Microsoft, the world's largest software provider, seems to be enjoying considerable success in the world of emerging technologies at the moment.
Just days after it won a patent for cameras embedded into TVs, PCs, and other devices that commonly sit in your living room, it has demonstrated a super-speedy English to Chinese translation programme in Tainjin, China.
Microsoft’s chief research officer, Rick Rashid, outlined the new software's history and operation, describing the two processes involved in the translation in an 8 November blog post: the first is the replacement of the English word with its Chinese equivalent, and the latter is the reorganisation of the words to correspond correctly with Chinese grammar and sentence structure.
Apparently, this technology was built on prior breakthroughs and heralds a move from the approach used in previous speech translation systems. Statistical models - which can better process the nuances of the human vocal range - have become the basis for the current generation of translation programmes.
The new software is claimed to represent an accuracy improvement of some 30 per cent on the current generation of English-to-Chinese translation models, which generally have an error rate of between 20 to 25 per cent. That makes for a ratio of one wrong word for every five spoken, whereas the new standard Microsoft is developing should only produce one incorrect translation per every five.
The translator also preserves the speaker's voice, allowing the translation to maintain its original vocal fidelity (as seen in the video, above.)
Rashid described the technology as “very promising,” but constantly alluded to the software as being a work in progress that will still need a few years to be market-ready.
“In other words, we may not have to wait until the 22nd century for a usable equivalent of Star Trek’s universal translator, and we can also hope that as barriers to understanding language are removed,” read a further statement from Rashid’s blog post.