CORPUS is a licensing and royalty protocol for music in the age of generative AI. Built by Sofilab GmbH in Munich and co-financed by Creative Europe, it lets musicians contribute works to a shared, rights-compliant training corpus — and participate economically in the value those models create. Contributors retain their rights; works are opt-in only. Training data never leaves CORPUS infrastructure; licensees submit model architectures to secure environments and export only the trained weights.

This journal is where we think out loud about that work. We publish essays, reports, arguments, and open questions on how music enters AI — and on whose terms. Five editorial postures shape what gets written here: Building (dispatches from inside CORPUS), Mapping (surveying the terrain), Questioning (taking apart received ideas), Proposing (new frameworks), and Making (from practitioners on their own craft). Contributions come from the CORPUS team and from invited guest authors: composers, researchers, industry lawyers, and engineers whose work touches the questions we care about.

In 2026 we are moving the contribution platform from closed alpha to public beta. REEF, our real-time adaptive model for interactive sound, is in active development. The legal form of CRPS (Corpus Participation Rights) is under evaluation. A foundation structure is being designed. We expect more of the journal to document those transitions as they happen.

If you have an idea for an article, would like to be interviewed, or want to contribute your own writing, get in touch.

The CORPUS team
The CORPUS team, from left to right: Luciano Pinna, Lawrence Hoggs, Yannis Vasilakis, Max Graf, Anja Gerscher, Lars Ullrich, Mathis Nitschke, Elsa Büsing, Jacob Andersen. Photo: Ingolf Hatz