Speech Recognition GPS Navigation

Research into speech recognition has been ongoing for over 25 years and has had proponents such as Defense Advanced Research Projects Agency (DARPA) and IBM. In recent years this significant body of research has begun to pay off with more consumer computing and electronics hardware incorporating speech recognition command systems.

A recent survey by VoiceBox, a leading speech recognition products supplier, revealed that there is a strong demand for speech-enabled features for personal navigation systems among consumers from Germany, Japan and the United States. They can be very useful for those who want to focus on driving, and if you’re often using the latest o2 business phones and their GPS plotting, voice recognition seems like a logical addition. Some 70 percent of the survey respondents had the belief that voice control of navigation functions would improve safety. Another speech solutions market leader, Nuance Communications, has conducted research (Automotive Voice Interface User Survey conducted by Maix Research) revealing that eight out of nine owners of speech-enabled in-car systems and GPS navigators regularly use the voice recognition capabilities.

While evidence indicates that most consumers believe that speech recognition voice control systems improve safety, there is conflicting evidence as to whether this is actually true.

A study, Cell Phones and Driving: Research Update , released by The American Automobile Association (AAA) Foundation for Traffic Safety suggests that using hands-free and/or speech-enabled technology while driving may be almost as dangerous as manually entering data into a physical interface. There conclusion was that the cognitive part of the distraction is just as dangerous as the physical part.

The University of Braunschweig’s (Germany) research study for Nuance found that speech recognition control systems significantly reduce driver distractions as compared to manual input of navigation and entertainment systems. Through analyzing drivers’ eye movements they discovered that, on average, speech recognition systems help keep drivers eyes on the road 200 percent to 300 percent more than manual input.

One thread that seems consistent is that any speech recognition system needs to be as simple to use as possible to reduce cognitive distraction just as, for example, any menu trees for GPS navigation displays need to be as intuitive and simple to use as possible.

There are two main types of voice recognition and command systems available. One involves consumers either memorizing different commands for different contexts. On-screen icons and text can provide hints as to the structure of the command instruction required. The alternative, and personally preferred, speech recognition system is one that understands conversational speech. Memorizing commands and checking the navigation interface is more “cognitively distracting” than being able to use normal conversational speech.

Voice Recognition and Control GPS Navigators

Pioneer’s AVIC-F500BT in-dash GPS navigator, selected as the Wall Street Journal Technology Innovation Award’s runner up implements VoiceBox’s Conversational Voice Search Platform. VoiceBox developed specialized software programs that overlay existing speech recognition engines like those from I.B.M. and Nuance. The software is designed specifically to interpret the driver’s intention and determine the correct context for a given instruction. The software is impressively complex being able to extract from conversational commands relevant from irrelevant words, filter out background noise and correctly determine user intent (intent recognition). If unable to resolve a query the VoiceBox software can ask clarifying questions.

VoiceBox and Nuance are now collaborating on applications that will combine Nuance’s speech technology and natural language suite with VoiceBox’s conversational voice search processes.

The Garmin 855 and 885T use a push-to-talk wireless remote mounted on the steering wheel to initiate spoken instructions. Using a vocabulary matching the buttons on the touchscreen display users can navigate the menus by spoken commands. Conveniently, the voice recognition does not require any initial training to improve accuracy. Speech recognition control can be used to speak any text on the screen, shortcut commands, popular places and addresses. While Garmin’s speech recognition is a useful tool it requires the use of a structured voice instructions. To see how Garmin’s speech recognition works have a look at the demonstration video by Garmin.

TomTom’s GO 920 and GO 920T use Nuance software for voice destination entry. Started with the pressing of a button, the speech software enables one to say street and place names as well as numbers. A limited number of word commands such as yes, no and OK can be used in conjunction.