@techreport{oai:ipsj.ixsq.nii.ac.jp:00211568,
 author = {渡邊, 悠希 and 坂本, 修一 and 星, 貴之 and 長谷, 芳樹 and 中野, 学 and Yuki, Watanabe and Shuichi, Sakamoto and Takayuki, Hoshi and Yoshiki, Nagatani and Manabu, Nakano},
 issue = {59},
 month = {Jun},
 note = {聴取により心が和む音声の合成を最終的な目的として，様々な感情を想定して発話された音声から認識・想起した感情を評価するとともに，その音声のもつ音響特徴量を分析した．感情評価実験の結果，癒やしの感情は感情次元の valence が正，activation が負の象限に付置されること，癒やしの程度において認識した感情と想起した感情との間に正の相関があることが明らかとなった．加えて，癒やしの感情とある程度の相関を有する音響特徴量は明らかとなった一方で，通常の音声合成等に用いられる低次の音響特徴量との強い相関は見られなかった．このことは， 癒やしの感情を想起させる音声を安定して合成するためには，より適切な音響特徴量を探索する必要があることを示唆している．, The goal of this study is to synthesize speech sound which induces relaxed emotion. As the preliminary study, we investigated how listeners felt when they heard speech sounds with various emotions. In addition, we analyzed the relationship between perceived emotions and the acoustic features of the speech sound with emotions. The results of the experiment revealed that relaxed emotion was categorized in the quadrant of positive valance and negative activation. Moreover, a positive correlation was observed between the recognized and evoked relaxed emotions. On the other hand, the strong correlation between low-order acoustic features commonly used to synthesize emotional speech sound and the amount of perceived relaxed emotion was not seen. These results suggest that more appropriate acoustic features need to be considered for the stable synthesis of the speech sound that induces relaxed emotions in the future work.},
 title = {聴取によって心が和む音声の合成に向けた基礎検討～認識・想起した感情および音響特徴量の観点から～},
 year = {2021}
}