Google on Sunday revealed that it could be the first to enable voice-to-voice translation for phones. Using the existing text conversion and voice recognition its search and from Android, the company is developing a system that could recognize a spoken foreign language and recite it back in the user’s native tongue. It would sort speech into segments and would analyze both the pattern of speech as well as Google’s own translation history to construct a better interpretation.
Translation services head Franz Och tells Britain’s The Times that a system whose voice recognition would "work reasonably well" should be ready within a few years. He didn’t mention which platforms might get such a feature beyond Android.
While it has at times been criticized for flawed speech-to-text translation in Google Voice, the company so far has one of the most advanced voice recognition systems of any smartphone OS maker to date. So far, it remains the only company to have voice-dictated map searching that lets Droid and Nexus One owners speak searches for locations based on their type and general area rather than having to look for specific addresses. Android 2.x also allows general voice-guided searching from its home screen.
