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 2019-12-13 13:00
Speaker-independent source separation with multichannel variational autoencoder
Li Li (Univ. Tsukuba), Hirokazu Kameoka (NTT), Shota Inoue, Shoji Makino (Univ. Tsukuba) EA2019-77
Abstract (in Japanese) (See Japanese page) 
(in English) The multichannel variational autoencoder method (MVAE) is a recently proposed determined source separation method, which uses a conditional variational autoencoder (CVAE) to learn the spectrograms of source signals given a source-class ID as an auxiliary input. The trained decoder distribution can be used as a universal generative model capable of generating spectrograms of all the sources involved in the training samples. The decoder distribution can then be exploited to estimate the spectrograms of sources in a mixture. The MVAE methods, including the original MVAE method and its fast version called FastMVAE, were shown to significantly outperform conventional methods under speaker-dependent conditions, where the target speakers are seen in the training dataset. In this paper, we investigate the performances of the two MVAE methods under speaker-independent conditions. To further enhance the ability of FastMVAE to estimate the latent space variables for unknown speakers, we propose a prior-aware inference algorithm based on the concept of product-of-experts. Experimental results revealed that the MVAE methods could perform well even under speaker-independent conditions.
Keyword (in Japanese) (See Japanese page) 
(in English) Multichannel source separation / determined source separation / multichannel variational autoencoder (MVAE), / FastMVAE / conditional variational autoencoder (CVAE) / / /  
Reference Info. IEICE Tech. Rep., vol. 119, no. 334, EA2019-77, pp. 79-84, Dec. 2019.
Paper # EA2019-77 
Date of Issue 2019-12-05 (EA) 
ISSN Print edition: ISSN 0913-5685  Online edition: ISSN 2432-6380
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 EA2019-77

Conference Information
Committee EA  
Conference Date 2019-12-12 - 2019-12-13 
Place (in Japanese) (See Japanese page) 
Place (in English) Kyushu Inst. Tech. 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Engineering/Electro Acoustics, and Related Topics 
Paper Information
Registration To EA 
Conference Code 2019-12-EA 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Speaker-independent source separation with multichannel variational autoencoder 
Sub Title (in English)  
Keyword(1) Multichannel source separation  
Keyword(2) determined source separation  
Keyword(3) multichannel variational autoencoder (MVAE),  
Keyword(4) FastMVAE  
Keyword(5) conditional variational autoencoder (CVAE)  
1st Author's Name Li Li  
1st Author's Affiliation University of Tsukuba (Univ. Tsukuba)
2nd Author's Name Hirokazu Kameoka  
2nd Author's Affiliation NTT (NTT)
3rd Author's Name Shota Inoue  
3rd Author's Affiliation University of Tsukuba (Univ. Tsukuba)
4th Author's Name Shoji Makino  
4th Author's Affiliation University of Tsukuba (Univ. Tsukuba)
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 ()
Date Time 2019-12-13 13:00:00 
Presentation Time 25 
Registration for EA 
Paper # EA2019-77 
Volume (vol) 119 
Number (no) no.334 
Page pp.79-84 
Date of Issue 2019-12-05 (EA) 

[Return to Top Page]

[Return to IEICE Web Page]

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