MUSICAL REPRESENTATION OF GENRE INFORMATION IN GENERATIVE ARTIFICIAL INTELLIGENCE: THE CASE OF SUNO
Öz
In generative artificial intelligence-based music systems, the genre name is one of the main textual prompts that influences the character of the music produced. This study investigates which indicators genre information is reflected through in musical outputs generated on the Suno platform and to what extent the style influence parameter changes this reflection. Designed as a qualitative case study, the research produced musical outputs in the genres of pop, rock, jazz, Turkish folk music, and arabesk using the same Turkish lyric pattern. For each genre, three outputs were generated at style influence values of 0, 25, 50, 75, and 100, resulting in a dataset of 75 music examples. The examples were examined through structured content analysis based on technical indicators such as duration, tempo, tonal centre, and vocal profile, as well as the codes of instrument-based genre representation, vocal-based genre representation, and melodic coherence. While genre representation in pop and jazz was found to be strongly established even at low style influence levels, this representation became more pronounced in rock as the parameter increased. Positive results were observed in arabesk productions, but this increase did not follow a regular pattern. In Turkish folk music, genre representation generally remained weak in terms of instrumentation, vocal character, and melodic coherence. The results show that the effect of the style influence parameter varies across genres and that this effect is related to how strongly the system already represents the relevant genre.
Anahtar Kelimeler
Generative Artificial Intelligence, Suno, Text-to-Music Generation, Genre Representation, Style Influence
Kaynakça
- Agostinelli, A., Denk, T. I., Borsos, Z., Engel, J., Verzetti, M., Caillon, A., Huang, Q., Jansen, A., Roberts, A., Tagliasacchi, M., Sharifi, M., Zeghidour, N., & Frank, C. (2023). MusicLM: Generating music from text. arXiv. doi: 10.48550/arXiv.2301.11325
- Bogdanov, D., Wack, N., Gómez, E., Gulati, S., Herrera, P., Mayor, O., Roma, G., Salamon, J., Zapata, J., & Serra, X. (2013). ESSENTIA: An audio analysis library for music information retrieval. In Proceedings of the 14th International Society for Music Information Retrieval Conference.
- Chopra, A., Roy, A., & Herremans, D. (2024). MIRFLEX: Music information retrieval feature library for extraction. In Extended abstracts for the late-breaking demo session of the 25th International Society for Music Information Retrieval Conference.
- Deruty, E., Grachten, M., Lattner, S., Nistal, J., & Aouameur, C. (2022). On the development and practice of AI technology for contemporary popular music production. Transactions of the International Society for Music Information Retrieval, 5(1), 35-49. doi: 10.5334/tismir.100
- Fabbri, F. (1981). A theory of musical genres: Two applications. In D. Horn & P. Tagg (Eds.), Popular music perspectives (pp. 52-81). International Association for the Study of Popular Music.
- Holt, F. (2007). Genre in popular music. University of Chicago Press.
- Huang, Q., Park, D. S., Wang, T., Denk, T. I., Ly, A., Chen, N., Zhang, Z., Zhang, Z., Yu, J., Frank, C., Engel, J., Le, Q. V., Chan, W., Chen, Z., & Han, W. (2023). Noise2Music: Text-conditioned music generation with diffusion models. arXiv. doi: 10.48550/arXiv.2302.03917
- Kennett, C. (2003). Is anybody listening? In A. F. Moore (Ed.), Analyzing popular music (pp. 196-217). Cambridge University Press.
- Krippendorff, K. (2004). Content analysis: An introduction to its methodology (2nd ed.). Sage Publications.
- Liu, V., & Chilton, L. B. (2022). Design guidelines for prompt engineering text-to-image generative models. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems (Article 384, pp. 1-23). Association for Computing Machinery. doi: 10.1145/3491102.3501825
