Speaker clustering for speech recognition using vocal tract parameters

M. Naito, L. Deng, Yoshinori Sagisaka

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

We propose speaker clustering methods for speech recognition based on vocal tract (VT) size related articulatory parameters associated with individual speakers. Two parameters characterizing gross VT dimensions are first derived from the formant frequencies of two vowels and are then used to cluster speakers. The resulting speaker clusters are significantly different from speaker clusters obtained by conventional acoustic criteria. Then phoneme recognition experiments are carried out by using speaker-clustered HMMs (SC-HMMs) trained for each cluster. The proposed method requires a small amount of speech data for speaker clustering and for selecting the most suitable SC-HMM for a target speaker, but gives higher recognition rates than conventional speaker clustering methods based on acoustic criteria.

Original languageEnglish
Pages (from-to)305-315
Number of pages11
JournalSpeech Communication
Volume36
Issue number3-4
DOIs
Publication statusPublished - 2002 Mar
Externally publishedYes

Fingerprint

Speech Recognition
Speech recognition
Cluster Analysis
Acoustics
Clustering
acoustics
Clustering Methods
Gross
Two Parameters
Experiments
experiment
Target
Vocal Tract
Experiment

Keywords

  • Speaker-clustering
  • Speech recognition
  • Vocal tract parameters

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering
  • Experimental and Cognitive Psychology
  • Linguistics and Language

Cite this

Speaker clustering for speech recognition using vocal tract parameters. / Naito, M.; Deng, L.; Sagisaka, Yoshinori.

In: Speech Communication, Vol. 36, No. 3-4, 03.2002, p. 305-315.

Research output: Contribution to journalArticle

@article{4c4ab15e0e0549bea3849fc444018c55,
title = "Speaker clustering for speech recognition using vocal tract parameters",
abstract = "We propose speaker clustering methods for speech recognition based on vocal tract (VT) size related articulatory parameters associated with individual speakers. Two parameters characterizing gross VT dimensions are first derived from the formant frequencies of two vowels and are then used to cluster speakers. The resulting speaker clusters are significantly different from speaker clusters obtained by conventional acoustic criteria. Then phoneme recognition experiments are carried out by using speaker-clustered HMMs (SC-HMMs) trained for each cluster. The proposed method requires a small amount of speech data for speaker clustering and for selecting the most suitable SC-HMM for a target speaker, but gives higher recognition rates than conventional speaker clustering methods based on acoustic criteria.",
keywords = "Speaker-clustering, Speech recognition, Vocal tract parameters",
author = "M. Naito and L. Deng and Yoshinori Sagisaka",
year = "2002",
month = "3",
doi = "10.1016/S0167-6393(00)00089-3",
language = "English",
volume = "36",
pages = "305--315",
journal = "Speech Communication",
issn = "0167-6393",
publisher = "Elsevier",
number = "3-4",

}

TY - JOUR

T1 - Speaker clustering for speech recognition using vocal tract parameters

AU - Naito, M.

AU - Deng, L.

AU - Sagisaka, Yoshinori

PY - 2002/3

Y1 - 2002/3

N2 - We propose speaker clustering methods for speech recognition based on vocal tract (VT) size related articulatory parameters associated with individual speakers. Two parameters characterizing gross VT dimensions are first derived from the formant frequencies of two vowels and are then used to cluster speakers. The resulting speaker clusters are significantly different from speaker clusters obtained by conventional acoustic criteria. Then phoneme recognition experiments are carried out by using speaker-clustered HMMs (SC-HMMs) trained for each cluster. The proposed method requires a small amount of speech data for speaker clustering and for selecting the most suitable SC-HMM for a target speaker, but gives higher recognition rates than conventional speaker clustering methods based on acoustic criteria.

AB - We propose speaker clustering methods for speech recognition based on vocal tract (VT) size related articulatory parameters associated with individual speakers. Two parameters characterizing gross VT dimensions are first derived from the formant frequencies of two vowels and are then used to cluster speakers. The resulting speaker clusters are significantly different from speaker clusters obtained by conventional acoustic criteria. Then phoneme recognition experiments are carried out by using speaker-clustered HMMs (SC-HMMs) trained for each cluster. The proposed method requires a small amount of speech data for speaker clustering and for selecting the most suitable SC-HMM for a target speaker, but gives higher recognition rates than conventional speaker clustering methods based on acoustic criteria.

KW - Speaker-clustering

KW - Speech recognition

KW - Vocal tract parameters

UR - http://www.scopus.com/inward/record.url?scp=0036497667&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036497667&partnerID=8YFLogxK

U2 - 10.1016/S0167-6393(00)00089-3

DO - 10.1016/S0167-6393(00)00089-3

M3 - Article

VL - 36

SP - 305

EP - 315

JO - Speech Communication

JF - Speech Communication

SN - 0167-6393

IS - 3-4

ER -