Suno AI와 Udio AI로 만든 음악 작품 분석

Abstract

This study aims to create and analyze music pieces using generative AI that can be used in the field of creative education. Suno and Udio AI are useful generators for vocal music creation, and they ask learners for input values in the form of prompts. As a first step for creation, we organized Suno and Udio’s creation methods and performed creation activities that are frequently duplicated in textbooks. The results of analyzing the created music pieces are as follows. First, Suno and Udio’s works maintain traditional functional harmonic structures in terms of harmonic progression and sense of composition. In particular, in the process of implementing the texture of music, they tend to flex ibly create linear low-pitched flows or transitional connections and repeat easy and simple harmonic patterns. Second, Suno and Udio’s works change in form each time they create music. In other words, they create music with very diverse structures rather than consistent forms, and in this study, only Suno’s work 1 confirmed a standardized 8-bar 2-part form. Third, Suno and Udio have considerable dissonance in the combination of lyrics and melody. All four works have lyrics and melodies that are not natural, which limits the ability to convey the meaning of the lyrics, and the lyrics are arranged without considering the beat and stress, making them difficult to sing along to. In addition, there was an error in the program where the lyrics and the sound source did not match, and the sound source was not accurately generated.