Sync Licensing 2026: How AI Music Gets Placed in Film and TV
Sync licensing is the business of putting music underneath pictures — film, television, advertising, games, trailers, branded social. It pays per placement rather than per play, and it runs on paperwork rather than playlists. This page covers how a sync licence is actually cleared, what the routes to a placement pay, and the part that decides whether AI-generated music can play in this market at all: provable ownership and a warranty you can honestly sign.
- Every sync placement needs TWO clearances, not one: the master (the recording) and the publishing (the composition). Beginners quote a fee for one and discover the buyer cannot use the track without the other.
- The buyer's risk is a lawsuit, not a takedown. That is why the contract asks you to warrant ownership and indemnify them — and why an indemnity you cannot honestly give is a contract you should not sign.
- The market is genuinely split on AI. Mid-market and long-tail libraries place AI-assisted music routinely; a large share of buyers refuse fully AI-generated tracks outright, and ASCAP, BMI and SOCAN register partially-AI works but not fully-AI ones.
- The structural advantage of a solo AI musician is boring and real: one owner, both sides of the copyright, no co-writer splits to chase. Clean paperwork closes deals that better music with murky splits cannot.
- Clearing artifacts is release prep and is not a substitute for disclosing AI use where a contract asks. They are different obligations, and conflating them is how people get sued.
Sync licensing is what happens when someone puts your music underneath pictures: a scene in a drama, a thirty-second regional ad, a game's menu loop, a trailer's second-act build, a brand's Instagram cut. It pays a negotiated fee per use rather than fractions of a cent per play, and it is a paperwork business before it is a taste business.
That last part is why sync licensing gets its own page here, separate from our AI music income playbook, which covers streaming royalties and catalogue economics. Sync runs on a different mechanism. A supervisor is not deciding whether they like your track — they are deciding whether they can prove, to a broadcaster's legal team, that using it will not cost anyone their job.
And that makes sync the one market where the provenance of AI-generated music has hard commercial consequences. Streaming platforms tag it. Sync buyers price the risk of it, and some refuse it outright.
A sync licence is two licences, and most people clear one
The single most common beginner mistake in this market is quoting a fee for a track you do not fully control. A piece of music is two separate copyrights, and a production needs both.
The master is the specific recording — the audio file itself. If you generated it, mixed it and released it under your own name, you control it.
The publishing is the underlying composition: melody, harmony, structure, lyrics. It belongs to the writers, and to their publisher if they have one.
A sync licence clears the master. A separate synchronisation licence clears the composition. Both must be signed before a single frame goes to air, and the two sides are usually quoted at matching amounts under a most-favoured-nations clause, so neither party can be paid more than the other. A supervisor who gets one and not the other has an unusable track and a wasted week.
This is where a co-written song quietly dies. Three writers, no split sheet, one of them unreachable, and the placement goes to whoever answers the email with clean paperwork instead. Deals are lost to admin far more often than to quality.
What the routes to a placement actually pay
There are four realistic ways music reaches a screen. They differ on exclusivity, on who takes what, and on how much of the work you do yourself.
| Route | Exclusivity | Typical split | Indicative fee | What you trade |
|---|---|---|---|---|
| Production-music library | Often exclusive per track | 50/50 master and publishing | $100–$2,000 per placement | Control of the track, sometimes permanently |
| Marketplace / royalty-free | Non-exclusive | Platform keeps 30–50% | $20–$500 per licence | Fee level, in exchange for volume |
| Sync agency / pitch service | Usually non-exclusive, term-limited | 20–50% commission on placements | $500–$15,000+ | Commission, in exchange for access |
| Direct pitching | None | 100% to you | Anything from $0 to five figures | Your time, entirely |
Two honest caveats on those numbers. First, they are ranges observed across published rate cards, library terms and creator reporting — not a schedule anyone will honour. Sync fees are negotiated against a production's music budget, and the same track can be worth $150 to a YouTube channel and $12,000 to a national campaign. Second, the top tier is largely closed to unrepresented catalogue regardless of what the music sounds like, because a network campaign will not take a chain-of-title risk on a stranger.
The route that suits most people starting out is non-exclusive: marketplaces and libraries that let the same track earn in several places at once. Exclusivity is worth signing when someone is actively pitching your catalogue and can show you placements, not merely a bigger headline percentage.
The warranty is the actual product
Here is the thing that reframes everything else on this page. When a broadcaster licenses music, their downside is not a takedown notice. It is a lawsuit, a pulled campaign, a re-edit of finished picture and a very expensive apology.
So the contract protects them, through two clauses that appear in nearly every sync agreement:
- A warranty of ownership. You state that you own or control both the master and the publishing, that the work is original, and that it contains no uncleared third-party material.
- An indemnity. You agree to cover their losses if the warranty turns out to be false.
Those clauses are the product. The music is what gets you into the conversation; the warranty is what closes it. And they have a direct implication for anyone working with AI generators: an indemnity you cannot honestly give is a contract you should not sign. If you do not know what your model was trained on, that is a risk you are personally absorbing, not a technicality you can leave to the library.
The practical version is to keep a paper trail per track before you pitch anything. Generation date and tool, the licence tier you generated under and what commercial rights it grants, any human contribution you made, your split sheet if anyone else touched it, and your ISRC for the recording. Our ISRC code guide covers that identifier and why cue sheets need it.
Where AI-generated music actually stands in sync
The market has not settled, and anyone telling you it has resolved in either direction is selling something. What we can describe accurately is the shape of the split as it stands in 2026.
Buyers are asking, directly. Submission forms and briefs increasingly include an AI-use question, and some go further and ask what the model was trained on. Rights holders who label AI involvement clearly — whether it is full generation, production assistance, or just admin and metadata — get treated as trusted suppliers. Vague answers disqualify tracks instantly, because a supervisor cannot pass a vague answer up to legal.
A substantial share of buyers refuse AI music entirely. Reporting from within the sync trade puts that figure close to half among the buyers who express a preference, and Berklee's In Sync survey of just over a thousand industry participants describes the same unsettled picture from the creator side. That is not a reason to avoid the market. It is a reason to target the half that does not refuse, rather than pitching blind and reading rejections as a verdict on your music.
Mid-market and long-tail placement is genuinely open. Libraries serving YouTube creators, podcasts, corporate video, indie games and social advertising place AI-assisted music routinely. This is the tier where volume, tagging and delivery discipline decide outcomes — and it is a real business, not a consolation prize.
Premium catalogues and trailer houses often exclude it. Explicitly, in their terms. Read them before you upload rather than after you are removed.
The PRO position is specific and worth knowing. ASCAP, BMI and SOCAN have updated their registration policies to accept compositions created partly with AI tools, provided there is human authorship in the mix. Works generated entirely by AI with no human contribution remain ineligible for registration. That matters directly for sync, because television placements generate performance royalties through cue sheets — a back-end that can exceed the upfront fee — and an unregistrable composition collects none of it. Our ASCAP vs BMI comparison covers which society to join.
Copyright registration is a separate question again. The US position is that fully machine-generated output is not registrable while work with meaningful human authorship can be. For a sync deal this matters less than people assume — buyers want usable rights and a credible warranty, not a certificate — but it shapes what you can defend later. Our Suno copyright explainer is the detailed version.
The advantage nobody mentions: you are the only owner
Now the part that runs the other way, and it is not a consolation. The structural weakness of most independent catalogue in sync is fragmented ownership. Three writers with no split sheet. A vocalist who was never contracted. A sample nobody cleared. A producer who now wants a point. Every one of those turns a two-signature clearance into a month of chasing, and supervisors work to picture deadlines that do not move.
A solo musician generating their own tracks under a licence that grants commercial rights has, by default, the cleanest possible position: one owner, both sides of the copyright, one signature, no chasing. For a supervisor under deadline, that is a genuine reason to pick your track over a better one with murky splits.
That advantage is only real if you keep it. Bring in a topline vocalist, paper it. Use a sample, clear it. Collaborate on an arrangement, split-sheet it on the day rather than after a placement offer arrives. Our AI music artist guide covers building the wider artist infrastructure this sits inside.
What libraries want in the file itself
Sync rejections are more often technical than aesthetic, and the technical bar is different from streaming's.
- Catalogue depth over singles. Most libraries want to see 30 to 50 usable tracks. They are buying a supplier, not a song.
- Alternate versions and edits. Full track, instrumental, plus 60-, 30- and 15-second cuts. Buyers cut to length and will take the track that already fits. Stems are a strong advantage where you can produce them — our Suno stems guide covers extraction.
- Loudness for the right medium. Broadcast television is mixed far quieter than streaming — around -24 LUFS against Spotify's -14 and Apple's -16. A track slammed for streaming is not broadcast-ready. Our AI music mastering guide covers the targets.
- Descriptive metadata. Mood, tempo, key, instrumentation, use case, whether it has vocals. Supervisors search by brief. "Track07.wav" is invisible.
- No audible AI artifacts. Metallic overtones, unnaturally even dynamics, smeared transients and odd stereo imaging read as amateur to a trained ear long before anyone reaches the paperwork.
That last point is where sync is stricter than streaming, and it is worth being blunt about. A distributor's classifier is a machine making a probabilistic call. A music supervisor is a person who listens to hundreds of submissions a week and hears an artifact immediately.
Clearing artifacts is release prep, not disclosure
Every generated track leaves its tool carrying passengers: a SynthID-class watermark woven into the signal, a C2PA provenance manifest naming the model, a spectral fingerprint unique to the generator, and secondary layers underneath. They are inaudible by design, and they are what distributor classifiers read.
Undetectr is the first and only AI music watermark remover — the one tool built specifically to remove what distributors scan for, rather than a repair suite adapted to the job. It clears all six layers in one pass, runs in the browser with nothing to install, takes under a minute per track, accepts MP3, WAV and FLAC from Suno, Udio and ElevenLabs Music, and masters to each platform's loudness spec in the same pass. SoundMatch checks for fingerprint collisions before you release, which matters more in sync than anywhere else — a collision surfacing after a campaign airs is a different order of problem from one surfacing on Spotify. It is €39 once for unlimited tracks, with a €19 Starter tier at 10 credits, and outputs for 150+ platforms.
Now the honest framing, because this is the part that gets people into trouble. Clearing artifacts is release prep. It is not a substitute for disclosing AI use where a contract asks. Those are two different obligations:
| Obligation | What it is | Where it applies |
|---|---|---|
| Artifact removal | Technical release prep on the file | Distributor classifiers, upload gates, delivery specs |
| Disclosure | A truthful answer you are contractually bound to give | Library submission forms, sync agreements, AI-use fields |
Removing a watermark does not make a false warranty true. If a library asks whether the music is AI-generated and you say no, you have misrepresented a material fact in a contract that carries an indemnity — and the tool you processed the file with is not a defence. Conflating the two is precisely how people end up sued rather than merely rejected.
The correct sequence is unglamorous: run the file through Undetectr so it is technically clean, answer the AI question honestly, and pitch the libraries that accept AI-assisted material. That combination is a working business. The other combination — clean file, dishonest form — is a liability with a delayed fuse.
Our Undetectr review is the sceptical long-form assessment, and our AI watermark remover benchmark covers the wider category. Both include the caveat that matters here: a remover will not fix a weak song, and it will not get anyone to hear it.
The honest verdict
Sync licensing is the best-paying market available to an AI musician and the least forgiving one on process. Everything that decides an outcome happens before anyone falls in love with your track: two clearances rather than one, a warranty you can stand behind, catalogue depth, edits, metadata, and a file with no audible tells — which is the one item on that list Undetectr handles for you.
The AI question splits the market roughly in half, and that half is not evenly distributed — it is concentrated at the premium end. Work the tier that is open, be straightforwardly honest about how the music was made, and keep your ownership position as simple as it started. One owner, both copyrights, documented. That is a smaller advantage than a hit song and a much more reliable one.
Where to go next: our music distribution services comparison for getting the same catalogue earning on streaming in parallel, AI music for YouTube for the creator-licensing end of the same market, and our methodology for how we test everything on this site.
Questions readers ask.
Sync licensing — short for synchronisation licensing — is permission to combine a piece of music with moving images. That covers film, television, advertising, video games, trailers, corporate video and branded social content. The licence is granted for a defined use: specific media, specific territory, specific term, and sometimes a specific scene. It is a negotiated one-off fee rather than a per-stream royalty, which is why a single placement can be worth more than a year of streaming on the same track.
Because a song is two separate copyrights. The master is the specific recording, usually controlled by the artist or label. The publishing is the underlying composition — the melody, chords and lyrics — controlled by the songwriter and their publisher. A production needs written permission from both, and the two are typically quoted at similar amounts, which is where the industry phrase "MFN", most favoured nations, comes from. If you generated the track yourself and wrote no co-writer into it, you control both sides and can clear them in one signature.
It varies enormously by budget tier and the ranges below are indicative rather than a rate card. Royalty-free library and YouTube-creator use typically lands between $50 and $500. Independent film, documentary and lower-tier TV usually runs from a few hundred to a couple of thousand. Indie game placements commonly sit in the hundreds to low thousands. National advertising and network television are the tier where five figures appears, and that tier is still mostly closed to unknown catalogue with no representation. Treat any single quoted figure with suspicion, including ours.
Some do and some explicitly do not, and the split is commercially meaningful rather than a technicality. Mid-market and long-tail libraries serving YouTube creators, corporate video and podcast production place AI-assisted music routinely. Premium catalogues, trailer houses and agencies working on national campaigns frequently exclude it, and many now ask directly at submission whether AI was used and what the model was trained on. Answer honestly. A library that finds out later removes your whole catalogue, not just the track.
In the United States, fully machine-generated output is not registrable, while a work with meaningful human authorship can be. That is the position to plan around, and it matters less to a sync deal than most people expect: what a buyer needs is the right to use the track and a credible warranty that nobody else will come after them for it. Suno and Udio's paid tiers grant commercial rights to your generations, which is what you are actually licensing on. Our Suno copyright explainer covers the ownership position in detail, and we are not lawyers — for a deal with real money attached, get one.
It is the clause where you promise that you own or control what you are licensing and agree to cover the buyer's costs if that turns out to be untrue. Broadcasters and agencies insist on it because their exposure is a lawsuit and a pulled campaign, not a takedown notice. If you cannot honestly warrant the chain of ownership on a track — an uncleared sample, an unclear co-writer split, a model output whose terms you never read — do not sign. That clause is the entire reason clean, documented ownership beats a better song with messy paperwork.
Build catalogue depth before you approach anyone; most libraries want to see 30 to 50 usable tracks rather than a promising single. Deliver instrumental versions plus 60-second, 30-second and 15-second edits, because buyers cut to length. Tag descriptively — mood, tempo, key, instrumentation, use case — since supervisors search by brief, not by artist. Then pick the route that matches your position: non-exclusive libraries and marketplaces to start, exclusive catalogues or an agency once you have something worth representing.
No, and treating those as the same thing is how people end up in breach of contract. Artifact removal is release prep: it clears watermarks, provenance manifests and generator fingerprints so a file is not rejected on technical grounds. Disclosure is a contractual obligation that exists independently — if a library's submission form or a sync agreement asks whether the music is AI-generated, the honest answer is the only safe answer. A clean file does not make a false warranty true.
The verdict, in one sentence: Undetectr.
Sync buyers hear artifacts before they read your paperwork. Clear the file first — Undetectr removes six artifact layers in one browser pass and masters to spec, €39 once for unlimited tracks — then disclose AI use wherever the contract asks.