Microsoft's New Software Decodes and Translates Speech in Real-Time
As technology comes ever so closer to mimicking science-fiction, Microsoft has unveiled a software that translates spoken words into different languages. This was demonstrated in Tianjin, China by Redmond's chief research officer Rick Rashid, in which his English speech was translated with only a slight delay into Chinese. Rashid's words were displayed on a screen in both English and Chinese, which was followed by the words being spoken in his own voice in Chinese. The technology is based on recent improvements in voice recognition and cognitive prediction, with the virtual translator managing to cope even when the spoken language sounds unclear. It processes each word according to the chosen language, which the software then outputs into speech, which is also arranged into a cohesive structure based on the new language. The end result is also processed using the original speaker's voice, which is sourced from about an hour's worth of automatically modulated vocal data from the user. As the technology is still somewhat in its' infancy, Rashid discloses it is not yet known what the limits of its accuracy are. But in the very near future, the language barrier could just be easily overcome with Microsoft's breakthrough.