Mountain View (CA) – The United States Patent and Trademark Office has approved Google’s patent application for a voice assisted search engine. Patent number 7,027,987 is titled “Voice interface for a search engine” and outlines how Google wants to improve voice recognition specifically for providing a the search results a user may be looking for.
According to Google, today’s voice recognition technologies are not suited to be used for carrying out serach queries. For example, hurdles include high error rates in correctly recognizing words and a lack of context sensitive search. In its patent application, Google explains an example of how voice recognition systems could confuse similar sounding word combinations, such as “White House” and “light mouse.”
To solve this problem, Google intends to apply a percentage weighting not only to individual words, but also to word combinations. Presumably since White House is searched for more often than “light mouse,” results to such a serach would be listed first.
Google has been testing a voice assisted search engine called Google Voice Search for some time. The engine allows users to call in search terms with a normal telephone and then view the results on a web browser. According to the GVS homepage, the test has very limited capacity and it has been taken offline.