Paper Abstract and Keywords |
Presentation |
2022-03-02 11:35
Study of Method for Improving Speech Intelligibility in Glossectomy Patients by Knowledge Distillation via Lip Features Kazushi Takashima, Masanobu Abe, Sunao Hara (Okayama Univ.) EA2021-81 SIP2021-108 SP2021-66 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In this paper, we propose a voice conversion method for improving speech intelligibility uttered by glossectomy patients. Because the glossectomy patients remove more than half of their tongue, their speech is less intelligibility compared to healthy persons. In a previous study, a voice conversion method using phoneme labels as auxiliary information in addition to audio information was proposed and it has been shown that the speech intelligibility is greatly improved. However, in actual situations, it is difficult to prepare phoneme labels corresponding to the speech content at the time of conversion. To solve the problem, by introducing knowledge distillation approach, we proposed to train a student model that synthesizes speech without labels, where a model trained using phoneme labels is used as teacher model. Although the performance of the method with knowledge distillation is better than that of the method without knowledge distillation, intelligibility was not imporved enough. Hence, we investigate a method that uses lip information as an additional feature in order to improve the performance of the student model. we extract bottleneck features from lip information and use them as additional input features by using a convolutional autoencoder. In the evaluation experiments, we evaluated the performance of the convolutional autoencoder and the conversion accuracy of each voice conversion method. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Voice Conversion / Glossectomy Patients / Deep Neural Network / CNN-Autoencoder / Knowledge Distillation / / / |
Reference Info. |
IEICE Tech. Rep., vol. 121, no. 385, SP2021-66, pp. 108-113, March 2022. |
Paper # |
SP2021-66 |
Date of Issue |
2022-02-22 (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 |
EA2021-81 SIP2021-108 SP2021-66 |
Conference Information |
Committee |
EA SIP SP IPSJ-SLP |
Conference Date |
2022-03-01 - 2022-03-02 |
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 |
2022-03-EA-SIP-SP-SLP |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
Study of Method for Improving Speech Intelligibility in Glossectomy Patients by Knowledge Distillation via Lip Features |
Sub Title (in English) |
|
Keyword(1) |
Voice Conversion |
Keyword(2) |
Glossectomy Patients |
Keyword(3) |
Deep Neural Network |
Keyword(4) |
CNN-Autoencoder |
Keyword(5) |
Knowledge Distillation |
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Kazushi Takashima |
1st Author's Affiliation |
Okayama University (Okayama Univ.) |
2nd Author's Name |
Masanobu Abe |
2nd Author's Affiliation |
Okayama University (Okayama Univ.) |
3rd Author's Name |
Sunao Hara |
3rd Author's Affiliation |
Okayama University (Okayama Univ.) |
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 |
2022-03-02 11:35:00 |
Presentation Time |
25 minutes |
Registration for |
SP |
Paper # |
EA2021-81, SIP2021-108, SP2021-66 |
Volume (vol) |
vol.121 |
Number (no) |
no.383(EA), no.384(SIP), no.385(SP) |
Page |
pp.108-113 |
#Pages |
6 |
Date of Issue |
2022-02-22 (EA, SIP, SP) |
|