Cloud gene expression programming based music emotion recognition method

Abstract

The invention discloses a cloud gene expression programming based music emotion recognition method. The method includes following steps: dividing music emotions into a plurality of emotion semantemes, and recording grades, to each emotion semanteme of multiple sample musical compositions, of P test takers; utilizing the cloud emotion labeling method to obtain a normal cloud model of each emotion semanteme of each sample musical composition; extracting voiceprint characteristic parameters of the sample musical compositions, corresponding each sample musical composition to the normal cloud model of the corresponding sample musical composition, and constructing to obtain a music emotion database; utilizing the music emotion database to construct an optimum relationship model of the voiceprint characteristic parameters and the emotion semantemes; and performing music emotion recognition to music according to the optimum relationship model and the voiceprint characteristic parameters of to-be-recognized music. The cloud gene expression programming based music emotion recognition method is based on ensemble learning and cloud gene expression programming, and the optimum relationship model of the music voiceprint parameters and the music emotions is constructed to recognize the music emotions effectively.

Description

technical field [0001] The invention relates to the field of music emotion recognition, in particular to a music emotion recognition method based on cloud gene expression programming. Background technique [0002] Music is the product of great human wisdom. It is not only a sound symbol, but also a language of emotional expression, which can directly cause people to have positive or negative emotions, and then affect people's life and work. As the research reported in the journal Nature pointed out that different music stimulates different emotions, and will lead to different responses in the human brain, affecting people's decision-making (Blood, A.J., Zatorre, R.J., Bermudez, P., and Evans, A.C. Emotional Responses to Pleasant and Unpleasant Music Correlate with Activity in Paralimbic Brain Regions.Nature Neuroscience,1999,2:p382-387); further research also pointed out that different background music can significantly affect customers' choices of wines from different origi...

AI Technical Summary

Problems solved by technology

[0006] However, the machine learning algorithms currently applied to the construction of music emotional cognition models still have many shortcomings. For example, it is still a difficult problem to avoid overfitting by determining the length of the decision tree; Bayesian network needs to know the probability of samples Density; the design and training of artificial neural networks are complex and time-consuming, and a training sample set with sufficient statistics is required; the classification performance of support vector machines depends on the form of kernel functions and the selection of parameters; it is difficult for evolutionary algorithms to find effective individual fitness values function to guide the direction of evolution

Images