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 2020-03-03 09:00
[Poster Presentation] A Comparison of Language Models for a Design of Reduced Phoneme Set
Shuji Komeiji, Toshihisa Tanaka (TUAT), Koichi Shinoda (titech) EA2019-152 SIP2019-154 SP2019-101
Abstract (in Japanese) (See Japanese page) 
(in English) Language models for a design of reduced phoneme set are compared each other.
The reduction of the phoneme set improves discriminability of phonemes under the condition
where the amount of training data is too small to train each phoneme model.
On the other hand, it increases homophones that yield degradation of speech recognition.
In the proposed approach, it is possible to reduce phonemes preventing degradation,
regarding pronunciation/word sequence confusion rate (PWCR) calculated from an $n$-gram language model.
Previously, we have evaluated the reduced phoneme set designed with trigram.
In this paper, we clarify whether unigram and bigram, which have less $n$-grams than trigram, can be
used to design reduced phoneme sets that prevent degradation of accuracy.
Accoding to the evaluation for unigram, inspite of about 1/10 calculation cost,
the difference of Word Error Rates (WERs) is within 0.2%
among these three language models.
Keyword (in Japanese) (See Japanese page) 
(in English) Automatic speech recognition / Phoneme set / Language model / n-gram / / / /  
Reference Info. IEICE Tech. Rep., vol. 119, no. 440, SIP2019-154, pp. 295-300, March 2020.
Paper # SIP2019-154 
Date of Issue 2020-02-24 (EA, SIP, 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 EA2019-152 SIP2019-154 SP2019-101

Conference Information
Committee SP EA SIP  
Conference Date 2020-03-02 - 2020-03-03 
Place (in Japanese) (See Japanese page) 
Place (in English) Okinawa Industry Support Center 
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To SIP 
Conference Code 2020-03-SP-EA-SIP 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) A Comparison of Language Models for a Design of Reduced Phoneme Set 
Sub Title (in English)  
Keyword(1) Automatic speech recognition  
Keyword(2) Phoneme set  
Keyword(3) Language model  
Keyword(4) n-gram  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Shuji Komeiji  
1st Author's Affiliation Tokyo University of Agriculture and Technology (TUAT)
2nd Author's Name Toshihisa Tanaka  
2nd Author's Affiliation Tokyo University of Agriculture and Technology (TUAT)
3rd Author's Name Koichi Shinoda  
3rd Author's Affiliation Tokyo Institute of Technology (titech)
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 2020-03-03 09:00:00 
Presentation Time 90 minutes 
Registration for SIP 
Paper # EA2019-152, SIP2019-154, SP2019-101 
Volume (vol) vol.119 
Number (no) no.439(EA), no.440(SIP), no.441(SP) 
Page pp.295-300 
#Pages
Date of Issue 2020-02-24 (EA, SIP, SP) 


[Return to Top Page]

[Return to IEICE Web Page]


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