Music licensing 101: what developers and platform builders need to know

Building a music-enabled app or platform? You’ll meet music licensing before you ship your first feature. Every song carries at least two copyrights, usually owned by different parties — and using one without the right licenses can stall a launch.

This guide breaks down music licensing for developers — no legal jargon, just the practical knowledge you need to build compliant music features. It’s an orientation, not legal advice: for anything binding, talk to an entertainment lawyer.

The two-part problem: composition and recording

Every song exists as two separate copyrighted works, and understanding the split is fundamental.

The composition is the underlying musical work — melody, lyrics, chord progressions. Think of it as the sheet-music version of a song.

The sound recording (the “master”) is a specific recorded performance of that composition. The same song can have many different recordings.

Different entities typically own these rights. A songwriter (and their publisher) holds the composition; a record label often holds the master of the commercial release, while an artist who re-records or self-releases may own a master themselves. The practical upshot: using a single song often requires permission from more than one party.

Performance rights: when music gets played

Performance rights govern the public playing of music — streaming, radio broadcasts, live venues, and background music in apps.

Performance Rights Organizations (PROs) such as ASCAP, BMI, and SESAC in the US license music on behalf of songwriters and publishers, then distribute royalties based on usage. They typically offer blanket licenses covering their catalogs, with rates that vary by usage type and audience size.

For developers, performance rights apply when your app plays music publicly — a fitness app streaming background music, or an app playing copyrighted intro music, both need performance licenses.

Synchronization licenses: music meets media

Sync licenses cover music synchronized with visual content: video uploads with background music, app trailers with soundtracks, in-game music.

Unlike performance rights, sync licensing happens directly between you and the rights holders — there’s no central clearinghouse. You negotiate master sync rights with the recording owner (often a label) and composition sync rights with the publisher. Costs vary widely with the song’s popularity, the duration and prominence of use, the territory, exclusivity, and whether the use is commercial.

Mechanical rights: the reproduction license

Mechanical rights cover reproducing and distributing a composition. When a streaming service plays a song, it owes mechanical royalties to the songwriter and publisher.

The Mechanical Licensing Collective (MLC) administers blanket mechanical licensing for eligible digital services in the US, collecting and distributing those royalties based on usage. Statutory mechanical rates are set through a regulatory process and change over time, so confirm the current rate with the MLC or a licensing professional rather than hard-coding a number.

Master recording rights: the other half

Master recording rights belong to whoever owns the actual recorded performance — typically a record label, though independent and self-releasing artists often retain theirs. To stream, download, or reproduce a specific recording you need permission from the master rights holder, separate from the composition rights above. The major labels control most commercial recordings; independents and self-released artists control the rest, and each is a separate negotiation.

Digital and platform-specific licensing

Digital distribution created additional categories:

Interactive (on-demand) streaming requires both mechanical and master recording licenses.
Non-interactive (radio-style) streaming requires performance and master recording licenses but not mechanicals.
User-generated content (UGC) platforms need bespoke agreements covering user uploads; major UGC platforms negotiate direct deals with labels and publishers.

Common developer scenarios

Music recognition and identification apps

Apps that identify songs generally don’t need licenses for the recognition step itself: recognition returns metadata, not audio. A music recognition API like AudD identifies a song and hands you structured data (artist, title, album, label, and — on the Startup plan and higher — ISRC and UPC). You only take on licensing obligations if your app then plays the song or distributes audio.

Podcast and audio platforms

Podcast apps face layered requirements: intro/outro and background music can require sync (and performance) licenses, and user-uploaded episodes containing copyrighted music need UGC agreements. Recognition helps here by identifying what’s in an episode so you can attribute or flag it.

Platforms that let users upload music-bearing content need comprehensive coverage: direct deals with labels for masters, agreements with publishers for compositions, and performance and sync licenses for playback and video. Recognition is the detection layer that tells you which copyrighted work appeared so the right policy can apply.

Gaming and interactive media

Games with licensed music typically need sync licenses for in-game music, master licenses for specific recordings, and composition licenses from publishers — plus performance licenses if the game streams music.

Rights detection and compliance

Building music features means detecting rights. When users upload content, you need to identify copyrighted material and handle it appropriately.

Content-ID-style systems scan uploads against reference databases. Music recognition APIs are the identification primitive underneath: they return song metadata so you can block unauthorized uploads, share monetization, attribute correctly, or generate usage reports for rights holders.

This is exactly where AudD fits. It identifies songs in audio without requiring a heavy enterprise integration:

from audd import AudD

# get your own token at dashboard.audd.io; "test" is capped at 10 requests/day
audd = AudD("test")

# Identify every song across a user's upload, with timestamps.
matches = audd.recognize_enterprise(user_upload, limit=10)

for m in matches:
    # On the Startup plan and higher, m.isrc / m.upc let you cross-check
    # against your licensing systems.
    print(m.timecode, m.artist, "—", m.title, m.label)

The enterprise endpoint chunks long audio or video and returns every match with timestamps (billed per 12 seconds), which is what a UGC scanner needs: one clip can contain several copyrighted works at different points. Recognition tells you what is present and where; the licensing decision is still yours to make.

A short clip from the microphone or a single file uses the standard endpoint instead, returning the one best match. Either way, recognition returns metadata only — it doesn’t grant any right to play or distribute the audio it identifies.

Working with rights organizations

Performance rights — contact ASCAP, BMI, and SESAC directly; they offer standardized licensing for most use cases.
Sync rights — identify rights holders (publisher and label) and negotiate directly; there’s no single clearinghouse.
Mechanical rights — the MLC administers blanket mechanical licensing for eligible US digital services; international mechanicals are handled territory by territory.
Master recordings — contact labels directly, or work through licensing agencies for independent catalogs.

International considerations

Licensing varies significantly by country, and a US license doesn’t cover European or Asian markets. Each territory has its own PROs (for example PRS in the UK, GEMA in Germany, JASRAC in Japan), its own copyright terms, and its own mechanical and sync structures. Plan for international expansion early; multi-territory deals take time.

Building compliant music features

Start by defining your exact use case: are you streaming full songs, playing short clips, or just identifying music? Each scenario needs different licenses.

Practices that hold up:

Implement content identification early, not as a bolt-on.
Build usage tracking into your architecture so you can report accurately.
Design for compliance from day one, and plan for international expansion.
Consult an entertainment lawyer for anything binding, and understand DMCA safe-harbor provisions where they apply.

The role of music recognition in compliance

Accurate identification underpins everything downstream. When you can reliably identify copyrighted content, you can generate precise usage reports, automate policy, detect unauthorized uploads, and attach proper attribution and metadata.

Tools like AudD’s recognition API give developers that identification layer — across a public catalog of 160 million songs — without complex enterprise onboarding, so compliance work doesn’t slow feature development.

Getting started

Define your use case, document how users interact with music in your app, and let that determine which licenses you need. Contact the relevant PROs for performance licenses (usually the most accessible), then approach publishers and labels for sync and master rights; independents are often easier to start with than majors. For the identification layer, get a token at dashboard.audd.io and read the API reference.

Related