基于特征旋律挖掘的二阶马尔可夫链在算法作曲中的研究与应用  

Research and application on second-order Markov chains based on feature melody mining in algorithmic composition

作　　者：郑银环 王嘉珺[2] 郭威 王备战[2] Zheng Yinhuan;Wang Jiajun;Guo Wei;Wang Beizhan(Dept.of Computer Science&Engineering,Xiamen Institute of Technology,Xiamen Fujian 361021,China;College of Software,Xiamen University,Xiamen Fujian 361001,China)

机构地区：[1]厦门工学院计算机科学与工程系,福建厦门361021 [2]厦门大学软件学院,福建厦门361001

出　　处：《计算机应用研究》2018年第3期849-853,共5页Application Research of Computers

基　　金：福建省中青年教师教育科研项目(JZ160238)

摘　　要：算法作曲是如今国内外许多研究者和作曲家都十分关注和颇感兴趣的一个新兴交叉学科研究领域。首先针对算法作曲领域内已有的各类研究做了资源整合、归纳,在此基础上,结合平时的探究与思考尝试设计和提出一种新的用于音乐片段分析与特征旋律挖掘的ISM(interval sequence mining)算法,即音程序列挖掘算法。该算法是从用于关联规则挖掘的Apriori算法思想中获得启发,结合相关乐理知识和作曲家作曲的心理过程来进行设计的,它以挖掘到的音乐片段中出现次数最多并相对较长的音程子序列作为该音乐片段的特征旋律,并进一步将该算法和二阶马尔可夫链模型相结合,用实验验证了其在生成具有给定训练乐曲音乐风格的新音符序列时的有效性,并以两者结合的方法探索性地创作了一曲名为《巴赫创意变奏曲》的新音乐小品,在算法作曲领域中作了一次新的尝试。Nowadays,algorithmic composition is an emerging interdisciplinary research area that many domestic and foreign researchers and composers are very concerned and interested in.This paper integrated and concluded all kinds of research resources of the existing algorithmic composition,then on this basis,combined with some inquiry and thinking,this paper designed and propose d a new algorithm called ISM(interval sequence mining),for analyzing music clips and mining its feature melody.The ISM algorithm got inspiration from the thought of Apriori algorithm which was used for association rule mining,combined with the psychological process related knowledge of music and the composer,the goal of the ISM algorithm was to find interval subsequence,which had highest frequency and relatively long length in the musical piece.This paper combined the ISM algorithm with the second-order Markov chain model,and validated the effectiveness of composing new music piece with given training in the generation of musical style by doing exploratory experiment.It composed a new musical piece named"Bach creative variations",which was a new attempt in the field of algorithmic composition.

关 键 词：算法作曲 特征旋律挖掘 马尔可夫链 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]