My read on Starkey was that they’d leapfrogged the More/Real with their AI: at least some proof of this exists with the real world performance of Genesis. I was actually thinking that if Oticon had either established a major step with true beam-forming to combat noise or a self educating AI there might be something worth talking about.
The change of the product cycle length is very interesting though.
Edit: The Starkey roadmap also indicates this AI development : 10 years ahead in 5 years - According to Brandon Sawalowic. (Sp?)
If you don’t want to hear about my bad experiences with Phonak, maybe stop trashing competitors when you’re clearly financially married to Phonak, hmmmm???
Just a thought. I certainly don’t appreciate these comments. Considering my bad experience with a financially-married-to-Phonak audiologist…looks like all of y’all are exactly the same peas in a pod.
I’m out of this discussion if all you’re gonna do is Oticon bashing because Phonak can’t match or even come close to their quality.
All y’all Phonak audiologists give audiologists a bad name. All you are are hearing aid pushers and brand salesmen. That’s it. Not real audiologists.
Marketing hype is the nature of the beast. All companies who market their products are guilty of it, and people love to shoot it down. But when the rubber meets the road, if the company’s products keep people happy and sell well, then it’s the merit of the product that sells, no longer the hype.
Regarding whether people understand the marketing “speak” of a product or not, I agree that it’s usually vague and make people scratch their head and say “huh? What does it even mean?” When it comes to more detailed description of a product in a whitepaper, it depends. Some company also keeps their whitepaper pretty vague, while other companies may share more details in their whitepaper.
Because of my technical background in electrical engineering, I personally find Oticon whitepapers sufficiently easy enough FOR ME to understand. If people with less technical background can’t understand what Oticon say in their whitepapers, it’s perfectly OK, too. Afterall, it’s not designed for non-technical people anyway.
Specifically about the DNN stuff in the Oticon whitepaper, I have no background in AI, even though I am a technical person. so I still had to invest significant time on a series of several YouTube technical videos that explain what DNN is and how it works first, before I could understand more easily what Oticon is saying about their DNN. But not understanding a whitepaper that glosses over how an AI DNN is implemented doesn’t automatically mean that it’s full of hype and nonsense. You may not understand it, but others may.
At least, unlike the Starkey 2-page whitepaper on their DNN AI, which I feel they don’t share enough details about how they implement their DNN and their new platform architecture, Oticon whitepapers are usually much more detailed (theirs on the DNN is a 10-page one) that gives the audience a clearer picture of how they implement it.
At the end of the day, the success in a product’s sales is the final proof in the pudding, regardless of how much people accuse them of hyping up their product. As long as they perform to expectation, people will buy. As @Um_bongo puts it in post #21 in this thread, he thinks that proof of Starkey leapfrogging the Oticon in AI is with the real world performance of Genesis, regardless of whether they have good whitepapers or overhyped marketing or not.
It comes down to performance that will lead to good sales. Although @x475aws and @brec here both think that the Oticon is too hyped up, they still ended up buying Oticon aids and are happy with it for the most part anyway.
In fairness, they have to give us a month or so notice for a launch if they want us to take off midweek clinic days to attend. Also I’m guessing the timetable is set like this to get the European launch up and running ahead of the AAA: Starkey don’t go there as they do their own Expo.
My notion is that it’s a strategic plan to make people think twice before encouraging clients to drop their dollars on the Genesis. Starkey has shown their hand already (it’s quite a strong one - like an 18 in Pontoon), even if Oticon can’t stack the pipeline in the short term, a conscientious independent dispenser would have to ask the question: what if Oticon has a 20?
New LE Audio/Auracast devices are coming out every week now, and bunch of stuff got introduced during CES. It’s not only Samsung’s phones but also TVs, tablets and other devices. Google’s and Sony’s phones are also LE Audio enabled and LE Audio Windows laptops are being demoed.
Apple probably will get on this train last, if ever, because they want to keep they walled garden walled and LE Audio is an open standard which is against their business practices. They only adapt open standards when it’s financially beneficial to them or are forced to.
Eventually they will be forced because Auracast is posed to be a t-coil replacement and it would be insane on their part to not support an accessibility feature (or EU will mandate it and they will have to, similar to what happened to chargin cables used by iPhones)
I should have been clearer about the kind of material I was referring to when I said,
I often scratched my head in reaction to various sentences, as in “huh? What does this even mean ?
I wasn’t referring to technical language. I’ve known what a “deep neural network” is for decades. (BTW, “deep” means more than one “hidden” layer between the input and output layers – not necessarily what a layperson would mean by “deep.”) Example of consumer-facing hype:
Thanks to recent advances in hearing aid technology, Oticon products work with the brain to process sound the way it’s meant to happen. By working with the brain, rather than just turning up the volume, Oticon hearing aids provide a more natural and pleasant soundscape, leaving you less tired and better able to remember more of life’s important conversations.
Oticon hearing aids give your brain access to the full sound scene – to support your brain in working more effectively
They help to ensure your brain receives the quality sound information it needs to remain healthy
This unique approach is called BrainHearing™, and it’s what makes Oticon hearing aids unique
“working with the brain” – what does that even mean? “provide a more natural and pleasant soundscape” – what does that even mean? “give your brain access to the full sound scene” – what does that even mean? “quality sound information” – what does that even mean? My guess as to what they’re talking about is that their noise discrimination and suppression is more selective than competitors’ because, y’know, they use an “on-board” DNN (professional-facing page):
Taking our BrainHearing technologies to the next level, we’ve created a
hearing aid that works more like how the brain works, because it
learned through experience. Oticon More™ is the world’s first hearing aid
that uses a fully trained, on-board Deep Neural Network (DNN) to process
the sound scene in a more precise and balanced way. This precise approach
provides the brain with optimized input from all types of meaningful sound,
giving patients better speech understanding with less effort and the
ability to remember more.
“more precise and balanced way”; “precise approach”; “optimized input” – what does that even mean?
Yeah so it’s low energy but they are not the same, what you can do is search for BTE and then open a new tab and search for LE Audio, this should show how different the profile’s are, LE Audio is a big deal right now and you wouldn’t get SIG making a big noise on this if they were the same thing.
Working with the brain → with normal hearing people, they have the ability to take the sounds in “as is” and use their brain hearing to do the processing (filtering, sorting out, focusing). They’re able to do it more easily because their normal hearing has access to the full dynamic range of the sound scene. With hearing impaired folks, their input dynamic range gets compressed a bit more (because lower volumes must get amplified), so instead of a 0 to 120 dB dynamic range for a normal low to high soundscape, the hearing impaired range will become tighter. For example, if the threshold hearing is now at 50 dB, the impaired user now only has a dynamic range of 70 dB to work with (120 minus 50). The subtle nuances in the normally wider range now get “squished” into a tighter range → harder to detect → harder to sort out/filter/focus. So the Oticon aids do many tricks to help balance out the various sound components in the sound scene and also clean up the diffused noise from the speech to help the brain of the impaired hearing do its processing more easily in a narrower dynamic range of hearing.
Provide a more natural and pleasant soundscape → this implies balancing the sound scene to help make the sound components stand out more. Maybe subdue some of the louder sounds by a lower margin so they don’t drown out other sounds. Increase the volume of the softer sounds by some margin to help make them more noticeable. Do noise reduction on droning mechanical sounds to reduce distractions. Use the voice detector to give speech priority.
Give your brain access to the full sound scene → this is the open paradigm. In a noisy environment, they don’t just automatically block out everything from the back and the sides. They still let all the surrounding sounds in but they balance them out first.
Quality sound information → beside balancing the sound scene mentioned above to help with the ability to recognize and hear the individual sounds better, there are other tricks they do, like binaural processing to help with the ability to determine the direction of where the sounds comes from; soft sound enhancement, feedback management. Then there’s also a wide input dynamic range of 113 dB SPL on the mics to accept loud enough (attack) sounds like musically percussive instruments. The Real adds better wind handling and sudden sound handling, etc.
Again, my personal opinion here:
All these words used when they describe their DNN because the DNN allows them the ability to take a sound scene and break down the sound components into more discrete components which gives them the ability to control these sound components of a sound scene more easily.
Before, with the OPN and the OPN S, they basically “balance” the sound scene by applying MDVR beam forming to reduce the mechanical noise sources while letting in all other sounds as a whole. And they have a voice detector to let them know when they can do further diffused noise reduction for the voices in the front. But with all the other sounds that are not droning mechanical noise or voices, there’s really no special way to treat them individually, so they just get “let through”. Those other sounds are not really “treated” in any other way, except that the mechanical noises get “balanced out”.
With the More and the Real, while they still keep MDVR beam forming for noise reduction, but now they can also take it one step further and make use of the DNN to break down the sound components in the sound scene further, and then rebuild them into a new sound scene to present to the user. By taking this approach, they can provide more user input to let the user define their level of experience, which situations are easy and which are difficult for them personally. This varies depending on the user. This is the “more precise and balanced way” (compared to the original OpenSound Navigator technology) because the DNN can give the user more precise control in defining which environment is easy and what is difficult for the user personally.
Also, beside the MDVR beam forming for general noise reduction up front, the DNN can now provide further noise reduction inside the DNN, called the “Neural Noise Suppression”. They can adjust the balancing between the discrete non-voice components with respect to the voice components to give better SNR to the voice components as a way to create noise suppression neurally (simply meaning in its digital neural network, not in the actual human brain).
This is different than how the OPN does it. The OPN subtracts the noise model of the back and sides that it creates from the front speech to clean up the diffused noise the front speech. This is similar to how active noise cancellation is done in headphones. But the DNN neural noise suppression no longer needs a noise model of back and sides to subtract from anything anymore. So the control of the various discrete sound components of the sound scene can be more precise → this is the precise approach they mentioned.
As for the “optimized input” that goes into the brain, it simply means that the brain of the damaged hearing would not have to deal with unprocessed raw sound scenes that makes it more difficult to decipher the information. Instead, the processed sound scene (with balanced discrete components in accordance to the user’s control) is prepared and “optimized at the input” that goes into the damaged hearing so that the brain of this damaged hearing has an easier job processing it.
But there’s no mention of xMems tech used. I think it’s too early. I’m aware of only one company who sent an earbuds prototype (with xmems) to a reviewer.
It’s confusing. BLE has been around quite a few years. Initially it was just for signals. Then Apple patented “Made for iPhone” as they figured out a way to use BLE for audio. Android followed years later with ASHA and now finally a universal standard of LE Audio seems within reach.
I have the Oticon More1 and Real1 aids paired yo my Samsung S23 phone and the connectivity to the phone is great as long as you don’t pay attention to the app and even the app stays connected better than it does to the iPhone. Yes both sets of aids are fully updated to latest firmware.