Some US researchers at Carnegie Mellon University -- who have clearly missed the basic gist of most traditional tellings of the Tower of Babel story -- are working on a nifty new technology for automatically translating bilingual conversations. Instead of speaking into a device and then waiting for the unit to transform the audio into text, translate that text and then output the translation in audio form (ala IBM's recently rolled-out solution
), the "Tower of Babel" translator allows the conversationalists to mouth the words they wish to speak, which it then translates on the fly and creates an audio overdub of the conversation. The upshot of this is that two people can "speak" to each other face to face in their own respective languages, with minimal delay or confusion. Electrodes are hooked up to the neck and face to sense the mouth movements, but unfortunately the system is still in its infancy. Currently it can handle a small vocabulary of 100-200 words at about 80% accuracy, and accuracy drops off significantly beyond that vocab. The system currently works with translating Chinese to English and English to Spanish or German. Obviously there's a long way to go, but we're looking forward to the day when we can all get along and chat it up Tower of Babel style -- heck, we might as well build a spankin' tall building while we're at it, yeah?