Google patents voice queries
Google has been granted a US patent for searching technologies using voice queries.
First applied for in February 2001, patent 7,027,987 covers a ‘voice interface for a search engine’ and names Google co-founder Sergey Brin among the inventors.
It covers methods for technologies to recognise a voice query and chop it up into its components which can then be manipulated to the form of a Boolean query. The remainder of the patent covers methods of refining the results based on different weightings.
However, it still appears there are technical barriers yet to hurdle. As Google states in the filing, searches are often short, lacking context, and yet have no restriction to the vocabulary used: this presents a problem to the traditional voice recognition model that relies on training recognition software over time in order improve accuracy.
However, Google remains coy about its plans for such an invention and has warned not to predict products on the basis of patent progress.
Even so, success at this level opens up the already lucrative context-based search advertising from the browser to include mobile phones, for example, or in-car services, where the driver needs to be looking at the road rather than a screen.
Google isn’t alone in working in this area. Much of the high-profile developer headhunting over the latter half of 2005 involved this area of technology. Kai-Fu Lee, Google VP of engineering and president for its China operations, was recruited in a high-profile legal tussle from Microsoft where he was working on voice applications in its Natural Interactive Services Division.