ABSTRACT

This study explores the educational potential of prompt engineering in AI-based music creation, focusing on the generative music platform SUNO. By designing prompts based on five variables—genre, instrument, mood, tempo, and technique—I generated and analyzed corresponding 120 audio samples to examine how prompt design affects musical outcomes in school music learning contexts. By using audio feature extraction and quantitative analyses such as similarity matrices, clustering, and dimensionality reduction, I found that detailed and well-structured prompts led to distinct and predictable musical results. In particular, prompts sharing key elements such as instrument or technique produced highly similar audio output, while more varied prompts yielded greater diversity. These findings highlight that prompt engineering can serve as a pedagogical strategy to guide student creativity, support process-oriented learning, and facilitate reflective music inquiry. Based on its finding, this study lays a practical foundation for the educational value of prompt-based AI music creation activities in music education.

Keyword :