以演算法作曲為基礎之分析於自動化伴奏與音樂合成研究：古典爵士樂案例分析

Abstract

本論文以演算法作曲 (algorithmic composition) 為基礎，使用馬克夫鏈 (Markov chain) 先產生伴奏和聲的部分，再為其譜上旋律，並使用決策樹 (Decision tree) 為其增添即興部分。

系統以爵士樂為主，分為兩大部分，第一為解構樂曲，分析和聲、音程、以及音高的部分：利用樂曲音符的出現率，以及和聲分析的所有法則，由此加以判斷樂曲之和聲進行走向，與旋律的音高排列方式。第二則是利用上述的分析系統，分析樂曲，並使用馬克夫鏈而產生自動伴奏，再為其加入旋律與節奏的進行規則，由此而產生完整的爵士樂曲 。

本論文提供問卷讓受訪者評論，問卷結果顯示專業的爵士樂手評價最高，其次為非音樂專長者，而古典樂背景則接受度較低。

Algorithmic composition, a type of automated music composition system invented among composers in which specific set rules are designed and its calculations are carried out by a computer. This thesis is based on algorithmic composition, to compose the part of harmony of accompanying with markov chain, and the part of improvisation with decision tree.

The system takes jazz as the main subject, and divides it into two major parts, including auto-analysis, which analyzes harmony, interval and pitch class, and the second part auto-composition which synthesize music elements.

This thesis offers the questionnaires to let interviewees comment on, and reveals that Jazz musicians make the highest point of comment. The second type of the comment is from the non musician, however, the classical musician make the lowest point of comment .