The Impact and Integration Path of Artificial Intelligence Composition Technology on the Traditional Music Composition Mode

Abstract

With the rapid development of deep learning technology, artificial intelligence composition technology has been able to generate musical works that are basically equal to traditional music composition in terms of sensory indicators, which has made a substantial impact on the main position of traditional music composition mode. In this paper, recurrent neural networks are used to learn the time-dependent features of musical sequences, and then generate adversarial networks to enhance the authenticity of the musical score, and express the music potential space in the form of probability distribution parameters. Rule-based algorithms are introduced to provide intervenable structural constraints, effectively generating scores with original musical style and rhythmic characteristics, and based on the above methodology, we explore the impact and integration possibilities of AI on the traditional mode of music creation. Experiments show that the PC and PI values of the AI composition method based on the RVAE-GAN model are improved by 8.57% and 5.89% compared with the Music Transformer model. The AI composition technique approaches the average level of traditional human work composition mode in terms of sensory acceptance and specific style maintenance. The above results confirm that the impact of AI on the traditional music composition mode is substantial, and the integration paths such as emotion-structure dual-drive and complementary style inheritance are feasible directions to cope with the impact.