Study on the composition rules for Chinese Jiangnan ditty

Rule-based system and genetic algorithm are artificial intelligence methods for algorithmic composition. Rule-based system is aimed at isolating the aesthetic code of a certain musical genre and using this code to create new similar compositions. In genetic algorithm, the chromosomes are evaluated by utilizing fitness function and how to define the fitness function is the most important key. This paper took typical Chinese traditional music-Jiangnan ditty as an object of study. Rule-based systems of Jiangnan ditty were established based on two methods. Firstly, acoustical features were extracted to characterize the musical genre of Jiangnan ditty. Secondly, a set of melodic features were established through statistical analyzing each piece of melody as a linear sequence of notes to describe pitch range, melodic interval, mode, note duration, timbre, rhythm pattern, and intervallic motive of Jiangnan ditty. According to the results, fitness function was defined, which would provide the judging basis for the subsequent genetic algorithm.