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 2008-03-20 15:15
[Poster Presentation] Unsupervised Phoneme Segmentation Using Mahalanobis Distance
Yu Qiao, Nobuaki Minematsu (Univ. of Tokyo) SP2007-198
Abstract (in Japanese) (See Japanese page) 
(in English) One of the fundamental problems in speech engineering is phoneme segmentation. Approaches to phoneme segmentation can be divided into two categories: supervised and unsupervised segmentation. The approach of this paper belongs to the 2nd category: that is we try to perform phonetic segmentation without using any prior knowledge on linguistic contents and acoustic models. In an earlier work, we have formulated the segmentation problem into a probabilistic optimization problem by using statistics and information theory analysis. We developed an objective function: summation of square error (SSE) based on Euclidean distance of cepstrual features. However, it is not known whether or not Euclidean distance yields the best distance metric to estimate the goodness of the segments. A popular generalization of Euclidean distance is Mahalanobis distance. In this paper, we study whether and how Mahalanobis distance can be used to improve the performance of segmentation. The essential problem here is how to determine the parameters (covariance matrix) for Mahalanobis distance calculation. We deal with this problem in a learning based framework and develop two criteria for determining the optimal parameters: MSV and MDV. MSV minimizes the summation of variance within-phoneme, and MDV tries to maximize the ratio of the variance between phonemes to the variance within phonemes. Both of them can lead to close form solutions by using matrix calculation. We carried out experiments on the TIMIT database to compare the proposed methods. The results indicate that the using of learning Mahalanobis distance can improve the segmentation performance.
Keyword (in Japanese) (See Japanese page) 
(in English) Unsupervised phoneme segmentation / Optimization / Mahalanobis distance / Learning distance metric / / / /  
Reference Info. IEICE Tech. Rep., vol. 107, no. 551, SP2007-198, pp. 69-74, March 2008.
Paper # SP2007-198 
Date of Issue 2008-03-13 (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 SP2007-198

Conference Information
Committee SP  
Conference Date 2008-03-20 - 2008-03-21 
Place (in Japanese) (See Japanese page) 
Place (in English) Univ. Tokyo 
Topics (in Japanese) (See Japanese page) 
Topics (in English) International Workshop (Mar 20), Speech Production, Speech Perception, Hearing and Speech, etc. (Mar 21) 
Paper Information
Registration To SP 
Conference Code 2008-03-SP 
Language English 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Unsupervised Phoneme Segmentation Using Mahalanobis Distance 
Sub Title (in English)  
Keyword(1) Unsupervised phoneme segmentation  
Keyword(2) Optimization  
Keyword(3) Mahalanobis distance  
Keyword(4) Learning distance metric  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Yu Qiao  
1st Author's Affiliation the University of Tokyo (Univ. of Tokyo)
2nd Author's Name Nobuaki Minematsu  
2nd Author's Affiliation the University of Tokyo (Univ. of Tokyo)
3rd Author's Name  
3rd Author's Affiliation ()
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 ()
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 2008-03-20 15:15:00 
Presentation Time 90 minutes 
Registration for SP 
Paper # SP2007-198 
Volume (vol) vol.107 
Number (no) no.551 
Page pp.69-74 
#Pages 6 
Date of Issue 2008-03-13 (SP) 

[Return to Top Page]

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