Are you talking to me? Voice technology and AI at CES 2018
Image credit: Getty Images
The CES 2018 tech show was all about voice: yours.
There’s no question as to who the real technology star is now: it’s you. Your voice is what hundreds of companies are vying to attract, with thousands of new products calling out for you to talk to them.
Voice-activated technology has erupted over the last 12 months since Amazon’s Alexa was informally crowned breakout technology champion of the CES 2017 consumer tech show. Seemingly by stealth, Amazon had snuck Alexa into a dizzying array of products and everywhere you turned, there she was.
Alexa was the name on everyone’s lips – literally – and Amazon had achieved this near-ubiquitous name-recognition without even having a stand at the gargantuan annual gadget-fest in Las Vegas.
Small wonder, then, that at CES 2018 Google went large – American large – to promote its Assistant and its clarion call, ‘Hey Google’. That legend and Google branding dominated both the show and the city of Las Vegas itself. ‘Hey Google’ was everywhere, including a complete graphical takeover of the Las Vegas monorail, which conveniently runs direct to the Las Vegas Convention Center.
It’s rare for one company to dominate CES in such a way, but voice is clearly such a key battleground for device dominance that Google felt compelled to create a dramatic splash. Google was unmissable at CES 2018, erecting its own giant outdoor space showcasing over 350 products that already respond to ‘Hey Google’ and giving away free Assistant-powered products from a giant gumball machine.
Unfortunately for Google, apparently no one had factored in the possibility of the torrential all-day desert rains of Tuesday – rains which broke a 120-day drought – that soaked the city and forced the company to close its stand for the day.
“Alexa, what’s the weather forecast for Las Vegas?”
All of this serves to show that voice is the new interface frontier, and right now everyone else is playing catch-up to Alexa. Alexa is already integrated into over 4,000 smart-home devices from 1,200 brands, with more than 30,000 Alexa ‘skills’ (Amazon’s term for voice apps) available. The first product a company can successfully place in the home is likely to become the gateway gadget to more of the same.
According to Strategy Analytics, Amazon’s Echo devices currently have around 67 per cent of the smart speaker market, with Google on 25 per cent. Google spent 2017 doggedly chasing Amazon’s heels, as the Seattle retail giant sought to consolidate its first-mover advantage in putting Alexa everywhere (into “tens of millions” of products, according to Amazon CEO Jeff Bezos). Both companies have been selling their voice products as loss-leaders, primarily to claw back market share, but also to ward off aspirants to the voice sector, such as Apple, which will shortly release its own much-delayed HomePod speaker powered by the familiar voice of iOS’s Siri.
The idea of Apple being viewed as a lightweight combatant in a leading-edge consumer technology fight seems incredible. That said, Apple and Siri’s position is arguably still stronger than Microsoft’s Cortana assistant – which has seen Alexa oust it as the voice assistant of choice on new desktop and laptop PCs coming in 2018 from from HP, Lenovo, Asus, and Acer. As for Samsung’s Bixby, that is now largely restricted to appearances on the South Korean company’s own devices and appliances.
According to global technology market analyst firm Canalys, in the first nine months of 2017, around 17.1 million smart speakers were shipped worldwide, but a whopping 16.1 million were shipped in the final quarter of the year, driven by Christmas purchases. Canalys predicts 70 per cent year-on-year growth in 2018, expecting to exceed 56 million units. There’s a lot of money to be made from voice: hence the drive to embed the functionality into everything from your toothbrush to your next car.
At CES 2018, both Google’s Assistant and Amazon’s Alexa were well-represented by new product offerings. Google unveiled forthcoming Assistant-based smart displays from JBL, Lenovo, LG and Sony, which capitalise on Google’s existing expertise in visual search, such as Maps and Google Photos, as well as all-important YouTube support – something Google has withheld from Amazon’s competing Echo Show video device. Third-party audio companies such as Bang & Olufsen and Anker also announced new Assistant-driven smart speakers.
Meanwhile, Assistant’s rival Alexa debuted on Acer laptops, as well as appearing in (or being “baked in to”, as the new voice vernacular has it) a diverse range of products – showers, mirrors, light switches, microwave ovens and cars. Toyota cars announced Alexa support, following Ford and Hyundai – automotive being an area Google is also pursuing with Android Auto – as did Vuzix with its Blade smart glasses. Whirlpool confirmed Alexa integration for its home appliances, although the company has wisely hedged its smart bets by also supporting Google Assistant.
The idea of the smart home is driving much of the uptake of voice-controlled devices, with myriad overlapping products targeting areas such as home security, lighting and the bathroom routine, be it the ADT Video Doorbell, Ring, Beam, Caséta, Streety, HiMirror Mini or CareOS. There was even a bath tub, from luxury plumbing company Toto, which simulates zero gravity so you can bathe like an astronaut (it uses water jets to replicate the feeling of weightlessness). This, though, was as nothing compared to Kohler’s $6,000 Numi smart toilet. Alexa-enabled, this luxury lav is voice-activated and has a heated seat, foot warmer, ambient mood lighting and a selection of cleansing and drying bidet options.
Smart toilets aside, the drive towards smart homes controlled by our voices may appear to be built on a simple premise – say something out loud; your gadget reacts. This will inevitably result in some peculiar products (laundry-folding robot wardrobes, anyone?), but the smart home is really an exemplar for several key converging technologies at the heart of the voice revolution: voice-recognition technology, internet and IoT connectivity, AI and machine learning.
The technology industry has already envisioned the near future in which your voice is the core controller for almost everything you do in life, anywhere you go.
“If voice is going to be everywhere, you want to see voice transition from the home, when you get into your car. You’d also want it at work,” says Steve Rabuchin, Amazon’s vice president of Alexa Voice Service and Alexa Skills. “We’re going off into auto and these other places because we believe voice is a natural interface and it should be ambient.”
Speaking at CES 2018, Baidu vice chairman Qi Lu said, “There will be no border between a home and a vehicle. Whatever you can do at home, you should be able to do it in cars.” The Chinese company is making inroads with voice with its own Duer-based devices.
Baidu’s – and China’s – intentions in the automotive sector were reaffirmed at the Nvidia press conference, where Nvidia CEO Jensen Huang announced his company’s partnership with Baidu, along with Volkswagen and Uber, adding these three companies to its long list of self-driving car technology partners. Nvidia has developed Xavier, an AI chip specifically made for self-driving vehicles, with neural network and deep-learning capabilities, “built to bring AI to every aspect of the driving experience”.
“There will be no border between a home and a vehicle. Whatever you can do at home, you should be able to do it in cars.”
Mercedes-Benz unveiled its own in-car voice-control system, eschewing Alexa or Google Assistant for its in-house Linguatronic AI system, officially branded as the Mercedes-Benz User Experience – MBUX. “Hey Mercedes,” as the salutation begins, will launch in the coming months, starting with the A-Class and eventually propagating to all Mercedes cars.
In describing the company’s vision and the decision to go it alone where voice is concerned, Ola Källenius, board member for group research and Mercedes-Benz cars development, said, “It needs to be a holistic user experience.” Much as Apple controls the whole ecosystem around its hardware and software, allowing for tighter control and deeper implementation, so Mercedes wants to avoid the problems inherent in relying on bolt-on third-party solutions. However, Källenius explained that, while its system is proprietary: “That doesn’t mean it’s an island. We’re not in competition with [other] systems. I see [Google and Amazon] as partners.” Indeed, Mercedes already supports both Apple CarPlay and Android Auto.
The MBUX experience, which uses two Nvidia chips, takes a natural-language approach, where the driver addresses the car directly, just as we do with Alexa, Siri, Cortana etc. A reflective instruction such as “Hey Mercedes, I’m too cold” will see the lights around the air vents glow red as your car raises the temperature by a few degrees. The infotainment system in MBUX-equipped cars has also been updated to respond to touch, as well as the swipe and pinch-zoom gestures familiar to anyone who has used a smartphone or tablet.
The same inspiration and transfer of technology ideas has been used by Byton for its upcoming Concept car – the first production model from the brand-new Nanjing-based Chinese car company, unveiled to the world at CES 2018. At the heart of its all-electric car design, inside the cabin, is what Byton describes as the “living‑room experience”.
Going some way beyond the traditional dashboard, the interior has a full‑width, edge-to-edge digital screen – what the demonstration driver referred to as a “coast-to-coast display” during E&T’s test ride at the show – that is 1.25m wide and 0.25m tall. Byton has designed its system to be operated and controlled with just five hand gestures. Amazon’s Alexa will also be built in for voice control, Byton referring to its car as the first vehicle that “also functions like a wearable”.
Byton’s aim is on connection, “building the next-generation smart device” and it’s 5G-ready.
It is the impending arrival of 5G, along with AI and robotics, that will enable the smart society and smart cities. When bandwidth is unlimited and data caps are no longer a concern, the Internet of Things will really take off – and voice will be central to it all.
The shift from desktop to laptop, then smartphone to smart device has been described by Dave Limp, head of Amazon’s devices, as “ambient computing” – something “less dedicated personally to you but more ubiquitous.” What will make voice ubiquitous are advances in the underpinnings of how these devices work.
Voice algorithms will become more refined, with greater standardisation across the industry. The vast amounts of voice data building up, as more and more people use the devices, will feed back in to the process, enabling machine-learning systems to evolve and become more intelligent. AI is crucial in local separation and local intelligence, helping voice systems to differentiate between human voices, identify who is speaking and cancel out the surrounding ambient noise.
At CES 2018, Ceva launched its NuePro chip, “a dedicated low-power AI processor family for deep learning at the edge” – local learning, rather than everything being sent to the cloud.
Taiwanese company MediaTek is also working on AI solutions on the edge with its new AI processor developed by MediaTek spin-off Intelligo. Described as “an intelligent DNN voice processor,” this system-on-chip processor is intended for “configurable deep neural networks” with a “highly efficient inference engine”, although it is only expected to learn 20-30 words. The idea is that this AI chip could be used for what MediaTek calls “decentralised processing”, whereby AI can be placed in specific devices and controlled entirely locally, such as “turn on” for light switches. For these simple listening and interpreting use cases, a limited vocabulary is fine.
Voice technology is still in its early phase, still developing and paths diverging. Some gadgets are likely to become multi-function devices accomplishing several tasks in one unit,while others will be dedicated to a single, routine task. There is also a sub-industry working on solutions for a hub or central controller for all the smart devices in your home, able to communicate seamlessly with Alexa, Assistant, Siri, Bixby et al, thus theoretically solving the problem of having multiple assistants at your beck and call.
Of course, the success of voice-activated devices is entirely predicated on the end users’ willingness to use them. Almost every article online about smart devices is followed by comments about privacy and how “no listening device is ever coming into my home!”
Still, times are changing and the recent rapid increases in voice uptake have been driven by an uncharacteristic openness by companies in sharing proprietary technologies. Google has followed Amazon’s lead in opening up Google Assistant to other companies in order to get Assistant into as many products as possible. Baidu, known as “China’s Google”, has also opened its Duer OS-based platform to developers. Who the product developer is, companies don’t care too much. Sheer numbers is what matters.
There is also an industry acceptance that loyalty is no longer a prequisite. Sonos, already a partner with Amazon, announced at CES 2018 that it will also support Google Assistant in its speakers this year. For its part, Amazon is happy for Alexa to work alongside other voice assistants, as will be the situation with Microsoft’s Cortana on forthcoming Acer laptops. The message is clear: just get our assistant in there and let the public decide!