Conversation Enhancement AI for True Wireless Stereo (TWS) Earbuds

In today's episode of the HearingTracker Podcast, host Steve Taddei interviews Giles Tongue, CEO & Co-Founder at Chatable. Giles is passionate about conversation enhancement technology, and his company is a frontrunner in the race to revolutionize the True Wireless Stereo (TWS) earbuds for the 1.5 billion people who suffer from hearing loss globally. Tune in for an update on the state of the art, and to find out what to expect from this rapidly advancing field of technology.

Podcast Transcript

Steve Taddei: Welcome back. I'm Dr. Steve Taddei and this is the Hearing Tracker Podcast.

Nowadays, the world of hearing technology feels rather chaotic. For example, you have electronics companies like Bose offering direct to consumer hearing aids. At the same time, hearing aid manufacturers are designing multifunction consumer earbuds. You don't even need to leave your house to get your hearing tested. But what about hearing care providers? The contradictions seem plentiful and there are so many changes happening all so fast. Well, in today's episode, I was joined by Giles Tongue, the CEO of an artificial intelligence startup called Chatable, and he feels that amidst this chaos we're on the right path and at the threshold of hearing technology that's personalized like never before.

So let's get into it.

Giles Tongue: So one of the big technologies that really started this big change is Bluetooth. Bluetooth has enabled the creation of the true wireless stereo earbud.

Steve Taddei: That is Giles Tongue. And in case you didn't already know true wireless stereo earbuds, or TWS, are pretty popular right now.

Giles Tongue: So there's 2 billion pairs of earbuds expected to be sold over the next four years.

I don't think we've ever seen a product category grows so fast in consumer electronics before,

Steve Taddei: But as we talked about last month, headphones have been around since the 1800s. So what has changed that's made them become so popular?

Giles Tongue: So TWS earbuds were first popularized because they enabled the convenient stereo streaming of music, media, phone calls with no cables.

So very convenient form factor. And over the last few years, we've seen active noise cancellation becoming ubiquitous to keep noise out and enhance the experience because, let's face it, nobody likes noise. With that comes transparency mode, also known as hear through or pass through, which was added so people could hear the world around them without having to remove their earbuds.

And of course, when we remove our earbuds, there's a risk of being disconnected or unpairing, which is a real faff.

Steve Taddei: Transparency mode is a serious innovation. By adding a few microphones to the earbud, you have the power to control the world around you. Well, at least the volume of it. You may remember this commercial Apple released in 2020 featuring transparency mode and a woman getting lost in her music while dancing around a city.

This technology is inspiring and we're already building off it in newer innovations.

Giles Tongue: What we're seeing more recently is hearing personalization, which is being added so that we can account for everyone's different hearing levels, hearing needs. And pioneers like Apple are tracking and showing noise exposure and safe listening levels through their AirPods, the watch, and the phone.

So a whole bunch of different hearing related features are coming at the same time.

Steve Taddei: So really we're just talking about TWS earbuds, becoming hearing aids right? Not at all Giles went on describing the true implications of this technology and even the possible generational impacts.

Giles Tongue: So your kiddo today is using ANC, personalized audio, transparency for live conversation, mostly because they don't want to take their earbuds out.

So they want to keep using their earbuds all the time and don't wanna disconnect. So they want to keep transparency mode on, and they're aware of safer listening from day one. So let's say that's a ten-year-old and as their hearing is changing over the years, they can monitor it. They can keep tweaking their personalization and changing accordingly to the usage.

So this means without any particular deliberate effort, the Airpod user is using and benefiting from a range of hearing health features, which are enhancing, protecting, monitoring their hearing. And this is now the norm for a whole generation. And some earbuds, you know forgetting about airports for a second, some earbuds don't even want you to use them without doing a hearing test first for personalized audio. It's the whole point of them.

Steve Taddei: This reminds me of a scene from one of my favorite movies.

That's right Back to the Future. In the second movie, Marty McFly finds himself in the future where his clothes automatically adjust to fit him.

While we regrettably didn't see hoverboards and flying cars in 2015, at least our hearing tech seemed to take a hint. And whether we realize it or not, our audio is becoming increasingly personalized. When we come back, we're talking artificial intelligence and the Chatable app.

So stay tuned.

Thank you for listening to the Hearing Tracker Podcast. If you know anyone interested in hearing tech and artificial intelligence, please consider sharing this episode with them. Giles also had excellent insights into chip development and the growing speed of innovation. And though we didn't have time to include that in this episode, it can still be heard in the unedited recording. That full conversation and more at Patreon.com/hearingtrackerpodcast.

So far, we've been speaking with Giles about some of the amazing technology at our fingertips. But Giles is more than just a tech guru. He's actually the CEO of an AI startup company responsible for an app called Chatable.

Giles Tongue: Yeah well Chatable is conversation enhancement AI for the TWS earbuds and hearing aids. We call it a new experience because you have to hear it to experience it.

And essentially what we're doing is upgrading transparency mode with conversation enhancement. So this means you can leave your earbuds in to have a conversation and the benefit of not having to take your earbuds out. And what Chatable is doing is removing background noise, reverb, music, other talkers. So then you can hear conversation easily.

Steve Taddei: So what we're talking about here is a phone app that processes audio to make speech clear, and it can be used with any ear worn device, like headphones and hearing aids. But wait, don't these technologies already offer noise reduction?

Giles Tongue: From a technological point of view, if we're comparing let's say traditional hearing technology, or DSP, it's a relatively light computational process that's comprising of multi-band graphic equalizers, some compression, sometimes beam-forming for directionality. What we're doing is using deep neural networks, performing a hundred million calculations per second in your ear. That's basically a complex black box, which takes in audio from a single mic, processes it really really quickly and outputs the new samples to your ear. So quickly that you'll never really know that any processing's been happening.

Steve Taddei: Let's take a moment and break that down. Traditional hearing aids have many processes. As Giles mentioned, there's equalization, which allows you to adjust the volume of different frequencies.

Think of your car and the base, mid, and treble knobs. Well, hearing tech offers a lot more control and you can make almost any adjustment you want. Like I'm doing right now with my voice. There's also compression, which can tame the loud peaks while accentuating quieter segments. Here's an example of hyper-compression where the peaks in my voice are being reduced by well over 10 decibels.

Giles also mentioned beam-forming microphones. Well, right now I'm using a highly directional type of microphone known as a shotgun mic. This loosely emulates the beam forming found on hearing aids. As I rotate the microphone away from me, notice how my voice gets quieter and less natural sounding. Together, all of these processes can help make speech clear. But Giles mentioned that Chatable uses something else.

Artificial intelligence. Now, if you're anything like me, these words flood your mind with images of dystopic futures, robots, fear, and humanity in peril.

And we have Hollywood to thank for that. But in reality, artificial intelligence is a little different.

Giles Tongue: So generally anything an animal or a human does that you describe as intelligent, is artificial intelligence if it's done by a machine. Another possible definition of intelligence is any process that the brain conducts.

So anything that replicates a brain process can be said to be artificial intelligence or artificially intelligent. Since we're replicating the intelligent brain process that were involved in how you hear speech and speech and noise, using artificial neurons in artificial neural networks we tick that box.

Steve Taddei: Giles had also mentioned deep neural networks, which as it turns out are a subset of sophisticated learning algorithms within artificial intelligence.

Giles Tongue: When you want to use a neural network based AI to solve a problem, you can sometimes find a brain that already solves the problem using neurons. If you then study the neural processing of that brain, and apply that to the process of developing neural network based AI, that's neuroscience led AI.

So the problem of speech and noise is already solved by the healthy human brain. So reverse engineering this solution is a good way to get an AI that does the same job. Reverse engineering the brain is called neuroscience. So we inform our AI work by looking at what the brain does when actively listening to speech.

So our AI is basically giving to you the output that you would achieve with a healthy functioning human brain artificial.

Steve Taddei: What this means for the Chatable app is that their algorithm architecture is based on the human brain. It analyzes the input, filters it, and resynthesizes speech. The result is individual volume control over both background noise and who you're talking to.

For example, this is what my voice might sound like if you and I were sitting across from each other at a noisy restaurant. I'm sure you can still hear me, but there's obviously a little distraction from all the noise. So let's try out the Chatable app.

Now that Chatable app is on and you're wearing a standard set of Apple earbuds. Now I'm increasing the volume knob for my voice. So you should hear me getting louder and clearer as I do so. And this is what it sounds like as I started reducing the background noise, you should hear it's becoming quieter while the sound of my voice stays relatively the same.

The Chatable app is available for download now, and that is their first-generation technology. But their team is working to bring this process on-ear with their third generation Edge AI

Giles Tongue: For the new Edge AI, what we've done is use the Knowles DSP, which is a tiny little chip which can work alongside... typically in an earbud you'll have an SOC, a system on a chip, which is basically running everything you need to have a functioning earbud. And then the Knowles chip runs alongside that in tandem with it. So we're entirely on that Knowles chip. So sound comes into the microphone, we process it on the Knowles chip, and then send it out to the speaker. And this is the first conversation enhancement AI without audible latency and the first to go on a chip in your ear.

And we've just finished adding personalization. So you can import an audiogram. If you're an iPhone user, your audiogram goes into Apple Health, and then we can pull that into our companion app, which you would use with your earbuds. And then we'd send up instructions to our AI, which is then personalizing your conversation enhancement specifically to what we've seen in your audiogram. Or you can use manual controls to personally adjust, you know, bass treble and that kind of thing.

Then the other thing we're adding now is hearing health features. So now you can see how much background noise we've actually removed and quantify the amount of noise exposure that you were exposed to and how much we've removed.

Steve Taddei: Artificial intelligence and deep neural networks have been around for decades, but advancements in technology and processing power have opened the gates and we'll see these algorithms improving features and speech clarity. And we're living this right now. Hearing aids are already using deep neural networks in lieu of the traditional rule-based processing.

Other companies like Chatable are pushing the boundaries with TWS earbuds.

Much thanks to our guests, Giles tongue of Chatable. To learn more, check out Chatable.ai. This episode was written, produced, and sound designed by me with help from Dr. Abram Bailey. Don't forget to subscribe and thank you for listening.