Paper Abstract and Keywords |
Presentation |
2023-03-01 10:10
Vocabulary-Set Decomposition and Multi-task Learning for Target Vocabulary Extraction in Japanese Speech Recognition Aoi Ito (LINE/Hosei Univ.), Tatsuya Komatsu, Yusuke Fujita (LINE) EA2022-102 SIP2022-146 SP2022-66 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
This paper proposes a target vocabulary extraction method for Japanese speech recognition models based on vocabulary set decomposition and multi-task learning. In the proposed method, when a certain vocabulary set, e.g., a series consisting only of katakana, is to be extracted, multi-task learning is performed by calculating the losses for the target katakana sequence and the non-katakana sequence. By combining the losses for the complementary sequence, the target vocabulary can be extracted with higher accuracy than in single-task learning using only the target vocabulary. Experiments show that the proposed method can extract the target vocabulary more accurately than the target vocabulary filtering and the single-task learning method using only the target vocabulary. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Automatic speech recognition / CTC / Self-Conditioned CTC / target vocabulary extraction / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 122, no. 389, SP2022-66, pp. 159-164, Feb. 2023. |
Paper # |
SP2022-66 |
Date of Issue |
2023-02-21 (EA, SIP, 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) |
Download PDF |
EA2022-102 SIP2022-146 SP2022-66 |
Conference Information |
Committee |
SP IPSJ-SLP EA SIP |
Conference Date |
2023-02-28 - 2023-03-01 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
|
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
|
Paper Information |
Registration To |
SP |
Conference Code |
2023-02-SP-SLP-EA-SIP |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
Vocabulary-Set Decomposition and Multi-task Learning for Target Vocabulary Extraction in Japanese Speech Recognition |
Sub Title (in English) |
|
Keyword(1) |
Automatic speech recognition |
Keyword(2) |
CTC |
Keyword(3) |
Self-Conditioned CTC |
Keyword(4) |
target vocabulary extraction |
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Aoi Ito |
1st Author's Affiliation |
LINE Corporation/Hosei University (LINE/Hosei Univ.) |
2nd Author's Name |
Tatsuya Komatsu |
2nd Author's Affiliation |
LINE Corporation (LINE) |
3rd Author's Name |
Yusuke Fujita |
3rd Author's Affiliation |
LINE Corporation (LINE) |
4th Author's Name |
|
4th Author's Affiliation |
() |
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 |
2023-03-01 10:10:00 |
Presentation Time |
20 minutes |
Registration for |
SP |
Paper # |
EA2022-102, SIP2022-146, SP2022-66 |
Volume (vol) |
vol.122 |
Number (no) |
no.387(EA), no.388(SIP), no.389(SP) |
Page |
pp.159-164 |
#Pages |
6 |
Date of Issue |
2023-02-21 (EA, SIP, SP) |
|