Presentation 2016-01-14
Objective evaluation of synthetic speech using association between dimensions within spectral features
Yusuke Ijima, Taichi Asami, Hideyuki Mizuno,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper proposes a novel objective evaluation technique for statistical parametric speech synthesis. A novel point of the proposed technique is that it utilizes the association between dimensions within the spectral features. We first analyze the subjective scores obtained with respect to the associations of spectral features of natural and various synthesized speech by using a maximal information coefficient (MIC). The analysis results show that the scores improve with weaker association. We then propose the proposed objective evaluation index, which uses a voice conversion technique to detect the associations for each speech. We perform subjective and objective experiments and evaluate the performance results obtained by comparing them with the obtained subjective scores and the conventional objective evaluation index, i.e., mel-cepstral distortion. The results indicate that our proposed objective evaluation index is more effective than the mel-cepstral distortion.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Statistical parametric speech synthesis / objective evaluation / spectral features / maximal information coefficient
Paper # SP2015-90
Date of Issue 2016-01-07 (SP)

Conference Information
Committee SP
Conference Date 2016/1/14(1days)
Place (in Japanese) (See Japanese page)
Place (in English) Sunpian Kawasaki
Topics (in Japanese) (See Japanese page)
Topics (in English) Synthesis, Generation, Prosody, etc.
Chair Kazunori Mano(Shibaura Inst. of Tech.)
Vice Chair Norihide Kitaoka(Tokushima Univ.)
Secretary Norihide Kitaoka(Tokyo City Univ.)
Assistant Takashi Nose(Tohoku Univ.) / Taichi Asami(NTT)

Paper Information
Registration To Technical Committee on Speech
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Objective evaluation of synthetic speech using association between dimensions within spectral features
Sub Title (in English)
Keyword(1) Statistical parametric speech synthesis
Keyword(2) objective evaluation
Keyword(3) spectral features
Keyword(4) maximal information coefficient
1st Author's Name Yusuke Ijima
1st Author's Affiliation Nippon Telegraph and Telephone Corporation(NTT)
2nd Author's Name Taichi Asami
2nd Author's Affiliation Nippon Telegraph and Telephone Corporation(NTT)
3rd Author's Name Hideyuki Mizuno
3rd Author's Affiliation Tokyo University of Science, Suwa(TUSS)
Date 2016-01-14
Paper # SP2015-90
Volume (vol) vol.115
Number (no) SP-392
Page pp.pp.27-32(SP),
#Pages 6
Date of Issue 2016-01-07 (SP)