Abstract
Recent studies indicated that depression is the main cause of mental disability. Therefore, reducing the incidence of depression has become an important issue nowadays. Among numerous therapies for depression, music therapy could effectively relieve depression and other psychological disorders. Currently, therapeutic music is selected according to subjective judgments of therapist. No objective mechanism has been developed to select therapeutic music. Music information retrieval (MIR) is the most prevalent method for music classification, which is an emerging research field that received growing attention from both academia and music industry. However, only a few research integrated the application of music therapy with music information retrieval with holistic music pattern analysis. Therefore, the aim of this study was to extract positive/negative music pattern by analyzing amplitude, frequency and tempo through data mining and reverse engineering. Three hundred 30-second-long song were adopted as training data. The pattern of pleasant/negative music will be identified and classified. This induction contributed to the creation of therapeutic music. The results of this study not only can be applied in the music therapy, but also practice in music services of related fields.