Conversational voice recognition technology from VoiceBox will be used in future Toyota cars for easier control of telematics. According to CNET News.com, the new VoiceBox technology will allow users to speak naturally, rather than confine them to a rigid list of commands. VoiceBox products, using algorithms called Knowledge Enhanced Search and Speech Recognition (KESSR), use contextual clues and prior usage to determine the intent of what might otherwise be an ambiguous command. Further still, the system will prompt the user with multiple options if it cannot decide how to proceed. According to PC Magazine, VoiceBox software understood 90 percent of sentences such as "Play me a song by the Rolling Stones" or "What is the weather like in Boston ?"
VoiceBox has also partnered with XM Satellite Radio to develop voice-search features for subscribers and with Johnson Controls, one of the largest tech suppliers to the auto industry. The first product of this partnership, a voice-search device for Apple’s iPod, will be released this year.
Voice-search is becoming increasingly necessary as we fill our lives with content-laden devices such as PDAs and iPods and need to quickly cut through the clutter of multiple feeds of information. Enabling users to use natural speech for these search tasks is a commanding step forward—and not just in attention-demanding environments like the automobile.
In an age in which voice recognition technology is as apt to connect my call to China as it is to Charleston, the thought of it quickly creeping into the automobile scares me.
Ages ago, when I was running a telecommunications software company with dozens of DSP engineers working on voice recognition software, a client at a major wireless carrier who was testing the technology sent me a clip from his local paper with a photo of a horrific automobile accident. He'd scratched out the caption under the photo and penned-in "Frustrated by the lack of precision of his hands-free cellular dialing system, Tom Smith rammed his car into a light pole yesterday." In my experience, the state of the art has not advanced far enough since then, especially to let it any closer to the drive train!
Posted by: Russ Pillar | January 27, 2006 at 06:12 AM
Fortunately, the VoiceBox partnerships appear to deal primarily with implementing better, less frustrating device operation in the automobile.
I think we're a long way from commands like "Car, turn left" to which it might respond "I'm sorry Dave, I'm afraid I can't do that."
Posted by: Micah Halsey | January 30, 2006 at 10:25 AM