Japan’s internal affairs ministry to enhance speech translation system for Olympics

JIJI

The internal affairs ministry plans to beef up its multilingual translation system to prepare for the 2020 Olympics in Tokyo, according to its chief.

The system, dubbed VoiceTra, instantly translates messages input in Japanese, English and Chinese, for instance, into another language and then reads them out. The technology can help remove language barriers, Internal Affairs and Communications Minister Yoshitaka Shindo said.

The ministry aims to have the system, which has already been put to practical use for tourists via an application for smartphones, translate conversations related to medical care and disaster responses. It hopes the enhanced system will appeal to foreign visitors who may be concerned about the safety and security of Japan after the March 2011 natural and nuclear disasters, an official said.

The system was developed by the National Institute of Information and Communications Technology, an independent institution under the ministry.

Since 2010, when voice translations became available in six languages, the entity has worked with research institutes abroad to make the system applicable to other languages. It currently supports 27 languages, with voice translations available in 17 of them. It also enables up to five people to hold conversations at the same time.

The ministry and the institute plan to raise the number of languages to 10 or so, mainly those of countries supplying a large number of tourists to Japan, including some Southeast Asian nations.

They aim to develop technologies to translate sentences consisting of 10 words or more and to accurately recognize speech even there’s background noise.

The system also will be improved to provide precise translations of subtle nuances, and it will be enhanced so that users can communicate and evacuate smoothly in the event of earthquakes and typhoons.

The ministry is considering taking steps, including increasing budget allocations, to promote the system.