On his blog dedicated to mobile technologies, Google announced a novelty for its voice recognition system. First introduced in 2008 in Google Maps on BlackBerry and on devices from Nokia and Apple, Google Voice Search allows mobile users to dictate a sentence or a few keywords to automatically generate a query on the Internet while use of the GPS chip to return results located. It will then be enough to dictate the word “pizzeria” for a map of nearby restaurants and their phone numbers.
Amir Mané and Glen Shires, product manager respectively and member of the technical team at Google, explained that before introducing this technology efforts have been focused on the development of a broad range of models for dictation understand any query and easily recognize “the focus, changes in tone, speed or other factors” in the voice of the user.
To truly recognize the requests of each mobile user, Google has just introduced on Android 2.2+ personalized voice recognition. Within the application, a new dedicated section thus made its appearance and allows users to prende part in this evaluation phase. Over time, it will be possible to develop and densify its own dictation model to obtain an increasingly optimized accuracy.
Recall that earlier this month, the Mountain View company announced the acquisition of Phonetic Arts, a company specializing in voice synthesis. Mike Cohen, head of ICT at Google, then declared: “Star Trek, they do not spend a lot of time typing on their keyboards. They just talk to their computers and they respond. ” It will be interesting to see what can be the combination of the analysis of a personal dictation coupled to an “intelligent” system of speech.
For now the new Google Voice Search is only available in the United States, the latter should then be deployed in other countries / languages.