Press release: Speech technology gets personal

Computers can convert written sentences into speech using a technology known as "text to speech". But, the user usually has no control over the voice that is used, or can only choose from a few standard voices. In applications such as translation of speech, where you can speak in English and hear the translation spoken in Japanese, wouldn't it be better if the Japanese speech sounded just like you, even though you don't speak Japanese?

In a new project, researchers from Europe and Japan are developing ways to make the computer speak using any voice in any language. The computer will learn how to sound like just you, simply by listening to you speak. You could then speak Japanese, Chinese, English or any other language you like, using your own voice.

They plan to demonstrate their new techniques by building a handheld personalised speech-to-speech translation system, that will allow conversation between people that speak different languages.