Speech Technology Magazine

 

Voice Enables Web Search Behind the WheelListen to this article in TTS, powered by Loquendo

A new multimodal application from ATX lets users search the Web from inside their cars.
By Adam Boretz - Posted Nov 17, 2008
Page1 of 1
Bookmark and Share

Drivers will soon be able to perform hands-free, speech-enabled Internet searches from the comfort of their car, thanks to Browse By Voice, a new application from ATX Group.

Browse By Voice provides an easy-to-use, multimodal interface that allows users to simply speak Google or Yahoo searches to the in-vehicle display—What is the weather in New York? or Where is the nearest Starbucks?—eliminating the need for typing text into a search engine.

“A user accessing the voice application is prompted to say a search phrase,” says Thomas Schalk, vice president of voice technology at ATX Group. “The phrase is essentially anything that could be typed into a search box and the recognizer converts the speech into text and the text is entered into a search box. And then you see the results [on the vehicle display screen].”

According to Schalk, the application’s design—which leverages speech as the primary component of the multimodal user interface—helps to ensure driver safety. “When we do any kind of speech-enabled service in the car, we have to be very aware of driver distraction,” he says. “And the key is to keep it simple and allow the driver to focus on driving, which means eyes on the road, hands on the steering wheel."

“You can speak something and glance at a display without being distracted,” he adds. “But if you were to have to type…that alone would cause driver distraction without a question.”

Built on a natural language, voice-based mobile search platform, users can speak naturally, are not required to memorize a list of commands, and do not need to perform directed searches—single searches that are broken into a series of individual steps. Browse By Voice can also account for regional dialects and adapts to a user's speaking habits, individual pronunciations, and word choices in specific, real-life, in-vehicle circumstances. “The system will adapt acoustic models and language models according to usage patterns,” Schalk says. “As people use this system over time, we perform usability analysis…and with pretty much all deployments there are improvements that happen over time.”

ATX—which in 2001 partnered with Mercedes-Benz to introduce the first Web-based information service into a production vehicle—has demonstrated Browse By Voice for several selected automobile manufacturers. According to Schalk, the service will be deployed in cars within one to three years. 

Additionally, ATX plans to expand the natural language voice platform to allow for hands-free text messages, email composition, and the management of social networking. “The whole idea is to extend beyond the browsing,” Schalk says. “Extend and use the same approach to allow short message dictation and that would include outbound texting, possibly some outbound emailing, and some other services that might require that same kind of flexibility as far as speech recognition handling goes.”

Page1 of 1