Nishiyama Rina, Nonaka Tetsushi
Graduate School of Human Development and Environment, Kobe University, Kobe, Japan.
Front Robot AI. 2024 Dec 9;11:1463477. doi: 10.3389/frobt.2024.1463477. eCollection 2024.
This study investigated whether a singer's coordination patterns differ when singing with an unseen human partner versus an unseen artificial partner (VOCALOID 6 voice synthesis software). We used cross-correlation analysis to compare the correlation of the amplitude envelope time series between the partner's and the participant's singing voices. We also conducted a Granger causality test to determine whether the past amplitude envelope of the partner helps predict the future amplitude envelope of the participants, or if the reverse is true. We found more pronounced characteristics of anticipatory synchronization and increased similarity in the unfolding dynamics of the amplitude envelopes in the human-partner condition compared to the artificial-partner condition, despite the tempo fluctuations in the human-partner condition. The results suggested that subtle qualities of the human singing voice, possibly stemming from intrinsic dynamics of the human body, may contain information that enables human agents to align their singing behavior dynamics with a human partner.
本研究调查了歌手在与看不见的人类伙伴而非看不见的人工伙伴(VOCALOID 6语音合成软件)一起唱歌时,其协调模式是否存在差异。我们使用互相关分析来比较伙伴和参与者歌声之间的幅度包络时间序列的相关性。我们还进行了格兰杰因果关系检验,以确定伙伴过去的幅度包络是否有助于预测参与者未来的幅度包络,或者反之亦然。我们发现,尽管在人类伙伴条件下存在节奏波动,但与人工伙伴条件相比,在人类伙伴条件下,预期同步的特征更明显,幅度包络展开动态的相似度更高。结果表明,人类歌声的微妙特质,可能源于人体的内在动态,可能包含使人类能够使其唱歌行为动态与人类伙伴保持一致的信息。