While many researchers have proposed various ways of quantifying recommendation list diversity, these approaches have had little input from users on their own perceptions and preferences in seeking diversity. Through an exploratory user study, we provide a better understanding of how users view the concept of diversity in music recommendations, and how they might optimise levels of intra-list diversity themselves. In our study, 17 participants interacted with and rated the suggestions from two different recommendation systems. One provided static top-7 collaborative filtering recommendations, and the other provided an interactive slider to re-rank these recommendations based on a continuous diversity scale. We also asked participants a series of free-form questions on music discovery and diversity in semi-structured interviews. User-preferred levels of diversity varied widely both within and between subjects. Although most users agreed that diversity is beneficial in music discovery, they also noted a risk of dissatisfaction from too much diversity. A key finding is that preference for diversification was often linked to user mood. Participants also expressed a clear distinction between diversity within existing preferences, and outside of existing preferences. These ideas of inner and outer diversity are not well defined within the bounds of current diversity metrics, and we discuss their implications.