Digital transformation has changed all aspects of life, including the music market and listening habits. The spread of mobile devices and music streaming services has enabled the possibility to access a huge selection of music regardless of time or place. However, this access leads to the customer's problem of choosing the right music for a certain situation or mood. The user is often overwhelmed while choosing music. Context information, especially the emotional state of the user, can help within this process. The possibilities of an emotional music selection are currently limited. The providers rely on predefined playlists for different situations or moods. However, the problem with these lists is, that they do not adapt to new user conditions. A simple, intuitive and automatic emotion-based music selection has so far been poorly investigated in IS practice and research. This paper describes the IS music research project "Moosic", which investigates and iteratively implements an intuitive emotion-based music recommendation application. In addition, an initial evaluation of the prototype will be discussed and an outlook on further development will be given.