REDMOND, Wash. — Nov. 30, 2011 — In the near future, when your car’s infotainment system is sending voice-dictated text messages and reading your e-mail to you, think about how someone first had to figure out how to teach your car to recognize words like “aunt Ethel,” “holidays,” “vegan” and “tofu turkey.”
Hands-free e-mail and texting through voice command are some of the most hotly anticipated features that will soon be available in vehicles with infotainment systems powered by the Windows Embedded Automotive 7 platform. These features enable drivers and passengers to have e-mail messages read to them, and to compose and send SMS text messages through their cars just as they do now through mobile devices — only they’ll do it by voice, not with keypads.
Drivers can reply to text messages using voice controls in which the system matches the driver’s reply to stored messages like “Running late” or “See you in 10 minutes.” This leads to a better overall driving experience and less distraction for drivers — allowing them to keep their eyes on the road and their hands on the wheel. A video from Steven Bridgeland, product manager for Microsoft’s Windows Embedded Business, demonstrates a prototype of these voice apps working together.
But who’s teaching your future car to talk and understand what you’re saying? That’s the result of a partnership between three Microsoft teams: Windows Embedded, Microsoft Research and Microsoft Tellme.
Forming a productive, cross-company partnership, Microsoft Research, Windows Embedded and Microsoft Tellme developed sophisticated voice search and audio processing algorithms for the Windows Embedded Automotive platform that help in-car systems recognize what the driver is saying. The team also helped pioneer the use of its sophisticated, cloud-based speech recognition in a major automobile manufacturer to provide drivers with audio updates on traffic, directions and weather conditions.
When they created the SMS Reply voice solution, developers collected thousands of real-world SMS replies and turned them into templates — for example, the sentence “See you in five minutes.” When a driver tells the in-car system to create an SMS message containing the words “see you in,” followed by a number and the word “minutes,” the system can quickly match that phrase to all relevant templates and insert the correct number of minutes. It will then repeat back a list of the most relevant SMS replies, and the driver picks the best-matched reply to send.
Speech recognition has come a long way in recent years, and recent breakthroughs in technology mean that progress is accelerating — no pun intended. For example, some in-car systems powered by Windows Embedded Automotive use cloud computing to make the product better as people use it: In these cases, what customers tell the system to do improves the system’s overall performance.
As Microsoft continues to develop the Windows Embedded Automotive software platform, drivers and passengers can look forward to communicating with friends, family and colleagues through e-mail and text uninterrupted as they go from computer to phone to car throughout the day. And we’ll be telling you all about it here on the Windows Embedded News Center — but for now, don’t forget to pick up that tofu turkey for aunt Ethel.
