Information: Join today and make your research activities more affordable! Technical workshop participation fees and annual registration fees are available at member rates.
Notice: [Important] Announcement of Changes to Registration Fee Payment and Manuscript Upload Procedures for IEICE Technical Meetings
IEICE Technical Committee Submission System
Conference Paper's Information		 Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2014-12-16 11:00
Speaker adaptation using speaker-normalized DNN based on speaker codes
Yosuke Kashiwagi, Daisuke Saito, Nobuaki Minematsu, Keikichi Hirose (Univ. of Tokyo) SP2014-118
Abstract (in Japanese) (See Japanese page) 
(in English) Recently, deep neural network (DNN) becomes one of the main streams of acoustic modeling for automatic speech recognition. Further, speaker adaptation techniques have been tested for DNN-based speech recognition, including one based on a framework of bias adaptation using speaker codes. This paper introduces speaker-normalized training to this framework and experimentally shows its effectiveness. In the conventional method using speaker codes, two kinds of networks of speaker-independent (SI) DNNs and subnetworks for speaker adaptation were trained sequentially. We expect that, by training the SI networks and the subnetworks simultaneously, this method can be tuned so that it can handle both SI information and speaker-dependent (SD) information more adequately. Further, different from the conventional method, the speaker code vector is generated through networks from a 1-of-$N$ speaker representation. This will reduce the training cost of the SI models and the subnetworks and avoid the over-fitting problem. Experimental evaluations using the TIMIT database demonstrate that our proposed training method can reduce the phoneme error rate by 5.7% relative.
Keyword (in Japanese) (See Japanese page) 
(in English) automatic speech recognition / acoutic model / speaker adaptation / speaker normalized training / deep neural network / / /  
Reference Info. IEICE Tech. Rep., vol. 114, no. 365, SP2014-118, pp. 105-110, Dec. 2014.
Paper # SP2014-118 
Date of Issue 2014-12-08 (SP) 
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 SP2014-118

Conference Information
Committee NLC IPSJ-NL SP IPSJ-SLP JSAI-SLUD  
Conference Date 2014-12-15 - 2014-12-17 
Place (in Japanese) (See Japanese page) 
Place (in English) Tokyo Institute of Technology (Suzukakedai Campus) 
Topics (in Japanese) (See Japanese page) 
Topics (in English) The 6th Symposium on Collective Knowlege 
Paper Information
Registration To SP 
Conference Code 2014-12-NLC-NL-SP-SLP-SLUD 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Speaker adaptation using speaker-normalized DNN based on speaker codes 
Sub Title (in English)  
Keyword(1) automatic speech recognition  
Keyword(2) acoutic model  
Keyword(3) speaker adaptation  
Keyword(4) speaker normalized training  
Keyword(5) deep neural network  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Yosuke Kashiwagi  
1st Author's Affiliation The University of Tokyo (Univ. of Tokyo)
2nd Author's Name Daisuke Saito  
2nd Author's Affiliation The University of Tokyo (Univ. of Tokyo)
3rd Author's Name Nobuaki Minematsu  
3rd Author's Affiliation The University of Tokyo (Univ. of Tokyo)
4th Author's Name Keikichi Hirose  
4th Author's Affiliation The University of Tokyo (Univ. of Tokyo)
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 ()
21st Author's Name  
21st Author's Affiliation ()
22nd Author's Name  
22nd Author's Affiliation ()
23rd Author's Name  
23rd Author's Affiliation ()
24th Author's Name  
24th Author's Affiliation ()
25th Author's Name  
25th Author's Affiliation ()
26th Author's Name / /
26th Author's Affiliation ()
()
27th Author's Name / /
27th Author's Affiliation ()
()
28th Author's Name / /
28th Author's Affiliation ()
()
29th Author's Name / /
29th Author's Affiliation ()
()
30th Author's Name / /
30th Author's Affiliation ()
()
31st Author's Name / /
31st Author's Affiliation ()
()
32nd Author's Name / /
32nd Author's Affiliation ()
()
33rd Author's Name / /
33rd Author's Affiliation ()
()
34th Author's Name / /
34th Author's Affiliation ()
()
35th Author's Name / /
35th Author's Affiliation ()
()
36th Author's Name / /
36th Author's Affiliation ()
()
Speaker Author-1 
Date Time 2014-12-16 11:00:00 
Presentation Time 90 minutes 
Registration for SP 
Paper # SP2014-118 
Volume (vol) vol.114 
Number (no) no.365 
Page pp.105-110 
#Pages 6 
Date of Issue 2014-12-08 (SP) 

[Return to Top Page]

[Return to IEICE Web Page]
The Institute of Electronics, Information and Communication Engineers (IEICE), Japan