Method for Simultaneous Speech Translation achieves 20% Speed Increase-Applications in Automatic Interpretation of Lectures and Meetings-

Information Science 2013/08/27

Professor Satoshi Nakamura and Assistant Professor Graham Neubig
Laboratory of Augmented Human Communication, Graduate School of Information Science

Researchers at the Augmented Human Communication Laboratory of the Graduate
School of Information Science at the Nara Institute of Science and
Technology (NAIST; President Naotake Ogasawara) have developed new
technology that allows for automatic speech translation to start translating
before the speaker has finished a sentence, allowing simultaneous speech
translation systems to move closer to the style of translation achieved by
human interpreters.
According to Professor Satoshi Nakamura and Assistant Professor Graham
Neubig of the laboratory, this technology can be used to achieve fast,
accurate translations in the simultaneous translation of news, lectures, and
other situations where long utterances cause traditional translation systems
to suffer from a large delay in translation results.
The results of the research were presented on August 29th at the Annual
Conference of the International Speech Communication Association
(INTERSPEECH) in Lyon, France, and on August 29th-30th at the "Innovation
Japan 2013 University Trade Show and Business Matching" event.


Back to Research Achievements