A structural language and automated performance of music

本研究旨在对音乐表演进行建模和分析。本论文主要研究三个问题：一是设计并实现了一个用于计算机辅助钢琴演奏的结构化语言MUSE;二是对专业钢琴家演奏的分析;三是在前两个问题的基础上对钢琴演奏进行建模与仿真。结果如下。1. MUSE语言的特点是将音符组织成一个层次化的数据结构，并将演奏实践参数化为动力学，agogics，发音，移位和阻尼器运动。整个系统被实现为由个人计算机托管的双电子钢琴。2.建立了一个识别预振速度的数学模型。借助这一模型，我们对几位钢琴家演奏的音乐进行了三拍子的韵律节奏的测量和分析。结果表明，不同的钢琴家在同一段音乐中实现的节奏模式有着明显的特征性区别。3.线性密度假设已经被引入以适应两个连续音符组的演奏速度的不连续变化，其中一个音符组由相同音符值d的音符组成，而另一个音符组由相同音符值d '、d'与d不同的音符组成。在这种情况下改变速度的经验公式进行了讨论和测试。

This research project aims at a modeling and analysis of musical performances. Three subjects have been dealt with: Design and implementation of a structural language MUSE for computer aided performance of piano music; Analysis of musical performances realized by professional pianists; and modeling and simulation of musical performances based on the first two. The results are as follows. 1. Language MUSE is characterized by the organization of musical notes into a hierarchical date structure and the parametrization of performance practice into terms of dynamics, agogics, articulation, shift, and damper motion. The entire system has been implemented as a dual electronic piano hosted by a personal computer. 2. A mathematical model has been developed for the identification of performing velocity. With the aid of this model, agogic metrical rhythms in triple time have been measured and analysed for music rendered by several pianists. The result shows a clear, characteristic distinction in rhythmic patterns realized by different pianists for the same passages of music. 3. The linear density hypothesis has been introduced to fit to a discontinuous change in performing velocity of two successive note groups where one consists of notes of the same note value,d, and the other consists of notes of the same note value, d', d' being different from d. An empirical formula for changing velocity in such situations has been discussed and tested.