Microsoft develops technology that can translate real-time voice
Microsoft is working hard to develop a technology that allows real-time translation. The remarkable thing is that the translation also can be spoken in the voice of the person who recorded it.
Microsoft has put online the video below which an early demo is given of the new technology. From about 6 minutes into the video shows that the spoken voice is recognized by the system and then translated into Mandarin. So far everything seems still quite normal. But Microsoft has gone one step further by translation be spoken by the computer. The most remarkable here is that the voice of the importer is imitated by the system. The speaker should be above first hour of entering text, but eventually only a few sentences of audio needed to pull this together. The technology is based on Deep Neural Networks, says The Next Web. The technology 'think' like the brains of a man and a man tries to imitate his mistakes and corrects them then. This is the technology currently to 30 percent better than its competitors, and the error rate is so much lower. Microsoft is expected to have a few more years of research and development needed to develop the technology to perfect. Nevertheless, it is already a very impressive technology that already seems to work remarkably well.
Microsoft has put online the video below which an early demo is given of the new technology. From about 6 minutes into the video shows that the spoken voice is recognized by the system and then translated into Mandarin. So far everything seems still quite normal. But Microsoft has gone one step further by translation be spoken by the computer. The most remarkable here is that the voice of the importer is imitated by the system. The speaker should be above first hour of entering text, but eventually only a few sentences of audio needed to pull this together. The technology is based on Deep Neural Networks, says The Next Web. The technology 'think' like the brains of a man and a man tries to imitate his mistakes and corrects them then. This is the technology currently to 30 percent better than its competitors, and the error rate is so much lower. Microsoft is expected to have a few more years of research and development needed to develop the technology to perfect. Nevertheless, it is already a very impressive technology that already seems to work remarkably well.
