Synaptics Calls for Voice Standards

Synaptics Calls for Voice Standards

SAN JOSE, Calif. — Emerging voice services need standards, said the chief executive of Synaptics as he works to close deals to buy Conexant and a unit of Marvell. Meanwhile, the company expects to be among the first to use a 40nm flash process for its display drivers.

With the Conexant acquisition, Synaptics will have an instant presence in silicon for voice services from Amazon’s Alexa, Baidu’s DuerOS as well as carrier services in Korea. The company positions the services as a natural extension of its work in touch interface and display chips.

“Voice will be more important in PCs, printers — in any system that you can think of touch or voice or displays will have a role,” said Rick Bergman, chief executive of Synaptics.

He called for greater ease of use in the underlying natural-language processing technology in an interview with EE Times.

“You don’t want 15 flavors of home assistants communicating in different ways. We need standards such as standard commands. You don’t want to learn different digital languages,” Bergman said. “It’s a little early, but maybe the ultimate vision is that AI gets so good it’s smart enough to figure commands out independent of device types,” he added.

An analyst agreed, noting consumer Internet of Things markets have developed more slowly that predicted, in part due to poor ease-of-use.

“Voice interfaces help, but natural-language processing is a misnomer. There is nothing natural about talking to today’s devices,” said Mike Demler, a senior analyst with the Linley Group.

Demler gave an example of a voice device that failed to respond to a simple request. His wife found from research online that it required use of the word “to” before the name of the device.

“I needed to say, ‘Play jazz to bedroom,’” Demler said. “Right now these devices are teaching us to speak their language rather than the other way around,” he added.

The good news is the market is moving. Conexant claims it has as many as 60 design wins for its far-field voice chip set. They include an HP PC, Harmon Kardon speakers and an LG refrigerator.

“There are so many potential applications — car speakers, lighting, thermostats, TVs, set tops…the nice part is we can participate in various ecosystems” such as services from Amazon, Baidu, Samsung and SK Telecom, Bergman said.

It’s a crowded field with Intel, Mediatek, Nvidia, NXP, Qualcomm, RDA Microelectronics, Realtek and Texas Instruments already engaged.

“We felt comfortable Conexant’s far-field technology and other audio algorithms were best in class. Its tough stuff to do, and they’ve worked on it for a decade,” Bergman said.

Next page: A toothpick-shaped touch-display driver

text

Sources project nearly 25 million smart speakers and other personal assistants could ship this year. Click to enlarge. (Images: Synaptics)
PreviousBluetooth Mesh Bucks Short Range Reputation, Consumer Roots
Next    Cypress CEO Enjoying 'High-Calorie Growth'