Refinement of landmark detection and extraction of articulator-free features for knowledge-based speech recognition

Jung In Lee, Jeung Yoon Choi, Hong Goo Kang

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Refinement methods for landmark detection and extraction of articulator-free features for a knowledge-based speech recognition system are described. Sub-band energy difference profiles are used to detect landmarks, with additional parameters used to improve accuracy. For articulator-free feature extraction, duration, relative energy, and silence detection are additionally used to find [continuant] and [strident] features. Vowel, obstruent and sonorant consonant landmarks, and locations of voicing onsets and offsets are detected within a unified framework with 85% accuracy overall. Additionally, 75% and 79% of [continuant] and [strident] features, respectively, are detected from landmarks.

Original languageEnglish
Pages (from-to)746-749
Number of pages4
JournalIEICE Transactions on Information and Systems
VolumeE96-D
Issue number3
DOIs
Publication statusPublished - 2013 Mar

Fingerprint

Speech recognition
Band structure
Feature extraction

All Science Journal Classification (ASJC) codes

  • Software
  • Hardware and Architecture
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering
  • Artificial Intelligence

Cite this

@article{73e9fe1a54fc4c5f9a6690b7ce99861f,
title = "Refinement of landmark detection and extraction of articulator-free features for knowledge-based speech recognition",
abstract = "Refinement methods for landmark detection and extraction of articulator-free features for a knowledge-based speech recognition system are described. Sub-band energy difference profiles are used to detect landmarks, with additional parameters used to improve accuracy. For articulator-free feature extraction, duration, relative energy, and silence detection are additionally used to find [continuant] and [strident] features. Vowel, obstruent and sonorant consonant landmarks, and locations of voicing onsets and offsets are detected within a unified framework with 85{\%} accuracy overall. Additionally, 75{\%} and 79{\%} of [continuant] and [strident] features, respectively, are detected from landmarks.",
author = "Lee, {Jung In} and Choi, {Jeung Yoon} and Kang, {Hong Goo}",
year = "2013",
month = "3",
doi = "10.1587/transinf.E96.D.746",
language = "English",
volume = "E96-D",
pages = "746--749",
journal = "IEICE Transactions on Information and Systems",
issn = "0916-8532",
publisher = "Maruzen Co., Ltd/Maruzen Kabushikikaisha",
number = "3",

}

Refinement of landmark detection and extraction of articulator-free features for knowledge-based speech recognition. / Lee, Jung In; Choi, Jeung Yoon; Kang, Hong Goo.

In: IEICE Transactions on Information and Systems, Vol. E96-D, No. 3, 03.2013, p. 746-749.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Refinement of landmark detection and extraction of articulator-free features for knowledge-based speech recognition

AU - Lee, Jung In

AU - Choi, Jeung Yoon

AU - Kang, Hong Goo

PY - 2013/3

Y1 - 2013/3

N2 - Refinement methods for landmark detection and extraction of articulator-free features for a knowledge-based speech recognition system are described. Sub-band energy difference profiles are used to detect landmarks, with additional parameters used to improve accuracy. For articulator-free feature extraction, duration, relative energy, and silence detection are additionally used to find [continuant] and [strident] features. Vowel, obstruent and sonorant consonant landmarks, and locations of voicing onsets and offsets are detected within a unified framework with 85% accuracy overall. Additionally, 75% and 79% of [continuant] and [strident] features, respectively, are detected from landmarks.

AB - Refinement methods for landmark detection and extraction of articulator-free features for a knowledge-based speech recognition system are described. Sub-band energy difference profiles are used to detect landmarks, with additional parameters used to improve accuracy. For articulator-free feature extraction, duration, relative energy, and silence detection are additionally used to find [continuant] and [strident] features. Vowel, obstruent and sonorant consonant landmarks, and locations of voicing onsets and offsets are detected within a unified framework with 85% accuracy overall. Additionally, 75% and 79% of [continuant] and [strident] features, respectively, are detected from landmarks.

UR - http://www.scopus.com/inward/record.url?scp=84878248577&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84878248577&partnerID=8YFLogxK

U2 - 10.1587/transinf.E96.D.746

DO - 10.1587/transinf.E96.D.746

M3 - Article

AN - SCOPUS:84878248577

VL - E96-D

SP - 746

EP - 749

JO - IEICE Transactions on Information and Systems

JF - IEICE Transactions on Information and Systems

SN - 0916-8532

IS - 3

ER -