Paper Abstract and Keywords |
Presentation |
2014-10-23 15:10
Automatic Speaker Estimation System for Long-term Analysis of Conversation with Nonverbal Information in Unconstrained Environment Masano Nakayama, Hiroaki Murata, Hitomi Yokoyama, Seiji Hotta, Kinya Fujita (TUAT) HCS2014-64 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We are developing group observation system under the unconstrained environments. In this paper, we proposed a method of speaker estimation, which is a part of the system. The proposed method includes human detection, sound position detection, noise removal and speaking pause interpolation. The result of human detection and sound position are combined as potential data to detect speakers. Comparing results of estimation to that of cording, the accuracy achieves 0.91 by precision and 0.86 by recall in the best case. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Speaker Estimation / Unconstrained Environments / Long-term Continuous Automatic Estimation System / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 114, no. 273, HCS2014-64, pp. 25-30, Oct. 2014. |
Paper # |
HCS2014-64 |
Date of Issue |
2014-10-16 (HCS) |
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 |
HCS2014-64 |
|