Future of Hearing Aids in Background Noise with Google Ai (Artificial Intelligence)

I’m going to tell you about recent research that may dramatically improve the way that hearing aids perform in background noise. The Holy Grail for individuals with a hearing loss is the ability to separate speech from background noise, whether you’re talking to one person or whether you’re talking to a group of people.

I recently saw a video on the Google AI blog that illustrates the ability to be able to separate out who you want to hear, just by looking at that individual. Inside of this software that they’ve created, it actually is pretty impressive. Instead of me trying to describe it to you, watch the video below (from 0.58 – 1.39) to see it in action.

What did you think? Now if you ask me that’s pretty impressive. But the number one most difficult listening situation that most people encounter is a noisy restaurant.

The next clip I’ll show you demonstrates their ability to separate out speech from background noise in a cafeteria setting (watch from 1.53 – 2.20).

Now I don’t know about you, but I feel like this technology inside of hearing aids would be a complete game-changer. Excited, I read the article connected to the research to see how much viability this would have in future generations of hearing aids.

It’s titled “Looking to Listen at the Cocktail Party: A Speaker-Independent Audio-Visual Model for Speech Separation” and was written by a number of Google researchers. Here are a few key points that I pulled out of this article.

  1. The researchers performed this on existing videos, meaning they didn’t do it in real time. So, before you get really excited about just taking this software, and putting it into a hearing aid, listen to what the authors had to say:

“It should be noted that our method does not work in real-time, and, in its current form, our speech enhancement is better suited for the post-processing stage of video editing.” – Google Researchers

  1. This technology is a huge leap forward for audio-visual processing. Previous generations of this type of software required the software to be trained by a certain voice. Meaning, if you were going to go to a restaurant with a certain friend, that certain friend’s voice would have to be programmed into the software so it knows what to actually look for.

In this particular software, they could actually take a generalized speech format so it would work with any individual in any situation which makes it more viable in a generalized setting, meaning you could go anywhere and communicate with anyone and use this particular technology.

  1. When they did this in a noisy bar situation, they were able to get rid of the background noise, but it significantly degraded the speech of the individual that they were trying to hear. This is very similar to the limitations we have in hearing aid technology today. Being able to remove a speech source from a variety of other speech sources can be very difficult.

So, if they’re going to make this a viable product inside of a hearing aid, they’re going to have to fix that as well.

  1. It’s important to remember that the researchers need visual information to be able to do this audio-visual separation, and hearing aids of course do not have visual information – at least not yet. You see, they are actually working on hearing aid technology that can sense brain activity for eye movement and if they can do that and combine it with this audio information, they may actually be able to determine who you want to hear based on where you’re looking.

While this type of technology is not yet available inside of hearing aids, it is research like this that gives hope to individuals with a hearing loss that struggle in a background noise situation and absolutely paves the way for future research for technology that will make its way into hearing aids.

In the meantime, one of the best ways to improve your ability to understand speech in a background noise situation is to have real ear measurements performed on your hearing aids. Now if you don’t know what real ear measurement is, I highly recommend you check out this video (https://appliedhearingaz.com/real-ear-measurement/).

One other way to dramatically improve your ability to understand speech in a background noise situation at the current moment is to use an assisted listening device, like a Roger Pen, a Roger Select, or a remote microphone. These types of devices will help cut through the background noise, because it takes the speech of the person you want to hear and sends it right into your hearing aids.

Video transcript

Video transcript

Everyone with hearing loss wants to hear better in background noise.  Well, in a recent blog post and video on the Google Ai Blog, researchers are working on software to do exactly that: https://ai.googleblog.com/2018/04/loo...  

In their video they show how they are able take audio and visual information to separate speech that you want to hear from noise that you don't. I personally find this software to be impressive.  Their video speaks for itself.  So I decided to read the entire research article to pull out key points about how it works and what the future implications might be for hearing aid technology.

1. It only works on existing videos right now - The researchers were very specific that their technology does not work in Real-Time.  In fact they stated, "It should be noted that our method does not work in real-time, and, in its current form, our speech enhancement is better suited for the post-processing stage of video editing."

2. This is a HUGE leap forward for Audio Visual processing - Previous technology required the training of software to learn each talkers voice so it knew what voice to focus on.  In this new software, it was trained with general speech information so it will work on anyone.  

3. Really noisy situations degraded the speech - When using this software for a Noisy Bar (not shown in the video) the researchers indicated that the speech was significantly degraded when separated from the noise.  This is a similar issue with hearing aid technology today.

4. There is no way for hearing aids to obtain visual information - The reason this software works is because they use Audio AND Visual information.  Hearing aids do not have a way to capture visual info at this time.

However, researchers are working on a way for hearing aids to capture electrical brain impulses that correspond with visual information so someday this technology might work in a hearing aid.While this Audio-Visual technology doesn't yet work inside of hearing aids, it does show promise for the future.  It is research like this that paves the way for new hearing aid technology so it is just a matter of time until they figure out a way to make it work.In the meantime, the best way to hear better in background noise is to have your hearing aids programmed using Real Ear Measurement: https://youtu.be/cHR0Oa6I-wYThe other way is to use a remote listening device like a Roger Pen, Roger Select, or a remote microphone.

Related videos

No items found.

Related articles

No items found.