got our health and the other features. And this is our, you know, of course, our Starkey AI and our processor, 51 hours of battery life. I mean, we’re balancing and we’re running a very efficient technology for the patient. Yeah.

So what comes next? I mean, sleep was really intriguing. And I told Aichin when I asked him how it was done, These are still the original Omegas. I’m just going to sleep in them to see if I’m capable of sleeping in a RIC. I’m curious if I can see—

Some people do. I actually got a letter last week where somebody was telling me about it because of their tinnitus and some other things. And yeah, it happens, but it’s there if people want it. And that’s— it’s people want options.

So what comes next to there? What else do you envision in the health metrics, and how are you working to coordinate that into Like an overall— like giving real insights to people versus just throwing numbers at them, which you’re already doing.

Yeah, no, we’re already giving the numbers. We provide what they want. I mean, the fall detection, when we did it, everybody’s like, oh, why would you do— nobody— this— it’s a hearing aid. Well, I like it when they say that because you’re doing something that people aren’t used to. And I still get every other month or numerous times a year letters from patients or a family member Saved somebody’s life. Our hearing aid, our fall detection, saved somebody’s life. For me, I think if you— we’ve all been talking so much about healthy hearing and hearing health, it’s the mental aspect, the mental wellness, somebody’s overall— because whether it’s loneliness, depression. So think about voice detection that could pick up somebody’s overall state of mind. Are they depressed? Are they excited? You know, somebody that’s in assisted living and the family member wants to make sure they’re taken care of, and, you know, they could tell if the patient’s okay with it. Are they in a depressed state of mind? Are they doing okay? Are they lonely? How much are they, you know, talking? So there’s a lot of other areas, and it’s the stuff that we take for granted, but when you when you’ll hear about it, then you’re like, oh, that makes sense. But so it’s connecting the overall mental wellness to the overall satisfaction of better hearing.

So using vocalization, in other words, to drive— Yeah, productivity and your own voice. You’re heading more in that direction, right? Because I’ve seen companies do things like that can identify a person’s emotional state for someone in a call center, for example.

Right, right, right.

Or there are people now working in the health arena where they can diagnose certain conditions by changes in vocalization patterns. Are you heading in that direction too? Yes.

Yeah. Yes. That is the direction. Yes, I can say that. I have said that before. And because I like talking about the future, I like talking about what we’re doing. I like talking about why can’t we do blood pressure, heart rate monitoring and stuff in the ear with our products. So there’s a lot of opportunity, but it’s also getting the horsepower, per se, the processor, the memory, and what we need to to do certain things. And it just, if it doesn’t, you know, takes a little bit longer, but we’re close.

Yeah, so presumably you’re a few years ahead in your thinking about what we want to do and what you’ve already modeled versus the processor power that’s available today. But you know Moore’s Law is still working.

Yeah, ’cause I know what we’re doing now and then what, you know, Achin and the team are working on. 7 years from now because we know what’s, you know, the possibilities are. It just takes a little bit to get to the possibilities. It’s not because we can’t do it. We’re just limited by what we could do based on technology. But we’re working on it just like everything. You get the innovation, you get the advancement, and you keep making improvements until it happens.

Well, I’m looking forward to seeing all what you do in the next years, audiologically, of course, because I’m a person with hearing loss, but also on the health side because It’s just really exciting, all the things you’re doing today. And I can imagine that— I know all the things that are possible.

It’s coming to Starkey.

Just watching it come to reality.

Yeah, coming to Starkey. And that’s the fun part is reality. We’re the Walt Disney World of hearing because it’s just, it’s a playground. You come over there and tour our tech center and our R&D, and there’s so many different projects going on. that just spending some time on each one, it just really motivates you. It motivates you to push harder and go further.

Well, I love where you’re going, and I really appreciate you spending some time with me. Congratulations again.

Thanks for being here.

Thank you. Welcome to my home away from home. If industrial-grade brutalism is your thing, this is the place for you. But why did I invite you here? These are the original Omega AIs, so I won’t get data. But I am curious if people can sleep well through the night wearing RICs in their ears. Let’s find out. You’re still here? I’m sorry your patience wasn’t rewarded, but as a side sleeper, it only took 2 hours before the press of the RICs against my ear made it sore and woke me up. I suppose more people will find it comfortable sleeping in the in-ear design rather than the RICs. But it still makes me curious how many people will actually sleep in them and take advantage of the sleep tracking features. Looking forward to finding that out in time. In the meantime, thanks for watching or listening to this edition of This Week in Hearing.