Apply evolutionary algorithms for music generation

Abstract

This bachelor’s thesis aims to explore the possibilities of generating new music pieces with the help of algorithms and artificial creativity. The work explores the potential of generating music with computers and reflects on whether the generated music has any musical relevance. To generate pieces of music a genetic algorithm with a tree structured representation was used. For each generation, individuals are evaluated through a combination of human and automatic components which may allow the piece to affect further generations. To investigate the musical relevance; a questionnaire with generated pieces was distributed on social media to people with different musical backgrounds. The results show that our tree structure may have musical potential. The results also shows the comparison of different rating methods and genetic operators and their performance.