Music emotion recognition system based on improved GA-BP

Abstract

In this paper, we proposed music emotion recognition and exploring system based on back propagation neural network and genetic algorithm (GA-BP). For each main rhythm of the music, emotion-related features are extracted and 8 emotion labels are assigned to them, then we find the function between the label and the emotion-related features by GA-BP, BP and the classic GA. Experimental results show classifier based on GA-BP get the highest classification rate of 83.33%. Moreover, based on the definition of music emotion vector, we provided an exploring system for people to search for a song in a fuzzy way.