Went along to a Google Open House event, at the end of last week, and it was interesting for a number of reasons.First, it emerges that Google UK's Mobile Application Services group (numbering some 300 people) contributes functionality to Google HQ, over and above localisation (apparently not a coincidence, but stemming from the UK's (relative) mobile telephony heritage).
Secondly, Hugo Barra - who heads up the group - gave an interesting demo of a "SatNav"-like app on the new Motorola Droid (or DEXT, for those of us in Europe).
It turns out that Google Maps (for mobile) applies intelligence to its 'estimated destination times'... It calculates the relative positions of different users along the same route to feedback more accurate info to users. This learned accuracy also applies to different times of the day... They are calculating journey times for that route and that time. (They triangulate your position, with the GPS info being passed automatically to the app.)
Thirdly, Barra was demoing the speech recognition aspects of the Android-based Droid and highlighted voice processing as a good example of cloud computing. Never thought of that before. Apparently, the direct audio file is uploaded and decoded in the cloud, where greater processing power can be thrown at the task of decoding someone's accent.
No 'training' of the voice recognition system is required.
For more info 'enter 'My Location' into Google's YouTube channel.
Thirdly, Barra was demoing the speech recognition aspects of the Android-based Droid and highlighted voice processing as a good example of cloud computing. Never thought of that before. Apparently, the direct audio file is uploaded and decoded in the cloud, where greater processing power can be thrown at the task of decoding someone's accent.
No 'training' of the voice recognition system is required.
For more info 'enter 'My Location' into Google's YouTube channel.