A degree of similarity based music search-recommendation algorithm using signal-power spectrum analysis

Abstract

Music is the art to express the thought and emotion of human that has close relationship with human life from the beginning of human civilization. Music search is not easy because of various characteristics of itself, but various methods for searching the music is being tried, recently. The music search technologies are used in various application areas such as title searching based on melody, recommendations for similar music. The previous methodologies have limitations on reliability and accuracy despite that the music recommendation service must satisfy various user requirements. In this paper the analysis method for DoS(Degree of the Similarity) between the music based on signal-spectrum analysis to implement the similar music search-recommendation system for the input music is introduced. The proposed method determines DoS of the music based on featured spectrum of an each music that is generated by frequency-based power spectrum analysis. © 2014 International Information Institute.