Automatic Music Composition System Based on Genetic Programming and Surrogate Model with Deep Learning

Abstract

Automatic music composition is one of the most difficult and attractive challenges in the artificial intelligence (AI) field. In order to tackle this challenge, an approach using interactive evolutionary computation (IEC) is drawing attention because IEC takes human emotions into consideration. The major problem in using IEC is that the number of evaluations from one user is limited due to user fatigue. To tackle this problem, a surrogate model is often introduced into IEC. An approach based on deep learning (DL) is also common in this field because of many quantitative futures. However, the approach hardly considers human emotions. In this study, we proposed the automatic music composition system based on IEC and a surrogate model called evaluation model. The model is constructed with a DL model, thus our system can compose music reflected human emotions quantitatively. The experiments are carried out to show the effectiveness of the proposed method.