Receptive music therapy is active hearing of music that is specifically selected to cause a certain effect on a person, such as pain reduction, mental opening, confrontation etc. This active, guided hearing could be helpful as a supporting ritual for patients at home and could extend traditional therapy. However, patients are often unable to select the music pieces that might be helpful for them in a current situation. We are suggesting a self-learning decision support system that allows a patient to answer questions on music anamnesis, is ready for inclusion into an electronic health record, and which enables a therapist to compile a therapeutic music program for the patient at home. Beyond this, the system also suggests appropriate music and duration of listening based on the patient's reported current mental state. In this paper, a concept for such a mobile system for receptive music therapy will be proposed.