ABSTRACT Querying music is still a disembodied process in Music Information Retrieval. Thus, the goal of the presented study was to explore how free and spontaneous movement captured by smartphone accelerometer data can be related to musical properties. Motion features related to tempo, smoothness, size, and regularity were extracted and shown to predict the musical qualities ‘rhythmicity’ (R² = .45), ‘pitch level + range’ (R² = .06) and ‘complexity (R² = .15). We conclude that (rhythmic) music properties can be predicted from movement, and that an embodied approach to MIR is feasible.

中文翻译：

---

从加速到节奏：智能手机评估的运动预测音乐的属性

摘要 在音乐信息检索中，查询音乐仍然是一个非实体的过程。因此，本研究的目标是探索智能手机加速度计数据捕获的自由和自发运动如何与音乐特性相关。与速度、平滑度、大小和规律性相关的运动特征被提取并显示以预测音乐品质“节奏性”（R² = .45）、“音高 + 范围”（R² = .06）和“复杂性（R² = .06）”。 15)。我们得出结论，可以从运动中预测（有节奏的）音乐特性，并且对 MIR 的具体方法是可行的。