关键词: 音乐特征, 音乐情感, 情感识别, 小波分析, Mallat算法

Abstract: In response to the issue that music-light show control system cannot automatically obtain the required music characteristic information, a kind of rough-emotion model which can be used for music-light performance was presented combined with the traditional Arousal-Valence model. In this model, the Mallat algorithm of wavelet analysis was used to extract comparative items and the ratio of judgment method of strength and rhythm was used to take two "soft-cutting" actions on music. Then the classification of music in the rough-emotion model and the extraction of music characteristic parameters could be achieved by the corresponding production rules of expert system. The simulation results show that the method can effectively classify music clips according to music emotion and extract characteristic elements. Meanwhile, these characteristic elements can satisfy the precision requirement of the music-light show control system on the time domain.

Key words: music feature, music emotion, emotion detection, wavelet analysis, Mallat algorithm