Quality-diversity for Synthesizer Sound Matching

Synthesizers have become commonplace in music production due to their ability create a diverse and unique timbre. However, many users find it difficult control synthesizer the complexity of relationship between synthesis parameters output sound. As such, sound matching task focuses on estimation that recreate target as closely possible within capabilities synthesizer. In this paper, we introduce quality-diversity approach problem by introducing novelty objective genetic algorithm-based matching. We show QD method allows for discovery collection matches are more compared found standard algorithms. These spread across space interpretable spectral features, making useful intuitive matches.