Study On Computer Analysis And Recognition Of Musical Emotion

The affective computing is another research focus in the field of intelligence Human-Computer Interaction, since the multimedia management technology rose. As music is now acting as the mainstream of multimedia resources, various kinds of electronic music technology has been produced and developed. Especially, music is a carrier of mood, mood is the connotation of music. To detect and classify the emotion of music is a new domain of multimedia content processing, which is also attracting more and more attention. In this paper, we focus on key problems of music effective computing and music mood recognition, and then present a discussion on the direction of further research in the future.This paper consists of three major aspects:Firstly it is analyzed that the emotion discerns, especially research history and current situation about the study on elements of expression in music and classification of music affection. At the end this paper presents a linguistic model which is used for our experimental expression of music emotion.Secondly, the compositions of the musical acoustics principle and digital audio as MIDI files are analyzed, which is regarded as the basis to draw and detect the essential feature of music. Then based on BP network, this paper proposed a method to automatically locating the melody of MIDI files, which is carrying the essence of music emotion.Thirdly, study on different methods for musical recognition was discussed, with a focus on the development, characteristic, model, operation principle and algorithms of BP network. An experimental musical emotion recognition system was developed, which proves the possibility of application BP network to music emotion recognition, and rationality and versatility of the BP network.At the end, we have a analysis on the technical application in the multimedia field, and further set up a typical application scenario and present it on the Matlab for emulation.