Novel music discovery concepts: user experience and design implications

Abstract

Current music consumers are facing an almost endless selection of music in online services to be accessed on-demand with a variety of devices. The focus has now shifted from providing on-demand access to massive music catalogs towards improving the user experience of the music services, providing new ways of finding relevant music from the massive online catalogs, and making music consumption a pleasurable experience. The key differentiation aspects for music services come largely from the user interface and the ways that music can be found or consumed.

This thesis belongs to the fields of human-computer interation (HCI) and music information retrieval (MIR). HCI is concerned with the design, evaluation and implementation of interactive computing systems and MIR targets to broaden the understanding and usage of musical data through research, applications and tools.

This thesis studies novel concepts for music discovery that are based on strong visual metaphors and stereotypes. The goal is to research the user experience (UX) of novel music discovery services and to formulate key design implications to support service development for music discovery. The research of music discovery prototypes consisted of three main phases: initial concept design phase, playful concept exploration phase, and iterative concept design phase. The thesis introduces, in total, ten prototype implementations of these novel concepts for music discovery. User evaluations of the implemented prototypes were conducted with Finnish active music listeners with both qualitative and quantitative research methods. This thesis contributes to both academic research on HCI in MIR and commercial music discovery service development.

The results provide insights to user experience with different types of novel music discovery services. Five novel music discovery services using the same content-based music recommendation back-end were compared and the comparison results are reported including both first impressions and longer-term usage. Additionally, the results of the studies introduce a wide set of future directions for each music discovery approach. These future directions enable service developers to further enhance the music discovery experience within these fields.

All but one of the proposed music discovery concepts work well for music discovery. The use of avatar characters and mood pictures for music discovery are the most promising ones. The results show that visual music discovery services have the potential to replace traditional music discovery services in different types of music discovery practices. The final contribution of the thesis is a set of 16 design implications for music discovery service developers.