Spoken term detection based on DTW

Jingyong Hou; Lei Xie; Peng Yang; Xiong Xiao; Cheung Chi Leung; Haihua Xu; Lei Wang; Hang Lü; Bin Ma; Engsiong Chng; Haizhou Li

doi:10.16511/j.cnki.qhdxxb.2017.21.004

Spoken term detection based on DTW

Jingyong Hou, Lei Xie, Peng Yang, Xiong Xiao, Cheung Chi Leung, Haihua Xu, Lei Wang, Hang Lü, Bin Ma, Engsiong Chng, Haizhou Li

School of Computer Science

Research output: Contribution to journal › Article › peer-review

7 Scopus citations

Abstract

Spoken term detection (STD) for low resource languages has drawn much interest. A partial matching strategy based on phoneme boundaries is presented here to solve the fuzzy matching problem in query-by-example spoken term detection with dynamic time warping. A variety of features were used to validate the strategy on the QUESST 2014 dataset. Tests show that this strategy is not only quite effective for fuzzy match tasks T2 and T3 but also effective for the exact match task T1. This strategy has significantly improved performance in fusion tests.

Original language	English
Pages (from-to)	18-23
Number of pages	6
Journal	Qinghua Daxue Xuebao/Journal of Tsinghua University
Volume	57
Issue number	1
DOIs	https://doi.org/10.16511/j.cnki.qhdxxb.2017.21.004
State	Published - 1 Jan 2017

Keywords

Dynamic time warping
Low resource languages
Partial matching
Spoken term detection

Access to Document

10.16511/j.cnki.qhdxxb.2017.21.004

Cite this

@article{2e43a62f40794175a83f15dfabc555a8,

title = "Spoken term detection based on DTW",

abstract = "Spoken term detection (STD) for low resource languages has drawn much interest. A partial matching strategy based on phoneme boundaries is presented here to solve the fuzzy matching problem in query-by-example spoken term detection with dynamic time warping. A variety of features were used to validate the strategy on the QUESST 2014 dataset. Tests show that this strategy is not only quite effective for fuzzy match tasks T2 and T3 but also effective for the exact match task T1. This strategy has significantly improved performance in fusion tests.",

keywords = "Dynamic time warping, Low resource languages, Partial matching, Spoken term detection",

author = "Jingyong Hou and Lei Xie and Peng Yang and Xiong Xiao and Leung, {Cheung Chi} and Haihua Xu and Lei Wang and Hang L{\"u} and Bin Ma and Engsiong Chng and Haizhou Li",

year = "2017",

month = jan,

day = "1",

doi = "10.16511/j.cnki.qhdxxb.2017.21.004",

language = "英语",

volume = "57",

pages = "18--23",

journal = "Qinghua Daxue Xuebao/Journal of Tsinghua University",

issn = "1000-0054",

publisher = "Tsinghua University Press",

number = "1",

}

TY - JOUR

T1 - Spoken term detection based on DTW

AU - Hou, Jingyong

AU - Xie, Lei

AU - Yang, Peng

AU - Xiao, Xiong

AU - Leung, Cheung Chi

AU - Xu, Haihua

AU - Wang, Lei

AU - Lü, Hang

AU - Ma, Bin

AU - Chng, Engsiong

AU - Li, Haizhou

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Spoken term detection (STD) for low resource languages has drawn much interest. A partial matching strategy based on phoneme boundaries is presented here to solve the fuzzy matching problem in query-by-example spoken term detection with dynamic time warping. A variety of features were used to validate the strategy on the QUESST 2014 dataset. Tests show that this strategy is not only quite effective for fuzzy match tasks T2 and T3 but also effective for the exact match task T1. This strategy has significantly improved performance in fusion tests.

AB - Spoken term detection (STD) for low resource languages has drawn much interest. A partial matching strategy based on phoneme boundaries is presented here to solve the fuzzy matching problem in query-by-example spoken term detection with dynamic time warping. A variety of features were used to validate the strategy on the QUESST 2014 dataset. Tests show that this strategy is not only quite effective for fuzzy match tasks T2 and T3 but also effective for the exact match task T1. This strategy has significantly improved performance in fusion tests.

KW - Dynamic time warping

KW - Low resource languages

KW - Partial matching

KW - Spoken term detection

UR - http://www.scopus.com/inward/record.url?scp=85016616225&partnerID=8YFLogxK

U2 - 10.16511/j.cnki.qhdxxb.2017.21.004

DO - 10.16511/j.cnki.qhdxxb.2017.21.004

M3 - 文章

AN - SCOPUS:85016616225

SN - 1000-0054

VL - 57

SP - 18

EP - 23

JO - Qinghua Daxue Xuebao/Journal of Tsinghua University

JF - Qinghua Daxue Xuebao/Journal of Tsinghua University

IS - 1

ER -

Spoken term detection based on DTW

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this