Phoneme boundary estimation using bidirectional recurrent neural networks and its applications

Toshiaki Fukada, Mike Schuster, Yoshinori Sagisaka

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

This paper describes a phoneme boundary estimation method based on bidirectional recurrent neural networks (BRNNs). Experimental results showed that the proposed method could estimate segment boundaries significantly better than an HMM or a multilayer perceptron-based method. Furthermore, we incorporated the BRNN-based segment boundary estimator into the HMM-based and segment model-based recognition systems. As a result, we confirmed that (1) BRNN outputs were effective for improving the recognition rate and reducing computational time in an HMM-based recognition system and (2) segment lattices obtained by the proposed methods dramatically reduce the computational complexity of segment model-based recognition.

Original languageEnglish
Pages (from-to)20-30
Number of pages11
JournalSystems and Computers in Japan
Volume30
Issue number4
DOIs
Publication statusPublished - 1999 Apr
Externally publishedYes

Fingerprint

Recurrent neural networks
Recurrent Neural Networks
Multilayer neural networks
Computational complexity
Model-based
Perceptron
Multilayer
Computational Complexity
Estimator
Output
Experimental Results
Estimate

ASJC Scopus subject areas

  • Hardware and Architecture
  • Information Systems
  • Theoretical Computer Science
  • Computational Theory and Mathematics

Cite this

Phoneme boundary estimation using bidirectional recurrent neural networks and its applications. / Fukada, Toshiaki; Schuster, Mike; Sagisaka, Yoshinori.

In: Systems and Computers in Japan, Vol. 30, No. 4, 04.1999, p. 20-30.

Research output: Contribution to journalArticle

@article{9f7af245b90744188dfe39cab1343592,
title = "Phoneme boundary estimation using bidirectional recurrent neural networks and its applications",
abstract = "This paper describes a phoneme boundary estimation method based on bidirectional recurrent neural networks (BRNNs). Experimental results showed that the proposed method could estimate segment boundaries significantly better than an HMM or a multilayer perceptron-based method. Furthermore, we incorporated the BRNN-based segment boundary estimator into the HMM-based and segment model-based recognition systems. As a result, we confirmed that (1) BRNN outputs were effective for improving the recognition rate and reducing computational time in an HMM-based recognition system and (2) segment lattices obtained by the proposed methods dramatically reduce the computational complexity of segment model-based recognition.",
author = "Toshiaki Fukada and Mike Schuster and Yoshinori Sagisaka",
year = "1999",
month = "4",
doi = "10.1002/(SICI)1520-684X(199904)30:4<20::AID-SCJ3>3.0.CO;2-E",
language = "English",
volume = "30",
pages = "20--30",
journal = "Systems and Computers in Japan",
issn = "0882-1666",
publisher = "John Wiley and Sons Inc.",
number = "4",

}

TY - JOUR

T1 - Phoneme boundary estimation using bidirectional recurrent neural networks and its applications

AU - Fukada, Toshiaki

AU - Schuster, Mike

AU - Sagisaka, Yoshinori

PY - 1999/4

Y1 - 1999/4

N2 - This paper describes a phoneme boundary estimation method based on bidirectional recurrent neural networks (BRNNs). Experimental results showed that the proposed method could estimate segment boundaries significantly better than an HMM or a multilayer perceptron-based method. Furthermore, we incorporated the BRNN-based segment boundary estimator into the HMM-based and segment model-based recognition systems. As a result, we confirmed that (1) BRNN outputs were effective for improving the recognition rate and reducing computational time in an HMM-based recognition system and (2) segment lattices obtained by the proposed methods dramatically reduce the computational complexity of segment model-based recognition.

AB - This paper describes a phoneme boundary estimation method based on bidirectional recurrent neural networks (BRNNs). Experimental results showed that the proposed method could estimate segment boundaries significantly better than an HMM or a multilayer perceptron-based method. Furthermore, we incorporated the BRNN-based segment boundary estimator into the HMM-based and segment model-based recognition systems. As a result, we confirmed that (1) BRNN outputs were effective for improving the recognition rate and reducing computational time in an HMM-based recognition system and (2) segment lattices obtained by the proposed methods dramatically reduce the computational complexity of segment model-based recognition.

UR - http://www.scopus.com/inward/record.url?scp=0033116017&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033116017&partnerID=8YFLogxK

U2 - 10.1002/(SICI)1520-684X(199904)30:4<20::AID-SCJ3>3.0.CO;2-E

DO - 10.1002/(SICI)1520-684X(199904)30:4<20::AID-SCJ3>3.0.CO;2-E

M3 - Article

VL - 30

SP - 20

EP - 30

JO - Systems and Computers in Japan

JF - Systems and Computers in Japan

SN - 0882-1666

IS - 4

ER -