Imagine this.
You’re sitting in a very noisy restaurant.
Your friend is speaking right across from you.
But at the same time,
the two tables next to you are chatting.
A waiter is collecting plates.
The coffee machine is running.
Someone is moving a chair.
There’s music playing in the background.
If someone then says:
“Can’t hear clearly? Just turn up all the sounds.”
You might find
that it doesn’t actually make things easier.
Because your friend’s voice gets louder,
but so do the clinking dishes,
the chatter from nearby tables,
and even the coffee machine sounds get amplified.
The real problem is never just about:
whether the sound is loud enough.
Instead, it’s about:
when so many sounds come in at once, which part do I really need to hear?
This is exactly the problem that the latest generation of AI hearing aids is tackling.
Phonak’s New EON Does More Than Just Amplify Sound
Swiss hearing technology company Sonova launched the new Phonak EON hearing aid platform on August 24, 2026.
Among them, the EON Sphere features Sonova’s third-generation real-time AI hearing technology and the new HYPERSONIC AI chip.
What’s truly remarkable
isn’t just that it has the “AI” label,
but what the AI actually does in your ear:
it separates speech from background noise in real-time.
Sonova explains that the Spheric Speech Clarity 3.0 uses retrained deep neural networks to continuously analyze the surrounding sounds as they enter the hearing aid, extracting speech from noises coming from different directions.
In other words,
the AI doesn’t wait until you’ve recorded a segment of sound to analyze it slowly;
it works constantly while you’re talking to someone.
But AI Doesn’t “Know” Who You Want to Hear
This is a very important distinction.
A hearing aid with AI is not reading your mind.
It doesn’t know:
“This person is your spouse, so they’re more important.”
Nor does it know:
“Those people at the next table are irrelevant to you.”
What it actually does is more like:
recognizing the sound environment,
distinguishing speech from other noise,
and enhancing speech clarity relative to background sounds.
Phonak’s other technology, AutoSense OS AI 8.0, continuously analyzes the current listening environment
and automatically selects the most suitable processing mode for the situation.
So the real progress isn’t:
that AI knows what you’re thinking.
But rather:
that AI can identify what’s happening right now much faster.
The Old Problem Was “Not Hearing,” Now It’s “Not Understanding Clearly”
These two challenges aren’t exactly the same.
If only one person is talking in the entire room,
amplifying sound might genuinely help.
But real life isn’t a recording studio.
The toughest scenarios are often:
family dinners,
restaurants,
weddings,
meetings,
train stations,
markets,
or multiple people chatting simultaneously.
In these environments,
the problem isn’t:
“there’s no sound.”
It’s:
too many sounds.
Therefore, the real challenge for hearing technology
is shifting from:
“amplifying sound”
to:
“pulling out important sounds from the rest.”
It’s Like Having an Audio Engineer Always Sitting Next to You
To put it simply,
imagine an invisible sound engineer sitting in the restaurant with you.
Your friend starts speaking.
The engineer immediately recognizes:
here is speech,
while the clinking dishes nearby
are background noise.
Someone else speaks behind you.
The system analyzes again.
The environment shifts,
a waiter walks by,
someone at the next table suddenly laughs loudly,
and it quickly adjusts once more.
This decision-making isn’t done every five minutes.
It has to happen continuously and rapidly during your everyday life.
This is why AI is perfectly suited for hearing aids.
Because this kind of task basically involves:
constantly analyzing large volumes of sound,
recognizing different patterns,
and quickly adapting.
Sonova Says Users’ Chances of Understanding Full Sentences Increase
According to Sonova’s own research,
wearing hearing aids with Spheric Speech Clarity 3.0 enabled doubles users’ chances of correctly understanding every word in complex noisy environments compared to when the feature isn’t active.
The company also states that listening effort can be reduced by as much as 37%.
These numbers are compelling,
but it’s important to remember:
this is research published by Sonova/Phonak.
It shows the technology’s effectiveness in their tests,
but doesn’t guarantee
every individual with hearing loss,
every degree of hearing impairment,
or every noisy environment will experience the same improvement.
Hearing aids still require personal adjustment for individual hearing conditions and real-world settings.
The more accurate way to put it is not:
“AI has solved hearing loss.”
But rather:
“AI is improving one of the toughest hearing aid challenges: speech in noise.”
Why Was It So Hard to Put This Kind of AI Inside Hearing Aids Before?
Because AI processing isn’t free.
Deep neural networks require continuous sound analysis,
which demands:
computing power,
and energy.
If you put a power-hungry AI into a hearing aid,
the result might be:
great sound processing,
but batteries drain quickly,
or the device becomes too large and uncomfortable.
That’s another key point about the new EON generation.
Sonova says the new HYPERSONIC chip improves EON Sphere’s AI processing efficiency by 37% over the previous generation, allowing the device to be 25% smaller and 20% lighter, while still offering battery life of up to about 38 hours.
This reflects a common trend in AI development.
What enables AI to enter everyday life
isn’t necessarily a sudden tenfold increase in intelligence.
Often, it’s more about:
making the same AI smaller, more energy efficient, and comfortable enough to wear all day.
This Is the Opposite of What We See with Generative AI
Every day, we talk about ChatGPT, Gemini, Claude,
and it’s easy to get the impression that AI’s value is to:
generate more text,
create more images,
produce more videos,
answer more questions.
But the hearing aid case takes a very different approach.
Its most important job isn’t:
to add more.
It’s:
to remove.
Or more precisely,
to reduce interference from unimportant signals,
making the important signals easier to hear.
When ten different sounds happen simultaneously in a restaurant,
AI’s value isn’t creating an eleventh sound,
but helping you focus on the one that truly matters among those ten.
What People Often Really Need Is Not More Information
This applies beyond hearing aids.
Today, when we unlock our phones,
we see:
emails,
LINE messages,
news,
short videos,
notifications,
documents,
meetings,
and search results.
AI can quickly generate even more content in seconds.
So the AI that will truly bring value in the future
may not just be:
“What else can I give you?”
but rather:
“Among all these things, which ones are most worth keeping?”
Hearing aids make this question very concrete,
because for someone sitting in a noisy restaurant,
trying hard to understand what family members are saying,
what they really need isn’t:
more sounds.
It is:
less interference.
And This Kind of AI Needs to Work in Real Time
When ChatGPT takes three extra seconds to reply,
usually it just means waiting a moment longer.
But real conversations don’t work like that.
If your friend says:
“See you tomorrow at 9,”
and the hearing aid takes three seconds to finish processing the background noise,
that sentence has already passed.
So this type of AI must meet a strict requirement:
it has to judge fast enough that you don’t even notice it’s doing so.
This is the importance of Edge AI—placing AI computation directly near the device.
Not all sounds first go to a distant cloud,
get processed there, and then sent back to your ear.
Massive sound processing must be done instantly on a device very close to the user.
This is a completely different path from the large data center-based AI we discussed yesterday.
While AI is becoming dependent on huge data centers,
it’s also shrinking to fit into:
phones,
earbuds,
hearing aids,
and other small devices we wear every day.
The Real Question Isn’t “Does It Have AI?”
In the future, almost every product will claim to have AI.
The real question should never be:
“Does it have AI?”
But rather:
“What original, difficult problem does the AI solve for the user?”
For hearing aids,
the answer is very specific.
Not writing articles for you.
Not generating photos.
But in a noisy world,
helping important words
stand out without being drowned by other sounds.
This may be one of the most human uses of AI.
Technology doesn’t always have to add more to the world.
Sometimes, the best technology
simply helps a person hear clearly again:
what the person sitting right in front of them is actually saying.
Today, let’s make progress with AI.
Learn one AI skill a day.
Save a little time each day.
Increase your capabilities a bit every day.
SasaDaily, growing with you.
Recommended Reading
AI Quick Q&A｜2026/08/02: Does AI's Complete Transcription of Speech Guarantee Perfect Accuracy?
