Telematics and speech recognition: Finally ready for prime time?

Humans are natural born talkers. But when it comes to communicating with machines, technology has too often been at a loss for words.

“The speech interface can enable comfortable, human-like interaction with a complicated piece of technology,” says Paul Sykes, segment marketing manager for infotainment, Automotive Business Unit, Renesas Electronics America.

“Speech technology is evolving so that, instead of having to use specific commands, you will be able to converse with it in a very natural way.”
While most major OEMs have implemented some kind of voice interface into in-car systems, it took Ford SYNC to make it sexy.

“Almost every commercial for Ford has a consumer talking to the car,” Sykes notes.

(Renesas provides 32-bit microprocessors for mobile multimedia applications; Sykes would not comment on any relationship between his company and Ford.)

Good listeners

Kia Motors America is another company aiming to differentiate itself with cars that listen and talk back to drivers.
Kia’s UVO in-car communications and entertainment system, expected to roll out in some models this year, promises a voice- and touch-activated interface for hands-free mobile phone use and SMS texting, as well as managing music files.

Using the mobile phone as the conductor of speech and data allows UVO to take advantage of the superior voice-recognition capabilities of phones, says Henry Bzeih, Kia’s national manager, connected car.

“The smartphone is good, and the servers they connect to are really good,” he points out. (For more on connected cars, see ‘Making the connected car a reality’.)

For in-car systems, speech recognition is limited by the amount of processing power available on the head unit.

When cars become connected to the Internet, though, the intense number-crunching necessary for a machine to interpret and respond to natural-language speech can take place on remote servers.

Communicating safety warnings

That strategy will be fine for infotainment, but it may not be good enough for communicating safety warnings, according to Dominique Bonte of ABI Research.

“The main advantage of off-board processing is that you can tap into very powerful computers; one disadvantage is there might be a small additional delay,” he says.

“You have to send the speech to the remote server to have it interpreted and then send the results back to the car.”

Another disadvantage, Bonte points out, is that the car would need to have coverage for off-board speech processing to work, and many remote places—places where your car’s system may be your only link to help—might have no connection.

So, Bonte says, “it should be off-board but also onboard in case you don’t have a connection.”

But there’s more to designing a good speech interface than voice recognition, adds Bzeih.

For example, some speech-activated systems require the user to go through multiple commands, such as saying, “phone” before you can tell the system whom to call.

Apps and safety

While a voice interface lets the driver keep her hands on the wheel and her eyes on the road, it’s a mistake to assume that it’s therefore safe.
Listening and responding to email is quite different from chatting with a passenger.

Using some in-car applications draws on the same parts of the brain that are used for driving.

If you’re listening to email, for example, part of your brain’s processing power is not available for paying attention to the road.

A 2001 study found that using a speech-based email system slowed a driver’s reaction time to a braking vehicle by 30 percent.

Because of this and other research, the telematics industry should keep its eye on the intersection between applications and safety and not assume that a speech interface is the panacea.

“The whole car industry is very focused on safety,” says ABI’s Bonte.
“But even if your hands are on the wheel and you’re looking straight ahead, just listening to information might represent a form of distraction.” (For more on driver distraction, see ‘How telematics can help prevent driver distraction’.)

For more on voice recognition and the future of the HMI, order Telematics Update’s must-read ‘Human-Machine Interface Report’ today.

Susan Kuchinskas is a regular contributor to TU.