X

Talk to the car with new tech

Voice-recognition technology could ensure that gadget-happy drivers keep hands on the wheel. Photos: Hands-free digital audio

Stefanie Olsen Staff writer, CNET News
Stefanie Olsen covers technology and science.
Stefanie Olsen
2 min read
For anyone who has driven behind someone yapping on a cell phone while coasting at 10 miles per hour in a 40mph zone, here's a disquieting thought: Things may get worse before they get better.

Cell phones aren't the only gadgets in the car anymore. MP3 players, satellite radio, personal digital assistants, DVD players and mapping tools are quickly becoming common accoutrements of the modern car. And that means drivers have more buttons, keypads and click-wheels to twiddle with while navigating the road.

Obvious safety concerns and legislation in various states are forcing car makers to look for a solution--and one may be on the way with new technology that lets people keep their hands on the wheel and "tell" their gadgets what to do with voice controls.

"There's fairly significant demand for 'button intensive' features in the car (like) dual climate zones and satellite radio," which has more than 120 channels, said Jim Pisz, national manager for partnerships at Toyota. "The future for us is in the ability to control all of these features by voice."

Of course, it may be a few years before mass-production vehicles synchronize electronic devices for voice control, but momentum is building for features that let people ask for driving directions or call a friend without using their hands.

Last week, for example, Toyota partnered with a relatively unknown voice-search specialist, called VoiceBox, in Bellevue, Wash. In development for roughly three years, VoiceBox's technology differs from established voice tech on the market because it allows people to speak conversationally to operate car electronics, rather than having them memorize and deliberately sound out commands.

The two companies are developing natural-speech technologies for Toyota's cars, but Pisz would only say that they'll become more common in cars within the next few years. "We're evaluating it at the highest levels," he said.

VoiceBox recently signed a major deal with XM Satellite Radio to add voice-search capability to its channel-rich service, which is available to more than 6 million people in the United States, many of whom listen in the car.

VoiceBox has also teamed with Johnson Controls, one of the biggest technology suppliers to the auto industry. One early product of their deal is a node that lets people search music on Apple Computer's iPod by voice in the car. The product is expected to be available this year.

Scion and Fiat

"Wherever you have a large menu of files to choose from--song files, phone contacts, local directories--voice technology is inevitable," said Veerender Kaul, research manager for advanced auto technology at Frost & Sullivan, a research firm.

Certainly, <="" news:link=""> for car navigation systems has been around for years, as it has for call centers. Many high-end to midrange vehicles like Lexus and Honda's Acura include voice command features for driving directions. But those technologies have long delivered a frustrating experience to consumers, thanks to a limited vocabulary of commands or poor recognition of synonyms and accents.

"The main problem is that most of the voice-based engines haven't been very reliable in the past," said Thilo Koslowski, vice president and lead auto analyst at market research firm Gartner.

VoiceBox's engineers think they can change that. It was founded in 2001 by Bob Kennewick, a Harvard University associate professor with degrees in economics and computer science. He recognized a fundamental problem with existing voice recognition, which required programmers to set up specific dictionaries for a given set of data, and then match speech to text. But users had to say the right words to make it work. Background noises could also muddle the translation.

His vision was to develop technology that could recognize the context of speech, picking up the right cues in a conversation to answer like a human would. For example, a request like: "Let me hear Cisco" could translate to the technology as a request to hear the singer Sisco, get a stock quote on the company Cisco Systems, or listen to the Johnny Cash song, "Cisco Spilling Station." The technology, which Kennewick developed, answers such a request by asking which of these options the person would like to hear.

In some contexts, the technology wouldn't need to provide options. A request for a stock quote on Microsoft followed by the name Cisco would immediately prompt another quote, for example. The VoiceBox technology would learn from experience by recognizing repeated requests for information and responding to personal preferences.

"It looks for clues in what you're saying and what you've said before to infer what you want, just like a human would," said Mike Kennewick, the founder's brother and current CEO of VoiceBox.

Mike Kennewick was a Microsoft business development executive for Windows in the 1980s. The brothers teamed up with two friends to found the company in 2001. It now employs more than 40 speech-recognition engineers.

According to Toyota and industry analysts, VoiceBox is one of only a handful of companies working on conversational voice-search technology.

OnStar, the telematics company inside General Motors, has long been an advocate of voice recognition in cars. Coming up on its 10th anniversary, OnStar has 4 million subscribers and its hardware is set to become standard equipment in GM cars by 2008. With the system, drivers push a button and talk to consultants to report an accident or unlock a door. (The company does about 1,000 airbag calls a month.) People also can make cellular calls from a built-in voice-activated mobile phone if they sign up with a program.

"When you push a button, the radio mutes automatically," said Chet Huber, president of OnStar. "We use voice recognition so your hands are on the wheel. You can say 'call home,' and it will call home.'"

Other big players in advanced search recognition, or voice recognition, include IBM, Microsoft and ScanSoft. Microsoft, for example, sells voice-recognition technology with its operating system for automobiles, but the system responds to commands rather than to contextual speech.

VoiceBox executives think the large companies are potential partners. Toyota, for example, plans to use VoiceBox technology on top of IBM software.

Microsoft's voice technology will be available in European cars made by Fiat next month. Fiat, which makes the Alfa Romeo, will offer voice command features for mobile phones and digital music players that can be hooked up to nearly 23 models of its cars, through Windows Mobile Automotive technology.

"Consumers fundamentally want voice to work--it's in our imagination with shows like the 'Jetson's' and 'Star Trek.'"
--Bob Kennewick, founder, VoiceBox

Honda's Acura runs Microsoft's auto operating system, but "="" software="" and="" hardware="" so="" that="" if="" a="" phone,="" mp3="" player="" or="" pda="" has="" bluetooth="" support,="" it="" will="" let="" person="" dial="" by="" voice="" ask="" for="" songs.="" its="" ipod="" search="" tool,="" johnson="" worked="" with="" apple="" computer.="" the="" mac="" maker="" said="" large="" number="" of="" consumers="" use="" music="" in="" car.="" technology="" is="" commercially="" ready="" lets="" drivers="" songs="" car.<="" p="">

For XM Satellite Radio, VoiceBox is likely to be a welcome feature, because it can give people easy access to relatively unknown features, like regular traffic updates, sports scores or stock quotes. People can say: "What's traffic like on 101?" and VoiceBox will tick off an answer. XM Satellite plans to introduce the product midyear.

Gartner's Koslowski said that Apple is also likely working with the auto industry on voice technology, as well as developing it in-house. He also suggested that all the Internet search companies such as Google will likely look at markets like the auto business, in which people need to search for files but don't have use of their hands.

"Consumers fundamentally want voice to work--it's in our imagination with shows like the 'Jetson's' and 'Star Trek,'" Kennewick said. "We think this will be a more compelling experience for people."

CNET News.com's Michael Kanellos contributed to this report.