CEO interview: Sonical launches ‘app store’ for hearables

CEO interview: Sonical launches ‘app store’ for hearables

Interviews |
By Nick Flaherty

A UK engineer is aiming to create an ecosystem of apps for hearables based around a full stack from chip to operating system, coming out of stealth mode at the CES show this week.

Gary Spittle, CEO of Sonical Sound Solutions in California, talks to Nick Flaherty about the creating an operating system and ecosystem not just for earphones and headphones but a new generation of healthcare devices based around chips using the RISC-V and ARM instruction set architectures.

Spittle has a long pedigree in the UK semiconductor business. A DSP engineer at Symbionics in Cambridge and then at Bluetooth pioneer Cambridge Silicon Radio, he moved to the US after CSR’s acquisition by Qualcomm and steering through the acquisition of audio experts apt-x in Belfast.

Working for Knowles, Dolby and Personic in the Valley set him up in the audio business.

“After CSR I moved out here to the Valley with a consultancy on ear-related issues [called Audicus] and have spent the last five years or so looking at the issues,” he told eeNews Europe. “There’s a huge gap in the technology and we estimate the software is five years ahead of any of the hardware so half the team are ex CSR and ex Qualcomm from Cambridge.”

Hearable innovation

It’s a lack of processing and connectivity that stifles innovation, he says.

“The real innovation of apple was the ability to run third party apps and there was a lot of resistance to that at the time. The same concept is happening in other areas,” he said, pointing to Apple’s Carplay software that allows cars to access apps on an iPhone.

“CarPlay is brilliant for me as a user and its brilliant for the car manufacturers as they don’t have to develop all that software. But how do we get an app from the phone to other devices that don’t have a screen so we spoke to hundreds of product companies. We found that there are dozens of features that companies want to put into their products and none of them can make them so we need apps so that third party developers can unlock these features, not just spacial audio but so many other things.”

Ear computer

“So they need an ear computer, there’s no way they can port their software to all the different chips out there. That’s what we create, an ear computer that makes products app-able, and takes things to the next level with an operating system.”

“We will show the full app store at CES and the first product not long after CES and then during 2023 there will be a handful of new products,” he said. “We’ve been filing patents like crazy.”

This is CosmOS, an operating system with a hardware abstraction layer (HAL) that allows code to run on ARM or RISC-V chips.

“CosmOS allows us to take apps from the cloud to the phone and push them to devices. We are starting with ear products and moving to other things, smart microphones, speakers, biometric monitoring and so on.”

“To do that for the ear we focussed on three main areas. We typically buy headphones to listen to music. We support that but its not our primary goal. The second area is listening to the world, and studies have shown everyone can benefit from enhanced sound and many companies are working on that.

“But what is really interesting is listening to your body. You can pick up vital signs, motion, all kinds of data and use that information to inform other things, for example combined with streaming music from Spotify along with biometric data that says this is more relaxing.

“We took a look at all the things that people want to do and found that the compute is nowhere close to good enough. Hearing aids cracked the battery life years ago, so we know we can get the performance


Sonical is working with French design house Dolphin Design, part of Soitec, on a RISC-V chip. “We are targeting 22nm and we have a huge amount of compute at our fingertips, we can stream audio and download apps simultaneously,” he said.

“We evaluated everything on the market and where we landed was the ease of use for the developers,” he said. “When you start designing your own processors and instruction sets it’s really hard, so we started with RISC-V with DSP extensions and it’s a really good fit for this kind of application.”

“There’s two designs in the mix. One is on a test chip but we are not fixed yet and evaluating it. It comes down to support. We haven’t made our final choice yet,” he said.

But this is not jut about a custom chip with custom software. “We have taken our stack and ported that to other chips with limited capabilities.”

This requires the hardware abstraction layer. “The HAL is customised for RISC-V and ARM so I can take an app with identical code on each of those. That’s why the developers love this. That’s how it has to be. But that’s really hard to do, it’s a lot harder than we thought.”

AI accelerators

Using RISC-V allows customisation but runs the risk of having a more fragmented market and more chips to support.

“We don’t want RISC-V turning into 100 variants and making our lives a misery so we are being quite selective in the chips that we support,” he said. He is also looking at the algorithms and AI accelerators that will boost the performance of the apps.  

“We are working with some of those chips but it is a headache making the apps work seamlessly,” said Spittle. “There are some algorithms that need the accelerators, low latency audio streaming with AI is hard to so and it needs a stable framework. So we have built a plugin that can talk to accelerators through the OS and we can run that on the accelerator or on the CPU.”

Bluetooth restrictions

Connectivity is another missing piece and he is dismissive of Bluetooth.

“Bluetooth is going nowhere and it takes so long,” he said “BLE (Bluetooth Low Energy) has taken seven years, and it has made no progress and even though its out here its nowhere where it needs to be,” he said. “If you start building a different chip you end up in a different place to Bluetooth.”

He does use Bluetooth for the initial handshaking, but then the multiradio chip takes over, interfacing via the phone or the even the charging case which also has its own battery.  

“We use BT for the basic interoperability and we piggyback on that,” he said, “It could be WiFi or UWB or something custom. The charging case will be very smart so I’m less dependent on the phone and that could be UWB streaming from the cloud for example.”

Going forward

The chip is not the way to make money, but it does demonstrate how much more performance is available as the apps become more sophisticated in  exactly the same way as the App Store for Apple or the Google Play Android app store.

“We use the apps to make our platform attractive in order to sell the chip and license the OS onto the chip and other chips and then we charge a license fee to the developers for the ease of use,” he said. “Right now the first customers will have a walled garden for their products but smaller developers can use App store or Google Play.”


If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News


Linked Articles