If one ever wanted to describe an intractable problem this is it. The hard reality is that for this to achieve near perfection, it willbe necessary to produce artificial cognition that simulatesourselves. The speaker needs to know when the wrong choice has beensupplied. It also needs to anticipate.
Yet we have already learned that we can work with near enough forincoming information and human cognition. The other way is quite adifferent matter.
Yet this is were a coaching protocol is necessary just as we do inlanguage teaching. Perhaps we can get there yet. It will still be agreat improvement in productivity.
Microsoft“universal translator” chatters Chinese in your own voice
Chris Davies, Nov 9th2012
http://www.slashgear.com/microsoft-universal-translator-chatters-chinese-in-your-own-voice-09256313/
Microsoft has developed a “universal translator” that notonly converts English speech into Chinese in real-time, butdoes so while preserving the speaker’s own voice.Demonstrated in China recently, the technology is based on jointresearch into Deep Neural Networks by the software giant and theUniversity of Toronto, Microsoft’s chief research officer RickRashid writes, using an hour’s worth of prerecorded speechexample data to cut together a new, translated mashup.
Microsoft firststarted talking about the universal translator project earlierthis year, revealing that the system can handle Spanish and Italianin addition to Mandarin Chinese. Recordings from both the speaker anda native Chinese speaker are required for the English-Chineseconversion, with the properties of the English speaker mapped onto afew hours’ worth of Chinese speech that can be reworked to suit abroad variety of phrases.
The big differencebetween early presentations of the technology and the October 25 demoin China is the degree of accuracy Microsoft has managed to achieve.Rashid says the word error rate has been cut by around 30-percent,meaning around one word in 7 or 8 is now incorrect.
“Of course, thereare still likely to be errors in both the English text and thetranslation into Chinese, and the results can sometimes be humorous”Rashid warns. “Still, the technology has developed to be quiteuseful.”
Further refinementwill be enabled when Microsoft loads in more training data, it’sexpected, as the Deep Neural Networks learn more about how the humanbrain processes audio. Still, there’s no telling when the systemmight show up in your Windows Phone.
Hiç yorum yok:
Yorum Gönder