One embodiment provides a non-transitory medium containing computer program code that performs an operation that includes retrieving, for each of two or more identified users, a respective musical preferences model representing a plurality of acoustical characteristics of musical content determined based on preferences of the respective user. Additionally, the operation includes generating a common musical preferences model representing a set of common acoustical characteristics of musical content for the two or more identified users, based on the musical preferences models for each of the two or more identified users. The operation also includes identifying a library of available musical content. Furthermore, the operation includes creating a playlist of two or more musical selections from the library of available musical content, based on the common musical preferences model.