Music, talent, events, and the platform underneath them. The words are written by people, the voices are synthetic, the artists are characters, and we would rather you heard that from us. How we make records.
Not in the passportLOOPCHILDA case study: the brief, the drafts that failed, and the rules that produced a listenable song about invisible infrastructure. What it cost.
Most AI music fails for one avoidable reason: nobody writes the words. What separates a commissioned song from a generated one, and what it costs.

A bright lead and a room full of her own voice answering.

A voice built to be chopped, over a drop that clears the room first.

A voice that sits lower than male R&B does, answered by its own falsetto.

The eager agent and the flat voice that answers it.

A countertenor who was told to sing lower and declined.