Music as a low-cost sleep aid is a promising way to improve the sleep quality of people. However, most available sleep music playlists are limited to generic, soothing songs, which do not take in account personalisation. In collaboration with the Neurology Department of the Reinier de Graaf hospital (Delft, The Netherlands), we explored a profile-based personalisation approach to deliver music that fits with people' sleep and music preferences. Through generative research, we collected people's preference data and proposed four, evocative sleep music profiles: The Explorer, the Diver, the Hunter, and the Observer. The results of the profiling evaluation suggest that the profile experience is credible, intuitive, and easy to use. Four profiles can reflect people's preferences, but may not be stable.