Learning other languages may become a thing of the past... not because everyone will begin speaking a single language, but because speech-to-speech translation is becoming increasingly sophisticated.
New speech recognition technology measures muscles in the speaker's face and neck to translate his or her speech into multiple languages. Such a device, developed by Carnegie Mellon University, was debuted at the International Workshop on Spoken Language Translation 2005, held October 24-25 in Pittsburgh. Currently, these monitors are in the form of electrodes, but eventually they could be implanted into one's throat. Instead of using grammatical rules, as does most contemporary translation technology, these new approaches use statistical analysis to make the translation.
The technology remains imperfect, but its developers believe that a commercially viable version could be ready in 10 years. Becoming multilingual, then, would go from spending years studying multiple languages to a simple surgical procedure.
Source: Roland Piquepaille's Tech Trends (ZDNet)
No comments:
Post a Comment