基於情緒之自動爵士旋律產生研究

Abstract

本論文為自動作曲系統方法之研究，有別於一般自動作曲系統，本論文將重心放在情緒結合音樂理論與動機結構上。我們將實際操作兩種不同概念的自動作曲系統，第一個系統是以音符參數的機率分佈為基礎，探討情緒參數與音樂各項參數的對應，進而產生出單旋律樂句。第二個系統是以音樂理論與動機結構為基礎，並且結合情緒參數與Pitch Class Set之理論深入探討調性、調式、和聲進行與動機結構的產生的方法，以完成樂句的產生。本論文實作出兩大自動作曲系統並且分別探討，兩系統藉由情緒參數做為觸發，分別將所產生出來的樂句做比對，分析其中差異並且進一步討論情緒、調性與動機結構的關係，以及三者之間的交互影響。希望透過本研究，讓音樂自動作曲系統上能更加細膩與完備。

This thesis is for studying Automated Composition System Method (ACSM), including all music motivation and emotion structure parameters, based on the theory of Western Music. The theory of music alongside with motivation structure is what will be focused on here, unlike under its normal circumstances. Two different conceptual practical operation systems will be tested for ACSM. The first is to use notes parameters on the basis for probability distributions, investigating emotional arguments with the corresponding music parameters. That will produce a single melodic phrase. The second is based on the basis of both of the theory in music and its motivation structure, combining emotion parameters with the theory of Pitch Class Set in its depth of tonality, tuning, and harmonic progression with generating method of motivation structure to complete the production of phrases. Two ACSM were practically tested and discussed separately in this paper, the two systems were triggered by four emotional arguments to compare all produced phrases respectively. The discussions to differences of mood, tonality, the relationship to motivation structure, and the interaction between the three are further analyzed. Perfection and the refinement of ACSM are to be achieved after this study.