An Interdisciplinary Study of the Unconscious Structures in AI-Generated Music Based on Suno

Abstract

Amid the wave of the Fourth Industrial Revolution, AI music composition has gradually evolved from a technical tool into a means of cultural and psychological expression. This paper adopts an interdisciplinary approach combining psychoanalysis and audio spectral analysis to explore whether AI-generated music exhibits the structural characteristics of a “non-human unconscious.” Based on 18 musical samples generated by the Suno platform, the study utilizes the spectral analysis software iZotope RX to examine five aspects—frequency distribution, dynamic range, noise profile, spatial characteristics, and peak energy—and cross-references them with subjective listening feedback from volunteers. On the theoretical level, the paper draws on Freud’s theory of repression, Lacan’s linguistic structure, and Deleuze’s concept of the desiring-machine to demonstrate how AI music, despite lacking subjective consciousness, may still present structural features that carry “cultural traces of the unconscious.” The study finds that while AI possesses neither awareness nor emotion, its output can nonetheless evoke emotional associations and projections in human listeners. This suggests a mirroring effect at the subconscious level. From the author’s perspective, AI music may be understood as a kind of “unconscious projection device,” offering new dimensions and pathways for reflecting upon the human psychological structure.