Voice Commands Coming Closer to Reality for Telematics
the real opportunities in telematics is to
provide voice enabled commands from the driver to the in-vehicle
navigation system. Few companies have tackled that challenge because of
the difficulties at the human-to-machine interface. VoiceBox Technologies seems to
have conquered that problem and is now
offering tier one and tier two auto suppliers like Johnson Controls
the technology to integrate their products with other on-board
VoiceBox will be working with Johnson Controls to develop hands free
phone dialing and in-vehicle music navigation (i.e. channel selection,
music genre choice, etc.). The company is already working
with XM Satellite Radio to co-develop a voice-search-enabled
reference platform. But the goal for VoiceBox is also to integrate with
any in-vehicle navigation system to provide information about localized
weather, sports events and traffic in order to assist with route
guidance around hazardous or congested areas. The company is hoping to
convince navigation system providers like Garmin and TomTom that its
solution would integrate smoothly with their existing products.
One example provided by VoiceBox that the company hopes to develop is a
way to integrate phone services with navigation. For example, if a
driver receives an inbound call to his or her phone which already has
caller id, the driver would use voice commands to identify the location
of the caller and, if desired, plot a route to the caller's address.
The competitive advantage of VoiceBox is that the company seems to have
solved the problem associated with voice recognition software. The
technology is speaker independent and is able to determine voice
inflection and gender within the first 30 to 40 seconds. It is also
adaptive so that over time the system "learns" more about the drivers
speech patterns. The company makes the assumption that voice technology
will never be good enough and so its objective is to interpret the
intention behind the phrases used by the speaker, especially if the
sentence is somewhat repetitive or disorganized.
VoiceBox is also focused on what it calls "cooperative conversation,"
whereby there is feedback at the human and machine interface (HMI). For
example, there may be confusion as to the voice directions provided by
the speaker. So the machine tries to recognize the phrase and establish
some level of confidence in its meaning in order to limit the set of
options. If the commands are clear, then they will be executed; if not,
the system will offer choices. The company believes that if a good
cooperative conversation model can be implemented, then users will find
it to be a successful experience.