Google planning mobile speech-to-speech translator

Google is working on high-accuracy machine translation and voice recognition

Google has been discussing new software under development which it claims could let mobile phones translate speech "almost instantly".

A report in The Sunday Times said that Google is adding voice recognition and automatic translation to its mobile applications. The firm already offers automatic text translations on its web pages, and has plans to launch the system for mobile devices.

"We think speech-to-speech translation should be possible and will work reasonably well in a few years' time," said Franz Och, Google's head of translation services, in an interview with the paper.

"Clearly, for it to work smoothly, you need a combination of high-accuracy machine translation and high-accuracy voice recognition, and that's what we are working on."

Och acknowledged that translation software still has some way to go, but said that a number of recent improvements suggested that it would soon be to up to the task of translating speech correctly.

"If you look at the progress in machine translation and corresponding advances in voice recognition, there has been huge progress recently," he said.

Automatic text translators are now reasonably effective, but voice recognition has proved more challenging, Och explained.

"Everyone has a different voice, accent and pitch," he said. "But recognition should be effective with mobile phones because by nature they are personal to you. The phone should get a feel for your voice from past voice search queries, for example."

It is this constant use that will make or break translation, according to Och, who claimed that the more it is adopted and used the better and more accurate it will become.

"Google is exploiting its vast database of web sites and translated documents to improve the accuracy of its system. The more data we input, the better the quality," he said.

When contacted Google would not comment further on the news.