Presentation 2013-01-31
Evaluation and Automatic Estimation of Voice Characteristic Similarity Using Isolated Vowels
Shohei TSUJIMURA, Yoichi YAMASHITA, Masanori MORISE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper describes a method of voice characteristic similarity focusing on difference of voice characteristics conveyed by instantaneous spectrum information. Speech data of sentences or words include various factors affecting voice characteristics, such as prosody and time-varying properties. Isolated Japanese vowels (/a/ and /i/) which are uttered with a fixed F0 frequency are used for a listening test and training a model of automatic estimation of the voice characteristic similarity. In the listening test, subjects were asked to judge identity for each pair of speakers. It is confirmed that it is possible to measure similarity of speakers by listening isolated vowels. Weighted Euclidean distance of acoustic features is a measure of automatically estimated similarity of voice characteristic MFCC parameters are used as acoustic features. The performance of estimation is increased by using high order MFCC up to 16. Weighted Euclidean distance shows higher correlation with listening test results than simple Euclidean distance.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Voice characteristics / Speech personality / Isolated vowel / Weighted Euclidean distance
Paper # SP2012-114
Date of Issue

Conference Information
Committee SP
Conference Date 2013/1/23(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Speech (SP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Evaluation and Automatic Estimation of Voice Characteristic Similarity Using Isolated Vowels
Sub Title (in English)
Keyword(1) Voice characteristics
Keyword(2) Speech personality
Keyword(3) Isolated vowel
Keyword(4) Weighted Euclidean distance
1st Author's Name Shohei TSUJIMURA
1st Author's Affiliation Graduate school of Science and Engineering, Ritsumeikan University()
2nd Author's Name Yoichi YAMASHITA
2nd Author's Affiliation College of Information Science and Engineering, Ritsumeikan University
3rd Author's Name Masanori MORISE
3rd Author's Affiliation College of Information Science and Engineering, Ritsumeikan University
Date 2013-01-31
Paper # SP2012-114
Volume (vol) vol.112
Number (no) 422
Page pp.pp.-
#Pages 6
Date of Issue