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 2013-07-25 17:30
Voice conversion using piecewise linear transformation with long-span features and state discrimination of the target speaker
Jun Ikeshima, Masayuki Suzuki, Daisuke Saito, Nobuaki Minematsu, Keikichi Hirose (Univ. of Tokyo) SP2013-56
Abstract (in Japanese) (See Japanese page) 
(in English) A method has been proposed for voice conversion. It is based on piecewise linear transformation with long time-span features, and state discrimination of the target speaker. The piecewise linear transformation consists of two methods in order to realize nonlinear transformation between source and target speakers as a set of linear transformations; dividing acoustic feature space into several regions and approximating a linear transformation for each region. This idea has been implemented as ``soft'' space division using GMM (Gaussian Mixture Model) and successfully applied to voice conversion. In the original/conventional method, space division is conducted using feature vectors of source speaker or joint feature vectors of source and target speakers. However, resulted division is not in good much with feature space of target speaker, when there are large differences between distributions of feature spaces of source and target speakers. In order to solve this situation, a new method of space division by features of target speaker is developed. Since features of target speaker are not available during conversion, ”affiliation probability” is estimated using discriminative modeling scheme. Also features of time-adjacent frames are added to the feature vectors (of discrimination and transformation processes), corresponding to the fact that the long term features are important for voice conversion similar to the case of speech recognition. The proposed method was shown to be effective for voice conversion through objective and subjective evaluations.
Keyword (in Japanese) (See Japanese page) 
(in English) voice conversion / piecewise linear transformation / discrimination model / long span feature / / / /  
Reference Info. IEICE Tech. Rep., vol. 113, no. 161, SP2013-56, pp. 19-24, July 2013.
Paper # SP2013-56 
Date of Issue 2013-07-18 (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 SP2013-56

Conference Information
Committee SP IPSJ-SLP  
Conference Date 2013-07-25 - 2013-07-27 
Place (in Japanese) (See Japanese page) 
Place (in English) Soho (togatta spa) 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Speech recognition and understanding, dialog system, etc. 
Paper Information
Registration To SP 
Conference Code 2013-07-SP-SLP 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Voice conversion using piecewise linear transformation with long-span features and state discrimination of the target speaker 
Sub Title (in English)  
Keyword(1) voice conversion  
Keyword(2) piecewise linear transformation  
Keyword(3) discrimination model  
Keyword(4) long span feature  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Jun Ikeshima  
1st Author's Affiliation The University of Tokyo (Univ. of Tokyo)
2nd Author's Name Masayuki Suzuki  
2nd Author's Affiliation The University of Tokyo (Univ. of Tokyo)
3rd Author's Name Daisuke Saito  
3rd Author's Affiliation The University of Tokyo (Univ. of Tokyo)
4th Author's Name Nobuaki Minematsu  
4th Author's Affiliation The University of Tokyo (Univ. of Tokyo)
5th Author's Name Keikichi Hirose  
5th Author's Affiliation The University of Tokyo (Univ. of Tokyo)
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 2013-07-25 17:30:00 
Presentation Time 30 minutes 
Registration for SP 
Paper # SP2013-56 
Volume (vol) vol.113 
Number (no) no.161 
Page pp.19-24 
#Pages
Date of Issue 2013-07-18 (SP) 


[Return to Top Page]

[Return to IEICE Web Page]


The Institute of Electronics, Information and Communication Engineers (IEICE), Japan