This statement is very vague and doesn’t really explicitly implies self learning.
"Allowing Edge to learn" can simply means to activate it so that it can start analyzing (or learn) the current environment the user is in to match it up with what’s in a pre-built database.
"Apply the best fit for that situation" is what it is, once the situation of the current environment has been analyzed and classified (or “learned”), it finds the matching situation in the pre-built database, then extract the parameters pre-defined to be best suited for that situation and apply those parameters’ settings to your hearing aids.
So, just a simple statement like that from Starkey doesn’t mean that the hearing aid can really self learn.
Training a neural network (like what Oticon did with the More), or using AI to build a smart database (like what I think Starkey did with the Livio Edge) is not the same as self learning. If you wear the hearing aid for 3 years, then it crapped out and you have to get new replacements, the replacement hearing aid will perform exactly the same as the original hearing aid. The original hearing aid does not get any smarter than the replacement hearing aid just because it’s had 3 years of self-learning experience from being worn by yourself, while the replacement hearing aid has 0 experience because it’s never been worn. The trained/accumulated “smart” is already built into all hearing aids just the same. They don’t evolve and self-learn individually to become different from each other at their core. Sure they’ll have different personalized settings stored, but their core smart is all the same and one HA does not take on a different experience from another as they progress over the years.
I think real self learning requires a very complex and self-sustaining digital eco-system that’s not easily fit inside a tiny chip of a tiny hearing aid.
I’m not arguing that the Livio Edge doesn’t do self-learning. I’m just saying that I don’t see any evidence of it described in its whitepaper, and the statement from Starkey about “allowing Edge to learn” does not imply self-learning for me.
Now the Widex Evoke claims self-learning, and I would buy that claim. That’s because they use their smartphone app to establish connectivity between the hearing aid back to their cloud based system. The user, by being presented with an A/B choice of settings, then to tell the system which is preferred for his/her current environment, would have their A/B choice along with the environmental data fed back through the phone app to the Widex cloud, into their self-learning eco-system, to train it better and better using connectivity with their hundreds of thousands of users (or millions? I don’t know how many they sold so far) over more and more exposure to more and more environmental data and more and more users’ preferences learned, and over more and more time, now that’s self-learning.
Maybe the Starkey Livio Edge AI does the “pushing” of data back into the Starkey cloud, but it’s not obvious to me through the whitepaper. Widex, on the other hand, is very explicit about this in their marketing claims. It seems like the Livio Edge is “pulling” from the database, and this database is either stored in the hearing aid or in the Starkey cloud and gets pulled through the Thrive control app. That part (where the database resides) is not clear to me.