In the past two weeks, Ford and Toyota have both come out against handing their dashboard electronics over to Google and Apple. They want to defend sunk investments in their own systems and keep the customers’ data—and the customers—to themselves.
That’s wonderful news for smaller companies like VoiceBox, a company in Belleview, Wash. that specializes in computer voice recognition.
"We allow people to differentiate their product,” says Victor Melfi, the company’s chief strategy officer. “If you’re worried about commodification of the car business, then why give it to Apple or Google?” He adds that the same logic applies to any company that uses talking machines. "AT&T spends a lot of money to get customers and keep them. Why cede that customer to Google?"
VoiceBox’s system deduces what a person is saying by using the context of the conversation to fill in holes that form when words get slurred or the background gets noisy. That cuts down on the number of times the machine has to bleat, "I'm sorry, I didn't understand what you said."
Just as important, the system uses context to manage a conversation so as to avoid having to drill down from one menu of options to another. "If you say, 'how’s the weather in Boston?' and then add, 'how about New York?' it knows you’re still talking about the weather," Melfi says.
He says the company used to train its system on subjects in its own audio lab. Now it crowdsources some training, à la the Mechanical Turk. The rest it handles with machine-learning techniques.
The obvious reason why voice-operated interfaces and other “embedded systems” are taking up more of the value of a car is the simple cost of developing computational wizardry. The subtle reason is that the emerging market puts more stress on such wizardry than on good, old-fashioned mechanical muscle.
“My daughter’s only care is if she can plug her iPhone into it,” Melfi says. “I could buy her a $19,000 car or a $90,000 car, and it’s all the same to her.”
Car companies are clearly worried about the declining interest in cars among the young. Only now are they beginning to understand that the top-end cars may not always be the best place to introduce their latest interactive features.
“Toyota put our system into the Lexus first,” notes Melfi. “Wrong! They should have put it in the Scion first! Kids are happy to play with the technology, to interact with it—but my mother-in-law doesn’t want to fool with it.”
VoiceBox is now in all of Toyota’s models. It’s also used by Fiat, Chrysler, Mazda, and, as revealed this week, by Subaru. VoiceBox revenues have risen by 80 percent or more in each of the past three years, and now it’s got its eye on non-automotive applications, like call centers and mobile devices.
It’s a complete strategy for the size-challenged tech company: being the non-Google, non-Apple alternative.
Philip E. Ross is a senior editor at IEEE Spectrum. His interests include transportation, energy storage, AI, and the economic aspects of technology. He has a master's degree in international affairs from Columbia University and another, in journalism, from the University of Michigan.