Research on Deep Learning Model Training and Cross-Ethnic Cultural Interaction Mechanism of Daur Music Inheritance in Heilongjiang Province

Daur ethnic group is a people with a long history and unique national culture, the ethnic group has been washed by a long time, and its traditional music has distinctive cultural characteristics. However, with the passage of time, the creative background of Daur traditional music continues to disappear, coupled with the lack of text and other issues, Daur traditional music heritage and rescue work is imminent. Therefore, this paper centers on the research of Daur music in Heilongjiang Province, analyzes the cultural characteristics of Daur music, and explores the cross-ethnic cultural interaction mechanism of Daur music. In order to solve the inheritance problem of Daur music, this paper combines information technology, uses generative adversarial network on the basis of deep learning, and constructs a MIDI-GAN Daur music generation model. Through model training and experimental comparison, the performance of the model in this paper is verified. And the generated Daur music is evaluated from both subjective and objective aspects, and the objective evaluation indexes such as confusion degree, pitch category, rhythmic consistency, and empty beat rate of the generated model are optimal. And in the subjective evaluation of the model each index score of 3.85, 3.52, 4.21, 4.11, the overall quality score of 3.92 points, also obtained the highest evaluation results, the model generated Daur music is more realistic, can convey the expected emotions, for the inheritance of Daur music provides a new idea of the problem.