A Perfect Format for a Program That No Longer Exists
I sat through the AVR+ presentation, and it was, as these things always are, immaculate on paper. On June 18, CISAC published the format and called it a turning point: a machine-readable JSON schema, more than six years in the making, the first working implementation of the Global Cue Sheet Standard 2.0. The engineering is real, and I do not quarrel with it. The schema draws its data elements from every format that ever mattered — the old AVR, RapidCue, Soundmouse, and the society formats of ASCAP, BMI, GEMA, KODA, PRS, SOCAN — and folds them into a single grammar. When the standard went to 2.0 in September 2024, recording metadata entered the cue sheet for the first time, which is how IFPI ended up in a CISAC room it had never bothered to sit in before: neighbouring rights now ride the same document as authors’ rights. Sylvain Piat says the format “has the potential to” — the conditional carrying more weight than he means it to — “massively increase the efficiency and timely processing of usages.” Jens Kindermann of GEMA calls it “a common language.” Credit where it is due. Better grammar beats worse grammar.
But a cue sheet is not a document. It is arithmetic. Every second of music under a program is a slice of a pool, and the durations are the payment. Which means the cue sheet carries a dependency the schema never names: it has to describe a broadcast that actually happened, exactly as it happened. That is the part no format can hold still.
Seconds times a weight
In Brazil, the arithmetic is not a metaphor; it is written into the regulation. Under ECAD’s Regulamento de Distribuição, each audiovisual work’s royalty is its share divided by the total musical duration, weighted by how each cue is used — and the weights are fixed on a twelfths scale: an opening or closing theme scores 12/12, a character theme 8/12, a live performance 6/12, a block theme 4/12, other works 2/12, and background music 1/12. The same table governs cinema, open TV, pay-TV and streaming video alike. Payment is seconds times that weight, and nothing softens the edges.
Now re-cut the program. A Turkish soap opera — we have had this problem precisely recently — is licensed in and reassembled for the local broadcast, and ECAD requires the broadcaster to report what actually aired, “fidedigna ao que foi tocado,” on a separate form reserved for foreign series. Scenes are trimmed, blocks resequenced. The origin cue sheet, however clean its JSON, now describes a program that no longer exists on the screen. Trim the seconds, and the payment falls. Recut a character theme into background wallpaper, and the same notes drop from 8/12 to 1/12 — an eightfold cut, executed by a local editor at the network who never saw the cue sheet and was not thinking about royalties. AVR+ will transmit whichever arithmetic it is handed, flawlessly. A perfect schema for the wrong edit is a perfect description of a payment that will be wrong.
Nothing in the standard reconciles the origin sheet to the aired cut. That reconciliation is a separate act, and it is done by hand.
And nobody has to file it
Here is the fact that ends the efficiency story before it starts. In Brazil, submitting the cue sheet is not mandatory. No regulation compels it — the União Brasileira de Compositores says so plainly, and calls this vital link one that “muitas vezes não é cumprido.” The document at the base of the entire structure, the one AVR+ exists to standardise, is optional, and its absence is an ordinary, documented cause of authors going unpaid.
When it does arrive, watch how it clears. The broadcasters pay ECAD and consider the matter closed — as far as a network is concerned, making the cue sheet work is ECAD’s job, bought and paid for. The societies sit closest to the repertoire but cannot build the sheet alone; they hold only fragments, and they help assemble what they can. Validation belongs to ECAD, which checks the sheet against its own secondary “listening” system, comparing paper to what went to air — and sometimes, when the show is foreign, against Wikipedia. I am not being cute. Wikipedia. That is the confirmation layer for what ran under a program made abroad. You cannot JSON your way out of a re-edit, and you cannot standardise a document nobody is required to send.
Born in Paris, adopted in Medellín — or not
Then there is the adoption problem, which is not technical. AVR+ asks producers to emit structured JSON. GEMA and the majors have the engineers, and they have the people who fly to the plenaries where these standards are born. But a standard is not adopted in a room in Paris. It is adopted, or it is not, by the second tier — the Colombian urbano label that has to know this format exists, understand it, and reach the systems that speak it, or the whole thing stays theory. That is exactly the layer where this kind of information does not travel. Even DDEX, the framework the industry likes to say it already runs on, is not shared knowledge one floor below the majors. A standard that the well-connected can implement, and no one else can reach, is not a standard. It is a moat with a press release.
The stakes are not theoretical. CISAC’s own study says AI could take a quarter of creators’ revenue by 2028. In May, BMI bought Soundmouse from Orfium to build the largest cue sheet database in the world — nobody spends that on a shrinking sector. The money is stampeding into audiovisual identification at the precise moment the identification still depends, at the far end of the chain, on a human, an ear, and an encyclopedia anyone can edit.
Duration is the money, and the money moves a second at a time. AVR+ counts beautifully. It still cannot tell you what it is counting.