Master versus publishing, completely

The master is the copyright in a recording. The publishing is the copyright in the song that was recorded, and the income that copyright earns. They are two separate works under the Copyright Act — owned separately, registered separately, licensed separately, and collected by different organisations — and one of them is legally a derivative of the other, which is why permission between them only travels in one direction. If you wrote and recorded your own song you hold both. Holding both is not the same as holding one thing.

  • A song is two copyrights: the sound recording, called the master, and the musical work, whose rights and income are called the publishing. The Copyright Office states that a copyright in a sound recording “is not the same as, or a substitute for” copyright in the composition.
  • The recording is a derivative work of the composition, so every recording depends on a permission from the song’s owner and no song depends on any recording of it. Somebody who records their own version of your song owns that recording and no part of your song.
  • The two sides register in different places. One Copyright Office application may cover both only when they sit on the same phonorecord and the claimant named for both works is the same party.
  • One song carries three sets of percentages — the master split, the writer share and the publisher share. They answer three different questions and no two of them are the same number.
  • The two copyrights do not always run on the same clock. In Canada a work lasts the author’s life plus seventy years and a sound recording lasts seventy years from first fixation, extended to at most seventy-five from publication.

What is the difference between the master and the publishing?

The master is the copyright in a recording; the publishing is the copyright in the song recorded, and the income that copyright earns.

Two words, one release, and almost nobody is ever handed the sentence that separates them. Here is that sentence, from an office with no commercial reason to keep it vague — the Copyright Office’s own circular on registering the two of them:

Sound recordings and musical compositions are considered two separate works for copyright purposes. Even though a sound recording is a derivative work of the underlying musical composition, a copyright in a sound recording is not the same as, or a substitute for, copyright in the underlying musical composition.

Read the middle clause again, because it is the one that gets skipped and it is the one everything else on this page runs on. The recording is a derivative work of the song. Not a sibling, not a partner, not the same asset seen from a different desk. One is built on top of the other, and that is a fact about direction rather than about importance.

Which produces the map. Every row below is a section of this page.

The masterThe publishing
What it isThe sound recording: one particular fixed performanceThe musical work: the song itself, independent of any recording of it
What people call itThe master, your masters, master rights, the recordingThe publishing, the song, the composition, the work
Who has to agree before somebody uses itWhoever owns that recordingWhoever owns that song
Where it is registeredCopyright Office as a sound recording; then a distributor, and SoundExchangeCopyright Office as a musical work; then a PRO, and The MLC or CMRRA
Who pays you for itYour distributor, and SoundExchange for US non-interactive playsYour PRO, and the mechanical collective in each territory
How long it lasts, USLife of the author plus 70 years — unless that side is a work made for hire, and then 95 years from publication or 120 from creation
How long it lasts, Canada70 years from first fixation, at most 75 from publicationLife of the author plus 70 years

One row is deliberately missing from that table: who authors each side. It is missing because it is the hardest question in this whole subject and it has two pages of its own — who authors a sound recording, and who counts as a songwriter. This page takes both answers as given.

What the table also cannot show is how strange the arithmetic gets. Two artists can each own their own recording of the same song and neither of them owns the song. One person can own a song and not a single recording of it. And you, having written and recorded your own record, own both — which feels exactly like owning one thing, right up until the first time somebody asks you to license half of it.

If somebody covers your song, what do they own?

They own the recording they made and no part of your song, and their arrangement takes no copyright without the song owner’s express consent.

This is the cleanest demonstration that the two copyrights are genuinely different things, because it is where a reader can watch one of them move and the other one stay put. Somebody records their own version of your song. At the end of it there is a new sound recording in the world, and they own it outright — it is their performance, their session, their fixation. There is still exactly one musical work, and it is still yours.

They do not get a piece of the song for their trouble, and the statute is unusually direct about it. Section 115 gives a cover artist the privilege of arranging the song:

A compulsory license includes the privilege of making a musical arrangement of the work to the extent necessary to conform it to the style or manner of interpretation of the performance involved, but the arrangement shall not change the basic melody or fundamental character of the work, and shall not be subject to protection as a derivative work under this title, except with the express consent of the copyright owner.

Two limits in one sentence. They can play it their way, within reason. They do not get a copyright in their way. The arrangement they spent three weeks on is protected only if you say so in as many words, and nobody has ever accidentally said so.

The route they used to make that recording — the compulsory licence itself, what it costs, and where the money for it goes — is a piece of its own, and the map of who collects what covers it end to end. What matters here is the direction. Your song can go and make a record with somebody else, and then another, and then twenty more, and each of those recordings is a separate copyright belonging to whoever made it. Your recording carries exactly one underlying song, forever, and cannot be detached from it by anyone at any price.

That direction is the derivative-work clause doing its work, and it is the whole reason what the recording’s owner can refuse is a shorter list than people expect.

One thing this page does not settle, said plainly rather than papered over: section 115 is a United States statute and its own text is scoped to phonorecords distributed in the United States. Whether Canada offers an equivalent route to record somebody else’s song without asking them was not established here from a primary source, so nothing above claims that it does or that it does not.

Who owns each one on a record you released yourself?

The song’s authors own the song and the recording’s authors own the recording, and on most self-released records those are two different lists of people.

Take the last thing you put out and write the two lists. The Copyright Office says who goes on each, in one paragraph of the same circular:

The author of a sound recording is the performer(s) featured in the recording and/or the producer(s) who captured and/or manipulated or edited the sounds that appear in the final recording. The author of a musical composition is the person who created the music and/or lyrics.

Two sentences, two rosters, and they overlap far less than anyone expects. The co-writer who sent a verse over WhatsApp and never came to the studio is on the second list and not the first. The session drummer who played the whole record and wrote none of it is on the first list and not the second. The producer who did both is on both lists, in two different capacities, for two different reasons, and will eventually be paid twice by two different organisations.

The Office’s own worked example is better than anything we could invent, and it is sitting in the group-registration guidance where nobody reads it. Three songs, written by two people, each of them recorded by one of the two:

Three unpublished compositions were written by Jim and Dwight: Composition 1 was performed by Jim; Composition 2 was performed by Jim; Composition 3 was performed by Dwight. Jim and Dwight may register the three unpublished compositions, but not the three sound recordings, on one group application.

One application for the three songs. Not one for the three recordings. Same two people, same three tracks, and the paperwork splits down the middle because the ownership already did.

Whether a given collaborator is an author at all is a test rather than a courtesy, and there is a different one on each side: what makes somebody a songwriter, and what makes somebody an author of the recording. If you paid a producer and put nothing on paper, what the money actually bought is more specific and less comfortable than the invoice suggests.

Write the two lists now, while everyone still likes each other. The moment you need them is the moment somebody else is writing them for you.

Why does one song carry three different sets of percentages?

A released song carries a master split, a writer share and a publisher share, and those three numbers answer three different questions.

Somebody in that room said “fifty-fifty” and everybody nodded at a different number.

Here is what they were each nodding at.

  • The master split divides the recording’s revenue among the people the recording’s owners agreed to divide it among. It has nothing to do with who wrote anything.
  • The writer share divides authorship of the musical work between the writers, conventionally out of 100.
  • The publisher share is the collection side of that same musical work — the half a publisher administers, which is why societies can count one song at 200%.

Those three do not add up, because they are not addends. A producer holding three points on the master and a twenty per cent writer share does not hold twenty-three per cent of anything: they hold two positions in two separate registers, and the only thing the two numbers have in common is a song title. Add them and you get a number that describes nothing that exists.

Each of the three has a page. The arithmetic of a split that will not divide, and what each society’s form actually accepts, is the rounding piece; both share conventions are explained in the boxes you would be filling in on the template; a point is one per cent of a number with three possible bases; and which document allocates which position is a list of eight of them.

The habit worth building out of all that is smaller than it sounds. Never write a percentage down without writing which of the three it is.

Where does each copyright get registered, and with whom?

Each copyright registers separately with the US Copyright Office, and each side is then claimed again with a different set of organisations that actually pay.

Two layers, and people conflate them constantly. The Copyright Office layer is about proving what you own. The society layer is about being paid for it. Neither one does the other’s job.

Start with the Office, where the rule is stated with its exceptions attached:

Since a musical composition and a sound recording are distinct works, separate registration applications generally should be submitted for each work. However, you may use one standard application to register a sound recording and an underlying musical composition when (1) the musical composition and sound recording are embodied in the same phonorecord and (2) the claimant for both the musical composition and sound recording are the same.

Condition two is the test, and the Office restates it later as a flat prohibition: a sound recording and a musical composition “can only be registered together if the same person or entity is named as the copyright claimant for both works.” The moment those two claimants differ — the moment you signed one side to somebody and kept the other — the single application is gone and there are two filings, forever.

There is a narrower door behind that one. The Single Application covers one recording and its underlying composition on three conditions: the composition and the sound recording are embodied in the same phonorecord, the author is the only performer featured in the recording, and the author is the only copyright owner of both works. That is the solo artist’s door, and it closes the day a featured guest appears on the track.

Now the layer that pays, one line each, because each of these has a page. The recording’s money reaches you through the distributor account you already have, and through SoundExchange for US non-interactive plays. The song’s money reaches you through a performing rights organisation and, for US mechanicals, The MLC — or SOCAN and CMRRA in Canada. Which collector holds which stream is a map of its own, and why a distributor statement is one door of two is where the shortfall gets explained.

Which leaves the shape of the failure, and it is close to universal. You registered the recording side on the day you released, because setting up a distributor is registering the recording side. Nobody handed you the other form. The recording side has been claimed since day one and the song side has been sitting there unclaimed ever since, quietly, on every release you have ever put out.

Do the two copyrights last the same length of time?

In the United States both run for the author’s life plus seventy years, unless one of them is a work made for hire.

Section 302 sets one clock for work created from 1978 onward: copyright “endures for a term consisting of the life of the author and 70 years after the author’s death,” and for a joint work it runs from the death of the last surviving author. On a record you wrote and recorded yourself, both halves are on that clock and both expire together.

Then subsection (c) changes the instrument entirely. For an anonymous work, a pseudonymous work, or a work made for hire, copyright “endures for a term of 95 years from the year of its first publication, or a term of 120 years from the year of its creation, whichever expires first.” No life in it anywhere. A fixed count from a date.

So sign one of your two copyrights away under a work-for-hire clause that actually works, and that half switches clocks while the other half stays on yours. One song, two expiry dates, decided decades earlier by a paragraph somebody dropped into a contract. Whether a grant can be undone before its clock runs out, and by when the notice has to be filed, is the recovery mechanism’s own page.

Does Canada divide them the same way?

Canada divides the same two copyrights, gives the recording to a different party by default, and runs a different clock on each.

The division holds. The defaults and the clocks do not, and the second of those is the one nobody mentions.

On the song side the default is the one you would expect: section 13(1) says that “the author of a work shall be the first owner of the copyright therein.” On the recording side Canada answers the other way round, giving the copyright to the maker rather than to the performers, and what that reversal does to a producer who was paid is worked out where it belongs rather than here.

The clocks are the part worth carrying away. Section 6 puts a work on the familiar term — “the life of the author, the remainder of the calendar year in which the author dies, and a period of 70 years following the end of that calendar year.” Section 23(1.1) puts the recording on something else entirely:

copyright in a sound recording subsists until the end of 70 years after the end of the calendar year in which the first fixation of the sound recording occurs. However, if the sound recording is published before the copyright expires, the copyright continues until the earlier of the end of 75 years after the end of the calendar year in which the first publication of the sound recording occurs and the end of 100 years after the end of the calendar year in which that first fixation occurs.

Put those two next to each other on one release. The song runs on your life. The recording runs on a date stamped on a hard drive, and it ends seventy-five years after you put it out at the very outside. In Canada the two halves of one song are built to expire at different times — on a record where you wrote everything, played everything, and never signed a thing.

What do you do about the half you have never claimed?

Find out which of the two has your name on a register, because on a self-released catalogue the missing side is almost always the song.

Open your distributor dashboard. That is the recording side, and it is working — the money may be small, but the plumbing is connected and something arrives.

Now name the place your song is registered. Not the recording of it. The song. If that takes longer than four seconds, you have found the half.

A distributor account registers the recording, nothing in that flow registers the musical work, and no society pays a writer it holds no record of. Nobody took your publishing. Esau at least got a bowl of stew.

Four things close it, in this order. Affiliate as a writer with a performing rights organisation. Register the works themselves with the mechanical collective for your territory. Name a publisher for each work, which can be you, because the publisher share is a job somebody has to be named to rather than a person who has to exist. Then do all three again for everything already out, one release at a time — the sweep has a checklist.

Two copyrights. You have been paid for one of them.

FAQ

If I wrote, performed and recorded the song by myself, do I still have two copyrights?

Yes. Writing a song and recording it produces a musical work and a sound recording, and one person holding both does not merge them into one asset. The Copyright Office runs a Single Application for exactly this case, on three stated conditions: the composition and the sound recording are embodied in the same phonorecord, the author is the only performer featured in the recording, and the author is the only copyright owner of both works. One filing, two works — and the day a featured guest appears on the record, the third condition fails and there are two filings again.

My co-writer did not play on the record. What do they own?

A share of the musical work and none of the sound recording. The Copyright Office names the authors of a sound recording as the performers featured in it and the producers who captured, manipulated or edited the sounds in the final recording, and a writer who was never in the room is not among them. That is why the split sheet and the master splits are two documents with two different lists of names, and why a co-writer holding half the song can hold nothing at all on the record of it.

My contract says “masters.” Does that include the songs?

The word names sound recordings, so a document that speaks only about masters is a document about one of your two copyrights. That is the starting point and not the answer: contracts define their own terms, and some define “masters” far more broadly than the word does standing alone, so what the document reaches depends on what else is written in it. This is the sentence to put in front of somebody who reads these for a living. Which instrument is doing which job is covered in the comparison of the three, and counsel starts cheaper than you think.

Do I have to register either copyright with the Copyright Office to own it?

No. The Copyright Office states that a work is under copyright protection the moment it is created and fixed in a tangible form, and that registration is in general voluntary. What registration changes is enforcement: the Office also states that you will have to register if you wish to bring a lawsuit for infringement of a US work. Owning a copyright and being able to sue over it are two different questions, and only the second one has a form attached.

Sources

  • US Copyright Office, Copyright Registration of Musical Compositions and Sound Recordings, Circular 56A — page 1: that sound recordings and musical compositions are two separate works for copyright purposes, that a sound recording is a derivative work of the underlying composition, and that a copyright in one is not the same as, or a substitute for, copyright in the other.
  • US Copyright Office, Circular 56A, page 2 — that a registration for a composition does not cover a recorded performance of it and a registration for a recording does not cover the composition; that separate applications generally should be submitted; and the two conditions under which one standard application may cover both.
  • US Copyright Office, Circular 56A, page 3 — the Single Application’s three conditions: the same phonorecord, the author as the only featured performer, and the author as the only copyright owner of both works.
  • US Copyright Office, Circular 56A, page 4 — who the author of each work is, and that the two may be registered together only where the same person or entity is named as copyright claimant for both.
  • US Copyright Office, Circular 56A, page 6 — the group-registration example of two writers and three recordings, where the compositions may go on one application and the sound recordings may not.
  • US Copyright Act, 17 U.S.C. § 115(a)(2) — that the compulsory licence carries a privilege to arrange the work to conform it to the performer’s style, that the arrangement may not change the basic melody or fundamental character of the work, and that it takes no protection as a derivative work without the copyright owner’s express consent. The compulsory licence itself, in both directions, is covered in the royalty-streams piece rather than here.
  • US Copyright Act, 17 U.S.C. § 302 — (a) life of the author plus 70 years for work created from 1978; (b) the last surviving author’s death for a joint work; (c) 95 years from first publication or 120 from creation, whichever expires first, for an anonymous or pseudonymous work or a work made for hire.
  • Copyright Act (Canada), section 13(1) — that the author of a work is the first owner of the copyright in it. Consolidation current to 2026-06-21.
  • Copyright Act (Canada), section 6 — the term of copyright in a work: the life of the author, the remainder of that calendar year, and 70 years after it. Consolidation current to 2026-06-21.
  • Copyright Act (Canada), section 23(1.1) — the term of copyright in a sound recording: 70 years from first fixation, and where published before expiry, the earlier of 75 years from first publication and 100 years from first fixation. Consolidation current to 2026-06-21.
  • US Copyright Office, Copyright in General — that a work is protected the moment it is created and fixed in a tangible form, that registration is in general voluntary, and that registration is required before bringing an infringement suit over a US work.

Every page cited here was read on the date at the top of this piece, and Circular 56A was read from the Office’s own PDF rather than from a summary of it. One limit worth stating rather than leaving you to find: that PDF’s body text extracts cleanly and its small-capital numerals do not, so nothing above quotes a circular number or a revision date from inside it. This is not legal advice. It describes US and Canadian law and no other, and a contract that touches either copyright is worth an hour of a music lawyer’s time before you sign it rather than after.

Keeping the register

Everything above is one filing problem wearing two hats, and the hats are the part software usually gets wrong. CatalogTracker keeps master splits and publishing splits per track, separately, each summing to its own 100, and generates them as two different documents. The part that matters for this page sits further down: a collaborator is a Party with roles rather than a name on a track, so the same person can hold a position on the recording side, on the song side, or on both, in whichever capacity they actually held. In development for iPhone.