<About awarded research>
The 2015 Best Paper Award was given to the following two papers. The former focuses on automatic estimation of accent sandhi only from input text, which aims at improving the quality of synthesized speech, and the latter introduces that technique to develop a web-based system to support learners and teachers of Japanese. The system is called OJAD and it is widely used by a huge number of learners inside and outside Japan.
Accent Sandhi Estimation of Tokyo Dialect of Japanese Using Conditional Random Fields
Development and Evaluation of Online Infrastructure to Support Teaching and Learning of Japanese Accent and Intonation
After publishing those papers, several new techniques are proposed and the accent sandhi estimation performance can be improved by those techniques. As for OJAD, a large number of participants in Japanese speech contests are using this system and we're planning to host an Online Japanese Speech Contest World Cup by using speech technologies and ICT technologies.