The role of musical feature extraction model MFE in multicultural music fusion innovation

With the development of society, multicultural integration in music creation has become a common trend. This topic mainly focuses on the research of music generation based on MFE. Firstly, an MFE method for multicultural music feature extraction is designed by GRU and TDNN algorithms, and then Transformer-XL is used as the base model, and its Mask mechanism is improved to establish a music generation model. A combination of subjective and objective methods is used to evaluate the model-generated music. The overall performance score of the model-generated melody is 4.23, which is 6.21%-94.82% higher than that of the comparison method. Better results were also achieved in chord accuracy and phrase stops. The model-generated arrangements are among the top scores in both online audition and offline performance evaluation scores, indicating that the model-generated arrangements in this paper reach the level of manual composition. The music generation model based on the improved Transformer-XL achieves ideal results overall, which can assist in the fusion of multicultural music creation and promote the diverse development of music creation.