Ford and Nuance Collaborate to Advance Speech Understanding and Intent

Ford is working with its voice technology partner Nuance Communications to find a way to make voice recognition even easier and more intuitive for drivers to successfully use, especially first-timers.

Ford and Nuance are testing a number of new algorithms, based on intent and meaning, for the voice recognition system in Ford SYNC that is powered by Nuance voice solutions and Microsoft's Tellme platform. These new techniques can flag common words and phrases that drivers might typically use in conversations, yet are not the specific commands or syntax required by today's SYNC system. By interpreting or understanding the user's request, SYNC will be able to execute the command or respond by coaching the user down the correct command pathway.  

"With each generation of SYNC, we have learned more about how drivers use the voice recognition system, and have continuously refined it so customers can do more and say more to get their tasks done more quickly and efficiently," said Brigitte Richardson, Ford global voice control technology and speech systems lead engineer. "With intent, we are examining how to take our voice recognition to the next level of command and control, helping further reduce the learning curve and improving ease of use, and ultimately building higher customer satisfaction. In the end, we want the user's wish to be the system's command."

New SYNC owner market research, conducted by Ford, shows that acceptance among owners of voice recognition continues to grow, with more than 85 percent of owners now using voice control while driving. That's up from 70 percent in the initial study conducted in the fall of 2010.

The true power of understanding intent comes into play if a user is unfamiliar with the exact commands for a SYNC feature or doesn't give enough information to complete a command. With intent algorithms at work, for example, if a user says, "I wanna call John Smith," the request could be accurately interpreted for the official SYNC command, "Call John Smith." Similarly, with a command such as, "I want to change the temp," the intent algorithms could aid the system in guiding drivers to provide more information about whether they want to make it warmer or colder in the cabin, and what temperature they would prefer.

"If the driver doesn't give enough detail or even uses common lingo, intent-based technologies can help the system ask simple questions for more information to complete the command, helping eliminate confusion and frustration, and boosting usability and satisfaction for customers," said Richardson.

Working closely with Ford's speech team, Nuance has created several language models for the core SYNC voice-activated functions and services, including phone, music, climate, navigation and traffic. With these language models, the SYNC system is able to interpret what the user is intending to do, and then complete the task.

"Nuance is continuously evolving its voice technology to provide drivers with a more natural and conversational interface," said Vlad Sejnoha, chief technical officer at Nuance. "By combining our advanced voice recognition capabilities with intent and language understanding, we're not only able to hear what drivers are saying, but better understand what they are looking to accomplish, be it listen to songs by Train or change the temperature to 75 degrees.

While work with intent algorithms continues in the Ford and Nuance research labs, customers have already seen a number of refinements and improvements to the SYNC voice recognition system found on more than 3 million vehicles today.

Since its launch in 2007, SYNC has been continuously improved and can now understand 100 times more commands than the original system, allowing users to take fewer steps and use more conversational commands to control SYNC features and services.  

Additionally, Nuance's voice technology at the heart of the SYNC system features "voice learning" capabilities that become more accustomed to the driver's voice over time. So while there is no training required, SYNC becomes familiar with the voice of its driver for a consistent experience. Nuance's natural text-to-speech technology is also behind "Samantha," the voice of SYNC, providing drivers with a more natural, conversational experience with their Ford vehicle.

All of these refinements are designed to encourage drivers to take advantage of the hands-free capabilities of SYNC and keep their focus on driving."It's only through continuous improvements that Ford will maintain a leadership position in voice-activated vehicle controls, helping our customers keep their hands on the wheel and eyes on the road," Richardson said.