An Enhancement of Japanese Acoustic Model using Korean Speech Database

Minkyu† Lee; Sanghun Kim

doi:10.7776/ASK.2013.32.5.438

All Issue

2013 Vol.32, Issue 5 Preview Page Next Page

An Enhancement of Japanese Acoustic Model using Korean Speech Database 한국어 음성데이터를 이용한 일본어 음향모델 성능 개선

30 September 2013. pp. 438-445

PDF XML

Abstract

In this paper, we propose an enhancement of Japanese acoustic model which is trained with Korean speech database by using several combination strategies. We describe the strategies for training more than two language combination, which are Cross-Language Transfer, Cross-Language Adaptation, and Data Pooling Approach. We simulated those strategies and found a proper method for our current Japanese database. Existing combination strategies are generally verified for under-resourced Language environments, but when the speech database is not fully under-resourced, those strategies have been confirmed inappropriate. We made tyied-list with only object-language on Data Pooling Approach training process. As the result, we found the ERR of the acoustic model to be 12.8 %.

Keywords

Speech recognition

Japanese speech recognition

Automatic speech translation

Language adaptation

Data pooling

Under-resourced language

본 논문은 일본어 음성인식기 신규 개발을 위해 초기에 부족한 일본어 음성데이터를 보완하는 방법이다. 일본어 발음과 한국어 발음이 유사한 특성을 근거로 한국어 음성 데이터를 이용한 일본어 음향모델 성능개선 방법에 대하여 기술하였다. 이종언어 간 음성 데이터를 섞어서 훈련하는 방법인 Cross-Language Transfer, Cross-Language Adaptation, Data Pooling Approach등 방법을 설명하고, 각 방법들의 시뮬레이션을 통해 현재 보유하고 있는 일본어 음성데이터 양에 적절한 방법을 선정하였다. 기존의 방법들은 훈련용 음성데이터가 크게 부족한 환경에서의 효과는 검증되었으나, 목적 언어의 데이터가 어느 정도 확보된 상태에서는 성능 개선 효과가 미비하였다. 그러나 Data Pooling Approach의 훈련과정 중 Tyied-List를 목적 언어로만으로 구성 하였을 때, ERR(Error Reduction Rate)이 12.8 %로 성능이 향상됨을 확인하였다.

키워드

음성인식

일본어 인식

자동통역

언어적응

Data Pooling

Under-resourced Language

References

1.Ulla Uebler, “Multilingual speech recognition in seven languages,” Speech. Commun. 35, 53-69 (2001).

2.C. van Heerden, N. Kleynhans, E. Barnard, and M. Davel, “Pooling ASR data for closely reslated languages,” in Proc. SLTU, 17-23 (2010).

3.Tanja Schultz and Alex Waibel, “Language Portability in Acoustic Modeling,” Speech. Commun. 10, 59-64 (2000).

4.Kenan Çarki, Petra Geutner, and Tanja Schultz, “Turkish LVCSR: toward better speech recognition for agglutinative languages,” In Proc. ICASSP, 1563-1566 (2000).

5.J. K. Lee, “Comparative study on korean and japanese formant values by korean speakers and japanese speakers,” Ono Yongu Studies in Linguistics. 15, 61-75 (1997)

6.Mecab: Yet Another Part-of-Speech and Morphological Analyzer, http://mecab.googlecode.com/svn/trunk/mecab/ doc/index.html, 2013.

7.The SRI Language Modeling Toolkit, http://www.speech. sri.com/projects/srilm, 2013.

8.S. H. Kim, I. Lee, and J. Park, “Developing fast/light korean recognizer through building FST_based search network,” in Proc. KSCSP, 25, 1, 21-24 (2008).

9.A. Constantinescu, and G. Chollet, “On cross-language experiments and data-driven units for ALISP,” In Proc. ASRU, 606-613 (1997).

Information

Publisher :The Acoustical Society of Korea
Publisher(Ko) :한국음향학회
Journal Title :The Journal of the Acoustical Society of Korea
Journal Title(Ko) :한국음향학회지
Volume : 32
No :5
Pages :438-445
Received Date : 2013-05-08
Accepted Date : 2013-07-05
DOI :https://doi.org/10.7776/ASK.2013.32.5.438

The Journal of the Acoustical Society of KoreaISSN:1225-4428(Print) 2287-3775(Online)한국음향학회

All Issue