Research on Multimodal Generation Strategy of Opera Style and Music Melody Based on Time Series Analysis

The application of artificial intelligence on the field of art can be used to assist the creation of musicians and provide new creative ideas for musicians.In this paper, firstly, an ARIMA model is established for the prediction problem of opera style, which is used to predict the trend of the development of opera style sequence, and the best model is selected according to the minimum information criterion and Bayesian criterion.Then an automatic music melody generation method based on the generative adversarial network framework is proposed, which applies the trained natural language generation model to music generation to textualize the music melody and reduce the model running time.In addition to this a barization music melody generation method is also used, which divides a large music melody into melodic segments and generates them segment by segment, reducing the difficulty of the model in generating the music melody.Finally, the Fourier transform method is used to extract the features of the music melody and complete the visualization of the music melody.The model ARIMA(2,1,1)(2,1,0)12 that best fits with the time-series prediction of the development of opera styles was identified through empirical analysis.The PB value of Leak-GAN_2 model in this paper is improved by 41.38% compared with MusicGAN.It shows that both the opera style prediction model and the music melody multimodal generation model constructed in this paper have better effect and certain advancement.