抄録
In automatic composing system, the system has to generate a music that matches to the user's preferences and favors. To solve the problems it is necessary to get parameters for numerizing and estimating user's preferences and favors. As related works, some interactive methods by which parameters and got ea from user's response were proposed. Some system also researched that automatically extract characteristic values from sample musics. In this paper, we propose an automatic composing system that generate melody matching to user's favors. This system consists complex network as a melody generation engine, and artificial neural network as a melody evaluation engine. By this system, we try to generate and evolve a prefer melody from sample musics.
