Melody Generation System Based on Generalization by Human Causal Intuition

摘要

In this paper, we import the Cognitive Biases to the field of music generation, and point out its merit. We adopt the Loosely Symmetric (LS) Model based on human illogical biases of cognition, which includes the symmetric bias and the mutual exclusivity bias, as a melody generation. We regard features of melodies as the simplest sequence of transitional events from preceding note to succeeding note. The system with built-in cognitive bias extracts feature quantities of existing melodies, and generates short melodies from features of existing melodies. Compared to melodies from Conditional Probability (CP) model and Rigidly Symmetric (RS) model, melodies from LS model affected by the symmetric bias and the mutual exclusivity bias. This result indicates that it is possible to generate melody in a more human-like way through the use of a system.