Musique Méta-Concrète is an interdisciplinary research-creation project exploring a new generation of sonic materials: sounds generated by large language models (LLMs) and contemporary AI audio systems. This project extends the historical trajectory of musique concrète—a practice that fractured the bond between sound and its physical cause—into a moment when the “material” of sound increasingly emerges from computational, non-physical, and statistically inferred sources. If Pierre Schaeffer’s sound objects sought to free sound from its origins, musique méta-concrète contends with the inverse problem: sound that originates nowhere recognizable.

​

The project proposes a new category of sonic object. AI-generated sounds exist in an ambiguous state: they are constructed from huge archives of recorded and textual data, yet lack any singular physical referent. They resemble field recordings, instrumental gestures, mechanical noise, or environmental atmospheres—but they arrive from latent space, not from microphones. Their sonic “materiality” is virtual yet affectively real. They occupy a liminal territory where familiarity meets uncanniness.

​

This project positions these emergent sound objects not as curiosities or substitutes for acoustic reality, but as legitimate artistic materials requiring careful handling, critical thinking, and new compositional approaches. musique méta-concrète takes seriously the idea that LLM-generated audio constitutes its own form of “meta-reality”—a speculative acoustics produced by statistical memory, collective cultural traces, and machine inference.

​

By integrating these materials with field recordings, archival fragments, instrumental sources, and live performance, the project creates hybrid sonic ecologies that challenge assumptions about authenticity, authorship, and the nature of listening.

​​

​​

​

The project arises within a lineage of sonic practices that includes:

• Musique concrète, which reoriented sound around the acousmatic and the recorded.

• Electroacoustic and acousmatic composition, exploring sound as abstract material.

• Post-digital aesthetics, treating artefact, glitch, and mediation as compositional tools.

• Contemporary AI art—a rapidly evolving field often dominated by novelty rather than critical artistic rigor.

 

However, AI-generated audio remains largely unexplored within electroacoustic discourse. Where early computer music pioneers such as Risset, Chowning, and Roads emphasized spectral control and synthesis, LLM-derived sound introduces a different paradigm: one shaped by pattern recognition, probabilistic inference, and the non-transparent internal structures of the model itself.

​

This project aims to situate AI in dialogue with these histories instead of outside them. It treats LLM audio not as an “AI effect” but as a new continuity within the evolution of sonic materiality.

​​

​​

​

A. Ontology & Materiality

• What is the status of a sound object with no physical source?

• How does one categorize a sound that mimics the material world but is not derived from it?

 

B. Listening & Perception

• How does the acousmatic experience shift when the listener knows the sound was never “real”?

• What new listening modes arise when origins are not simply hidden but fundamentally absent?

 

C. Authorship & Ethics

• To what extent does the training corpus of an LLM constitute an uncredited archive?

• How do we navigate cultural entanglement when machine-generated audio encodes patterns from diverse, often unacknowledged sources?

 

D. Method & Creation

• What compositional strategies allow AI-generated materials to function not as gimmicks but as integrated, expressive sonic objects?

• How do “meta-concrete” sounds coexist with recordings, live performers, and spatial audio environments?

​

 

​​

​​

Musique Méta-Concrète proposes that AI-generated sound represents the next evolution of abstract sonic material—one that is not captured but conjured, not recorded but inferred. By extending the concrete tradition into a speculative, computational realm, the project redefines what a sound object can be, and what forms of listening will matter in an era when sound is increasingly unmoored from its physical origins. This project aims not only to create new works but to establish a framework for artists, scholars, and the public to critically engage with a rapidly changing sonic world—one where the material of music is no longer only found in the world, but also in the shifting, latent architectures of the models we build.

​​

​