News

Linguamatics natural language speech recognition allows control of appliances (21/11/2005)

Can you imagine calling your iron from work and telling it to switch itself off? Or perhaps ringing while stuck in a traffic jam to order the kettle to be boiled on your arrival? Researchers at Cambridge-based Linguamatics Ltd have developed a way of talking to appliances, bringing the concept of an "intelligent" home one step closer.

Speech systems are becoming widely used for such applications as in-car navigation, and as an alternative to call centres. However the heavily scripted approach used in these systems does not transfer well to the dynamically changing home environment.

The Linguamatics solution overcomes this problem with natural language processing, a technology that allows us to talk naturally to a computer. It uses a system of ontologies - controlled vocabularies that describe the appliances and the relationships between them, ensuring that appliances and services can be changed without reprogramming the system. "This is a huge step towards a home which can understand what we want of it, and respond accordingly," said Dr David Milward, CTO of Linguamatics. "The really ground-breaking element is how natural the whole system feels - the user doesn't have to adapt his or her speech in order to interact with each appliance."

The Linguamatics' system, in partnership with Loughborough University, is currently being demonstrated at the Advantica Test House, set up by energy and utility services company, Advantica. Linguamatics have also been working with several partners in the EU TALK Project, including BMW and Bosch, on further exploring the possibilities of speech recognition for use in the home and car. This includes being able to programme the house itself with commands such as "turn on the hall light when the front door is opened."

The uses are not all as futuristic as they sound - it could enable the elderly or the disabled to stay in their own homes supported remotely by a warden.

"Effective speech recognition has important implications," said Milward. "For example it could be used by somebody with severe arthritis who struggles with switches and dials. It is also easy to implement on a large scale - for instance in a nursing home or hospital - to allow bed-ridden patients to adjust their surroundings, or to control sophisticated equipment with simple commands."

www.linguamatics.com