What if you could control devices and track down any information you want simply by talking? Solid efforts are now under way to make this pipe dream of the speech-recognition community a reality. ScanSoft, VoiceBox Technologies, and others are eyeing new ways to make the "open sez me" model of device and information control work.
At January's Consumer Electronics Show VoiceBox showed off its conversational-language processing technology and announced deals with Toyota, XM Satellite Radio, and Johnson Controls. If you're familiar with phoning a call center and hearing a prompt such as "Say Sales or Representative," then you know speech-recognition engines are already very accurate for limited vocabularies. But VoiceBox's software uses multiple speech-recognition engines and filters for determining context, making natural sentences understandable.
In a demonstration for PC Magazine, the software understood more than 90 percent of sentences such as "Play me a song by the Rolling Stones" or "What is the weather like in Boston?" Toyota is building VoiceBox's technology into telematics devices in its cars, and XM Satellite Radio is integrating it into its hardware so that drivers can ask for jazz stations or inquire about traffic without taking their hands from the wheel. Johnson Controls is using it in telematics devices so users will be able to link an iPod to a car via Bluetooth and speak the names of songs to play.
"Now that mobility and digital media are ubiquitous, 2006 is going to be the year for voice search," says Mike Kennewick, VoiceBox's CEO. Meanwhile, ScanSoft and music database company Gracenote are developing ways to integrate voice with digital music players. Pretty soon, talking to your tunes may not be such a crazy idea.