Paper Abstract and Keywords |
Presentation |
2019-12-06 13:55
[Poster Presentation]
Analysis and Subjective Labeling for Emotional Speech Database JTES Mai Yamanaka, Takashi Nose, Yuya Chiba, Akinori Ito (Tohoku Univ.) SP2019-39 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We have constructed JTES, a prosodic balanced emotional speech database containing 50 sentences of 4 emotions of 50 men and women, totaling 20,000 utterances.
However, JTES had variations in the emotional intensity and expression, which could adversely affect emotion recognition and speech synthesis.
In this paper, in order to investigate the variation of speaker's emotional expression for improving the accuracy of speech emotion recognition, the subjective label (listener label) is given by the listener and the result is analyzed.
The subjective labels matched the emotional label of the speaker with an accuracy of 70% but some voices that could not be correctly classified were distinguished into neutral.
There is the possibility that these speeches affect emotion recognition.
Crowdsourcing was used to assign large-scale subjective labels. In order to request many people to works, we introduced some speeches to easily check the reliability.However it was not possible to use it as an index for reliability.
About 70% of the match rates between speaker labels and subjective labels for each labeler were 70%, but there was also one with less than 50%.
The labeling bias of each labeler was vectorized and compared with the average of all labeler's vectors, and 60% of the labelers were distributed at a value of about 0.5 away from the average. Since then, the number has decreased to three.
This indicates that there are individual differences in labelers.
When the emotional labels assigned to each labeler were vectorized and compared with the average of all labeler vectors, the distance was found to be distributed in the form of a normal distribution from 0.2 to 0.9, and it was clear that there were large individual differences depending on the labeler.
This indicates that there are individual differences in labelers. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
emotional label / subjective label / crowdsourcing / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 119, no. 321, SP2019-39, pp. 61-66, Dec. 2019. |
Paper # |
SP2019-39 |
Date of Issue |
2019-11-29 (SP) |
ISSN |
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) |
Notes on Review |
This article is a technical report without peer review, and its polished version will be published elsewhere. |
Download PDF |
SP2019-39 |
Conference Information |
Committee |
NLC IPSJ-NL SP IPSJ-SLP |
Conference Date |
2019-12-04 - 2019-12-06 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
NHK Science & Technology Research Labs. |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
The 6th Natural Language Processing Symposium & The 21th Spoken Language Symposium |
Paper Information |
Registration To |
SP |
Conference Code |
2019-12-NLC-NL-SP-SLP |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
Analysis and Subjective Labeling for Emotional Speech Database JTES |
Sub Title (in English) |
|
Keyword(1) |
emotional label |
Keyword(2) |
subjective label |
Keyword(3) |
crowdsourcing |
Keyword(4) |
|
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Mai Yamanaka |
1st Author's Affiliation |
Tohoku University (Tohoku Univ.) |
2nd Author's Name |
Takashi Nose |
2nd Author's Affiliation |
Tohoku University (Tohoku Univ.) |
3rd Author's Name |
Yuya Chiba |
3rd Author's Affiliation |
Tohoku University (Tohoku Univ.) |
4th Author's Name |
Akinori Ito |
4th Author's Affiliation |
Tohoku University (Tohoku Univ.) |
5th Author's Name |
|
5th Author's Affiliation |
() |
6th Author's Name |
|
6th Author's Affiliation |
() |
7th Author's Name |
|
7th Author's Affiliation |
() |
8th Author's Name |
|
8th Author's Affiliation |
() |
9th Author's Name |
|
9th Author's Affiliation |
() |
10th Author's Name |
|
10th Author's Affiliation |
() |
11th Author's Name |
|
11th Author's Affiliation |
() |
12th Author's Name |
|
12th Author's Affiliation |
() |
13th Author's Name |
|
13th Author's Affiliation |
() |
14th Author's Name |
|
14th Author's Affiliation |
() |
15th Author's Name |
|
15th Author's Affiliation |
() |
16th Author's Name |
|
16th Author's Affiliation |
() |
17th Author's Name |
|
17th Author's Affiliation |
() |
18th Author's Name |
|
18th Author's Affiliation |
() |
19th Author's Name |
|
19th Author's Affiliation |
() |
20th Author's Name |
|
20th Author's Affiliation |
() |
Speaker |
Author-1 |
Date Time |
2019-12-06 13:55:00 |
Presentation Time |
90 minutes |
Registration for |
SP |
Paper # |
SP2019-39 |
Volume (vol) |
vol.119 |
Number (no) |
no.321 |
Page |
pp.61-66 |
#Pages |
6 |
Date of Issue |
2019-11-29 (SP) |
|