Paper Abstract and Keywords |
Presentation |
2017-08-28 16:10
Improvement and Evaluation of a Machine Translation System from Japanese Texts to Sign Language CG Animations for Weather News Naoto Kato, Taro Miyazaki (NHK), Seiki Inoue (NHK-MT), Tsubasa Uchida, Makiko Azuma, Shuichi Umeda, Yuko Yamanouchi, Hideki Sumiyoshi (NHK), Nobuyuki Hiruma (NHK-ES), Yuji Nagashima (Kogakuin Univ.) WIT2017-19 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We have been developing a machine translation system from Japanese texts to Japanese Sign Language (JSL) CG animations for weather news. To overcome some problems of the initially-developed machine translation system, we improved the system by increasing our JSL corpus size, developing our new translation method, introducing clause-based and phrase-based CG animations, producing a new CG avatar, adding facial expressions to the avatar and synthesizing mouth patterns. We conducted a series of experiments for the improved system and obtained an accuracy increase from 3.4 to 4.0 in 5 grade evaluation. Furthermore, we found the improved system has less number of claims pointed by the evaluators than our initially-developed system. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Sign Language / Japanese / Machine Translation / Corpus / Computer Graphics / Weather News / / |
Reference Info. |
IEICE Tech. Rep., vol. 117, no. 188, WIT2017-19, pp. 27-32, Aug. 2017. |
Paper # |
WIT2017-19 |
Date of Issue |
2017-08-21 (WIT) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
WIT2017-19 |
|