Novice-Centered Application Design for Music Creation

Atsuya Kobayashi, Tetsuro Sato, Kei Tateno

Abstract: In the field of music production, the integration of machine learning-based technologies, particularly in the form of intelligent digital audio workstations and smart plugins, has gained traction. However, challenges arise when designing for novices, as controlling the models requires musical knowledge, and this could decrease the motivation to create. To mitigate these issues and serve the needs of novice users, we developed MARLEY, an application that employs machine learning to recommend loop sound samples and facilitate music prototyping. The user study revealed that participants enjoyed the experience, indicating its potential usefulness in the early stages of music production, while providing feedback for further improvements. Future work will focus on enhancing the model’s accuracy and explainability, as well as evaluating the tool’s usability and its impact on creativity support for novices among a broader and more diverse user base.