An Emotion-Based Method to Perform Algorithmic Composition

Abstract

The generative music using algorithmic composition techniques has been developed in many years. However it usually lacks of emotion-based mechanism to generate music with specific affective features. In this article the automated music algorithm will be performed based on Prof. Phil Winosr’s “MusicSculptor” software with proper emotion parameter mapping to drive the music content with specific context using various music pa-rameters distribution with different probability control, in order to generate the necessary music emotion automatically. When the emotion scenario varies, the generative music will be logically made via the emotion and context control based on the emotion music classification method. This innovative technique not only generates the emotion music according to the scenario, but also plays the different content of the music every time to make listeners feel “fresh”. The emotion music classification method and the automated music development can be analyzed as the reference for the input of the automated music program. The result shows the proposed method generating music emotions successfully such as happy, angry, sad, and joy, with the correspondent parameter mapping between music and emotion. Although this paper only demonstrates the possibility of emotion-based algorithmic composition, hopefully the proposed idea can be extended to apply into the fields including multimedia and game, to make the background music automatically generated any time according to the context changed by the interaction between human and machine.