Monday, February 8, 2010

Google Rebuilds the Tower of Babel with Real-Time Language Translation

Google Rebuilds the Tower of Babel with Real-Time Language Translation: "

googlelogo6.jpgIf our attempts at getting such simple information as bus schedules or account balances from automated voice recognition systems are any indication, then we imagine Google has a lot of work to do in its latest endeavor - real-time, spoken-language translation.



According to the Times UK, Google is working on developing software for a mobile phone that would translate what you were saying into the language of the speaker on the other end of the line and vice versa.


Sponsor



As you may have noticed, Google already has a hand in the translation business, with its web page translation service. Google Translate currently translates between 52 languages, which includes a number of languages with completely different alphabets.



The Times UK spoke with Franz Och, head of Google's translation services, who said that this new service should be up and running and 'work reasonably well in a few years' time.'



'Everyone has a different voice, accent and pitch,' said Och. 'But recognition should be effective with mobile phones because by nature they are personal to you.'



Och is referring to the fact that the software would have the opportunity to learn your accent, dialect and general manner of speaking over time, becoming more accurate. But we can only imagine the difficulty of the task ahead, especially with languages such as Mandarin or Cantonese, which are tonally based. In Mandarin, for example, the word 'ma' can have four different meanings according to the tone used. If the speaker uses the first tone, a constant high pitch, then the word means 'mother'. If they use the third tone, a dropping then rising pitch, however, the meaning changes to 'horse'.



The fun doesn't stop there, the Times UK article points out, as handling the vast number of accents and dialects is also an immense task. Much like the web-based translation that Google does, though, the system would become more accurate over time, essentially learning from its experience.



We hope that one of the first things it learns is not to call our new Chinese friend's mother a horse.


Discuss



"

3 comments: