Presentation 2022-02-28
Estimation of speech emotional intensity model using impression ratings
Megumi Kawase, Minoru Nakayama,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We used deep learning to estimate emotional intensity from speech. In our previous study, we considered emotional intensity as 10 categories and estimated emotional intensity by categorization, but the flexibility of this method was insufficient. To solve this problem, in this study, we calculated the perceived intensity values using two evaluation methods by means of an emotional perception evaluation experiment of speech. As a result, we confirmed the nonlinearity of the intensity values of each emotional intensity category in the emotional perception evaluation experiment. As a result, the estimated intensity values with an average correlation coefficient of 0.73 with the set intensity values of speech were obtained in the emotional intensity estimation model using perceived intensity values.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) speech / emotion / intensity / deep learning / emotional perception evaluation experiment
Paper # HIP2021-66
Date of Issue 2022-02-20 (HIP)

Conference Information
Committee HIP / ITE-HI / ASJ-H / VRPSY
Conference Date 2022/2/27(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Online
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Shuichi Sakamoto(Tohoku Univ.) / Takehiro Nagai(Tokyo Inst. of Tech.) / Hiroaki Kato(NICT) / Shinji Nakamura(Nihon Fukushi Univ)
Vice Chair Yuji Wada(Ritsumeikan Univ.) / Sachiko Kiyokawa(Nagoya Univ.) / / Shuichi Sakamoto(Tohoku Univ.)
Secretary Yuji Wada(NTT) / Sachiko Kiyokawa(NICT) / (NHK) / Shuichi Sakamoto(Kinki Univ.) / (Kanagawa Univ.)
Assistant Yuki Yamada(Kyushu Univ.) / Daisuke Tanaka(Tottori Univ.) / Ippei Negishi(Kanazawa Inst. of Tech.)

Paper Information
Registration To Technical Committee on Human Information Processing / Technical Group on Human Information / Auditory Research Meeting / SIG-VRPSY
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Estimation of speech emotional intensity model using impression ratings
Sub Title (in English)
Keyword(1) speech
Keyword(2) emotion
Keyword(3) intensity
Keyword(4) deep learning
Keyword(5) emotional perception evaluation experiment
1st Author's Name Megumi Kawase
1st Author's Affiliation Tokyo Institute of Technology(Tokyo Tech)
2nd Author's Name Minoru Nakayama
2nd Author's Affiliation Tokyo Institute of Technology(Tokyo Tech)
Date 2022-02-28
Paper # HIP2021-66
Volume (vol) vol.121
Number (no) HIP-381
Page pp.pp.42-47(HIP),
#Pages 6
Date of Issue 2022-02-20 (HIP)