Music Expert Witness — Hire a Forensic Musicologist

A music copyright expert witness provides the musical understanding the courts require, and answers questions the courts cannot answer themselves. When two pieces of music are alleged to be the same, what exactly do they share, and is the shared material the kind of thing copyright protects? Judges and juries are not trained to dissect a melody, separate cliché from originality, or weigh a melody in the context of its harmony against centuries of prior art. That is the work. The expert performs the analysis, documents it, and explains it in terms that a non-musician can follow and that an opposing expert cannot easily distort and refute.

The role is somewhat narrow on purpose. The expert describes the musical facts in service of the finder of fact—the court. The musicological expert does not “decide infringement.” Infringement is a legal conclusion for the court and the jury. The expert provides the understanding that leads to the correct finding.

“Thrice is he armed that hath his quarrel just.” — Shakespeare, Henry VI, Part 2
The leading expert tells you whether your quarrel is just, arms you when it is, and tells you straight when it is not.

What the expert actually does

Forensic musicology work is, well, “forensic,” meaning it applies scientific knowledge and techniques to the justice system.

The forensic analysis process begins with building an understanding of the pertinent works, most often through independent transcription. A musicologist writes down the notes, and independence importantly characterizes the whole process. The expert does not accept another’s notation, or a web-service waveform, as ground truth. It’s first principles. The expert transcribes and otherwise directly gains an understanding of the works at issue because it’s the foundation of all else. A mistake early in the process can lay the foundation for poor logic, propagate throughout the analysis, and corrupt the observations and the conclusion.

Such errors, nevertheless, are ordinary and not the exception. Poor transcriptions or correctly transcribed transcriptions that are completely mischaracterized are both pervasive. Why?

The possibilities are too numerous to list. There are omissions, things overlooked, inept mistakes, and sometimes ept ones. And then there are the secondary and tertiary poor observations these produce or promote: rhythms flattened to make two figures line up; a pitch misheard, so an interval changes character; at an even more basic level, two works compared before anyone reconciles their keys. And it follows certainly that an analysis built in any way on someone else’s potentially inherits all of it. Assumptions are fraught things. The expert needs to have a scientific mindset.

So, to the extent I can prevent it, I refuse to be led at all. I conduct a prelitigation analysis in deliberate ignorance of what the client believes they hear and what the other side claims, until my own transcription and comparison are complete. Only then do I read others’ observations, including those of my client, so I can address them and serve the client well. Opposing musicology is not ignored, quite the opposite, but it mustn’t steer. An independent analysis is better served if no one else’s hand reaches the wheel first.

The second failure is the more pernicious — correct observations made to carry weight they cannot bear. A fragment of a few notes, a repeating but mundane figure, a pairing of ordinary elements gets elevated to significance and presented as a smoking gun. A misunderstood vehicle called “selection and arrangement,” applied at a threshold so low or otherwise miscontrued that any two songs might well clear it. (Both “misunderstood” and “vehicle” are doing appropriately damning work there.) The transcription can be accurate, and the characterization can still be wrong.

Calling trivial material trivial, and reserving significance or substantiality for material that earns it, is the heart of the discipline.

After transcription comes comparative analysis. The expert compares melody, harmony, rhythm, and structural elements element by element, identifying precisely which elements overlap and testing each one against the question, “Is this element protectable expression, or is it part of the common vocabulary of music that copyright leaves free for everyone?” Courts refer to this as filtering, in itself an oversimplified and often misapplied process. But kept reasonably straightforward, a descending scale is by itself not an ownable element. A common two or three-chord vamp is not ownable. Common elements are, by definition, not themselves protectable.

An assortment of such commonalities brings us back to “selection and arrangement.” But when a musicologist is considering the case for selection and arrangement, which, properly applied, is completely valid, the protectable material is the specific, original combination of otherwise common elements. The law treats that combination as only thinly protected. The oft-cited precedent (I’m not a lawyer, by the way.) Satava v. Lowry, 323 F.3d 805 (9th Cir. 2003), holds that an arrangement of individually unprotectable elements earns copyright only when the selection and combination are themselves original, and then only against near-identical copying. That’s what the analysis needs to find.

AI presents its own factors. At the time I’m writing this, an AI-created work cannot enjoy copyright protection. The curation itself, that is, the selection of such a work does not constitute authorship even though it has a human element. When the work at issue is AI-assisted, the protectable-expression analysis begins at first identifying which elements the human author contributed. It would be a significant divergence here if we went further into it. But another article on Musicologize, Can I Copyright a Song I Made with AI? covers how the Copyright Office draws that line and what it means for the infringement analysis.

Then comes prior-art and common-source testing. A similarity that looks damning can dissolve when the same figure turns up in songs that predate both works. Well-applied, prior art can be one of the strongest instruments in the work. It arguably cuts in one direction more than the other, and that’s perfectly acceptable. Sometimes that’s just science. If a phrase is common to a genre or traces to a shared earlier source, it is far less protectable, and the claim of its originality and inference of something else having been copied from it weakens or fails. Conversely, a claim can survive this test and come out stronger. But in many cases, when the shared figure turns up in songs that predate both works, the question is nearly settled. Neither party owns what was already common, and the claim narrows or falls. Not finding one does less. It does not prove the figure is distinctive. That alone would not be logical. No one can search all of music, and the absence of a match is not the presence of originality. A claim is logically stronger for surviving a rigorous search. The empty search illustrates better than it decides. A thoughtful, credible analyst says exactly that, rather than dressing a null result as a verdict.

The expert’s output is a written report of appropriate depth and, when the matter reaches it, testimony. The report states what was analyzed, how it was analyzed, and what was found. It maps each conclusion back to a specific finding. It does not overstate. And it makes an earnest attempt to teach, to show, and not merely tell. A good rule of thumb in creating a strong report is that it should read like something a second analyst could replicate. In a sense, that is exactly the standard it will be held to.

When an attorney engages one, and why earlier beats later

A music copyright expert witness is useful to both sides, and the most valuable engagement is almost always the early one.

On the plaintiff side, the first questions should not leap to “how to win.” Begin with the strength of the claim; whether the claim is viable. Viable does not equate to real. A potential plaintiff hears a similarity and feels wronged. No small point. In my experience, potential plaintiffs are never scheming. They’re sincere. The similarity is intrinsically real in some measure at that point. The musicologist expert looks at it extrinsically. An early viability read tells the attorney and the client whether the alleged similarity is musically meaningful and whether it falls within protectable expression or unprotectable convention. Here, the advantages are equally beneficial in both directions; knowledge is power. A clear no, delivered early, is invaluable, and there’s a named “Three C’s” algorithm at Musicologize that stresses delivering such “no” findings to the client clearly, compellingly, and compassionately. I do a client no favors by leading them into battle with a weak foundation. A well-considered “no” saves a client from spending years and a small fortune chasing a claim that the music does not support.

Very obviously, a well-reasoned, clear “yes” established early shapes the complaint and strategy from the start. Ask your lawyer what that’s worth.

On the defense side, the value is mirror-image. A defendant accused of copying needs to know fast whether the accusation has musical substance. Often it does not, and a rigorous analysis showing that the shared material is commonplace discourages the exposure before it grows.

Brands and agencies didn’t write the music (generally), but they get sued anyway. Of course the deep pocket is a named defendant. One can argue that the brand presumably gained the most from the work at issue, so a claim that names them is often rational rather than simply predatory. Suing the party that profited is ordinary.

It can negatively affect things, though. And the expert needs to be professional throughout it:

Unfortunately, sometimes a claim, even beneath the service, becomes somewhat opportunity-based, where the merits are thin, and the target is qualified by its balance sheet, the suit can be reduced to a kind of arbitrage. The nuisance value of settling is set above the cost of bringing a weak case, and the gap is the profit. This play works, if distastefully, because the defendant cannot be sure, quickly, that the music is on its side, and paying to be rid of the doubt looks cheaper than fighting. Musicologist articles are chock-full of my opinion that, in too many cases, more oxygen is afforded than they deserve, and expert-witness forensic musicologists bear some responsibility, along with a court system that does its best but will hopefully improve.

Originality Clearance

When the defendant is a brand, an agency, or a music supervisor, a preemptive or preventive early read is an originality clearance service. Increasingly the fact pattern involves AI-generated material with attribution receipts. It catches a problem before release, when fixing it is cheap, rather than after, when it is a lawsuit. A rigorous read before release removes the musical uncertainty that the aforementioned arbitrage depends on. If the track was checked and the shared material is common, the doubt that made paying rational is gone. A documented read is also evidence of good faith, and that matters most where the exposure is worst. Enhanced statutory damages turn on willful infringement (roughly; still not a lawyer here), a defendant knowing and going ahead anyway, and a clearance opinion on file is hard to reconcile with that. So the same preemptive read that drains the nuisance claim also takes the willfulness exposure off the table. What the lawyers do with that is theirs.

Earlier beats later for a reason that does not change. Before filing, the analysis shapes the decision to proceed. After filing, it is constrained by positions already taken and deadlines already running. The expert who is brought in at the start can prevent the mistake. The expert brought in at the end can, to a considerable extent, mostly describe it.

There is a related distinction worth keeping straight. Copyright infringement is a legal standard. Plagiarism has most of the same components and is often a reputational and commercial problem. The two terms are used interchangeably, but plagiarism is more accurately the moral wrong of copying as opposed to the technically legal wrong. The thresholds may be similar or not, and the same forensic analysis applies to both. The musical facts do not care which forum is worried about them.

Admissibility, and the standard that decides what gets heard

Not everything an expert says gets heard. Testimony has to clear a reliability bar first, and that bar is mostly a demand to think scientifically and logically, the same demand the rest of this work makes.

By now, readers know I am not a lawyer, but the rule is worth stating plainly. So I did some research. Federal Rule of Evidence 702, amended effective December 1, 2023, asks the court to find, by a preponderance of the evidence, that an expert reliably applied a reliable method to sufficient facts before the testimony comes in. The amendment tightened things. The governing cases are Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579 (1993), and Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999), which extended the same scrutiny to non-scientific experts, musicologists included. “Daubert” is, famously, the scrutiny that disqualifies poor musicology — sort of its nickname.

A good test: what clears that bar is a method another analyst could run and reach the same place. Reliability, in practice, is replicability. The works are transcribed from the recordings; the comparison runs element by element; the applicability of prior art is documented well; and the conclusions map back to a finding that someone else can check. Rigor is the bar, not merely credentials and certainly not bravado. The courts that have looked at musicology and tried to find the optimal role, limits, lanes, etc. One might look at Swirsky v. Carey, 376 F.3d 841 (9th Cir. 2004), which treats musicological comparison as the proper method for the extrinsic test. Smith v. Jackson, 84 F.3d 1213, 1218 (9th Cir. 1996), holds that test to external, objective criteria, satisfied by analytic dissection rather than an overall feeling of similarity. Tisi v. Patrick, 97 F. Supp. 2d 539 (S.D.N.Y. 2000), shows the other end of it. The court granted the defense summary judgment because the claimed overlap was a shared key of A major, a common tempo, and a chord progression common to much rock music, all of it running through prior art by other artists. The elements said to be copied were not copyrightable. The pattern is reasonably plain. Analysis that isolates protectable expression survives. A pile of shared commonplaces, however confidently presented, does not.

Musicologize spends a lot of pixels on how the system sometimes fails, but let me say plainly what I think of the system. It is noble. The cases gathered here are a court working to get the question right, and let a qualified expert help it do exactly that. The rules, limits, lanes, etc. are meant to protect that work, not to punish it. But I’ll still complain when it fails.

The musicologists, too. I would like to think that most of us are noble and work in good faith. There is an old razor (I looked this up, too. It’s Hanlon’s) that says, “Never attribute to malice what carelessness explains.” (I softened it a little.) When an analysis goes wrong, I’d rather attribute it to the difficulty of the work itself. So, be grateful when the limits do their job. The discipline is not built to catch ignoble “experts” but to prevent people from fooling themselves, which is the harder and more common problem.

What separates genuine method from courtroom theater

The difference shows up in a handful of failure modes that a competent opposing expert can exploit.

Overclaiming is perhaps the most damaging tell. An expert who says the work was “clearly copied” has stepped past the analysis into a legal conclusion. (Watch me do it now. Bound to happen.) Under-documenting is quieter. An analysis that is thorough but poorly expressed is harder to testify to, and opposing counsel will recast it. Conflating intuition with analysis is the subtle one. A trained ear is useful for deciding what to examine. It is not itself a finding. “Golden ears” is a pejorative. Show, don’t tell. And once again, science and logic: an expert who leads with the conclusion and builds backward to it has, too likely, produced something fragile. The analysis is supposed to test the intuition, not insist upon it.

There are a handful of simple errors that illustrate poor analysis, even typifying it, unfortunately. Confusing major and minor keys, or modal mixture, or borrowed chords, and then assigning function wrongly to illustrate imagined similarities. Or worse, probably: failing to transpose two works to a common key before comparing them. These are basics, but they happen. And an even minimally competent opponent will find them and use them to discredit the rest of the analysis and destroy the claim.

Say the experts on both sides are credible and diligent. An expert who acknowledges the opposing analysis’s strongest point and explains why it is not dispositive is far more credible than one who waves it away. Courts and juries credit the analysis that is clearest and most truth-facing. Theater, on the other hand, is the confident verdict announced as though the method, the evidence, and an earnest desire to make the court understand it were beside the point.

How many coincidences is too many?

The hardest argument to weigh is that of accumulation. An analyst lists a dozen shared traits, a key, a tempo, a rhythmic figure, a cadence, a melodic fragment, and packages them as too many coincidences to be chance. Sometimes that is exactly right. A stack of genuinely unlikely, genuinely independent matches can make compelling evidence. Other times, though, it’s a laundry list of common features dressed up to imitate some concocted improbability. Which one it is depends! How likely that particular stack is to arise on its own, is a question of (sing along with me!) science and logic, and not merely of how many items can be listed.

The reductive instinct is to count. Everyone wants a rule of thumb. “How many notes in a row??” And to a forensic musicologist, it reads as some ages-old story passed around for generations. It probably hasn’t been more than a few months since I heard a genuine music star opine, wrongly, on how many notes, where the line is known or agreed to be drawn. Science, logic, and forensic musicology dictate that no such line can exist. (You saw it here!) At least notes are facts, though! Even with notes as plain facts, the answer depends. The follow-up question, perhaps, should be: how many coincidences? But a coincidence is not a plain fact. Two works sharing a figure is something you can point to. Whether that sharing is a coincidence, and how unlikely a coincidence it is, is a matter of judgment about probability, and it varies from one item to the next. Counting coincidences treats them as equal units. They are not. Coincidence is by degrees.

The law says originality is a low bar. In Feist Publications v. Rural Telephone Service Co., 499 U.S. 340 (1991), the Supreme Court held that originality requires only independent creation plus a modicum of creativity. Clearing it, though, is hardly the measure. It’s just whether you get turned away at the door. Mere originality makes its protectability neither strong nor weak, and not at all specific or focused. All songs have their originality vested more in certain elements than in others. And pertaining to the tactically appealing low bar that Selection and Arrangement sometimes presents, the Court was blunt about mere aggregation. “Common sense tells us that 100 uncopyrightable facts do not magically change their status when gathered together in one place.” Protection for an original combination is “thin.” Thin is not none. It is real protection but with a narrow reach. Copyright is consistent and reasonable, guarding original selection and arrangement. Whether a later work sits inside such a protected arrangement or outside it, among the commonly shared building blocks, is not a line a rule can draw. So, rely on the rule only to the extent that it should matter — as the doorman. It still really turns on how many features align, which ones, how distinctive each is, and how they combine, and, as ever, there’s no number fixed in advance as determinative. Repeating “it depends” is not ambiguity. It acknowledges complexity, and helping the courts in resolving it is a service a competent and honest musicologist provides, where a checklist very often cannot.

Logic pervades. It’s where the accumulation argument really shows its failings. The move from “they share many things” to “the odds of that occuring by chance are tiny, so it was copied” is a known fallacy, and a court named it almost sixty years ago. In People v. Collins, 68 Cal. 2d 319 (1968), an expert multiplied the probabilities of several shared characteristics into a number that made innocence look impossible. Mark Twain usually gets credit for “There are three kinds of lies: lies, damned lies, and statistics.” and it holds. Multiply enough maybes together and you can prove anything is nearly impossible. Here, the California Supreme Court reversed. Mathematics, it warned, “a veritable sorcerer in our computerized society… must not cast a spell over” the trier of fact. The defect was that the traits were never shown to be independent, and multiplying probabilities that are not independent yields “a wholly erroneous and exaggerated result.” Musical traits are rarely independent. A chord progression constrains the melody above it. A groove implies a family of rhythms. Multiplying their separate rarities into a one-in-a-million coincidence is the Collins error. Two more failures travel with it. The rarities are asserted, listed, counted, and perhaps multiplied, but not measured. So, again, logic. Calling a figure distinctive is not the same as showing, from prior art for example, how distinctive. And the shared combination is usually drawn after both songs are known, which lets the analyst pick, from melody, harmony, rhythm, structure, and lyric, whichever handful happens to line up. Choose the target after the shots land and any two songs share a bullseye. We do not gladly tip our hat to such delusion or chicanery. The law strives to be noble. So does the forensic musicologist expert witness.

And this accumulation effect does not usually start with the expert. It starts with the client. The expert might strive to be compassionate. The moment someone believes they were wronged, or senses an opportunity, the belief comes first and the evidence is gathered to fit it. That’s just people. People notice what confirms what they already feel and wave off what runs counter to it. This is ordinary, and it operates with sincerity. A songwriter convinced their song was taken will hear the resemblance in every shared interval and discount every difference, and mean none of it dishonestly. That is why I do not start from the client’s belief. I finish my own read before I read theirs. The pull is understandable, and correcting for it is most of what an independent analysis is for.

Return to that pithy if damning phrase. Hanlon’s razor. None of this requires assuming bad faith. The analysis stops one question too early. It establishes that a combination exists and is original enough to protect, and treats that as the finish line. The real question is whether the similarity makes independent creation implausible.

Why this analysis is hard to replace

For better or worse, forensic musicology is complex. The work teaches something the framework around it easily forgets. Precedent is treated or relied upon as a guiding light, but is better understood as a well-intentioned set of conventions, built case by case, with courts doing their best with disparate facts. The conventions are useful. They are not guardrails that also dictate truth; they were never built to be. A holding about two other songs was an answer to those songs. If it applies to the matter at hand, great. If not, don’t force it.

The principles hold. Originality is a low bar. Common building blocks are free. You cannot multiply probabilities that are not independent. Those travel from one case to the next because they’re not just precedent, they’re logic. But each comparison is its own. Music carries too many factors, interacting in too many ways, for a simple platitude to carry the day. The same discipline binds the analyst. A precedent does not rescue a weak musical case, and an expert who hides behind one has skipped the work that actually decides it. That work is listening to these two pieces of music and accounting for everything in them.

Litigation and Public Record

My written analysis appears in public federal court filings. Recent matters include:

Bouchet p/k/a Rawyal v. Woods et al. United States District Court, Western District of New York. No. 6:25-cv-06588-MAV.

Gary Frisby v. Sony Music Entertainment et al. United States District Court, Central District of California. No. 2:19-cv-01712-GW-AGR.

Brandon Saunders (EBA Executives, Inc.) v. Christopher Brown et al. United States District Court, Central District of California. No. 2:21-cv-09237-DSF-GJS.

Additional consulting analysis appears in matters not reflected in public filings, including pre-litigation and confidential engagements.

I also bring firsthand experience as a copyright owner in U.S. federal copyright litigation, distinct from my work as a retained expert. I know what a copyright claim asks of the party who brings it.

Work with Musicologize

Musicologize, the practice of forensic musicologist Brian McBrearty, is a decade-plus body of published forensic analyses of real music copyright disputes, including Shallow, Levitating, Stairway to Heaven, the Ed Sheeran cases, and Dark Horse. Each is worked the same way, separating protectable expression from common building blocks, judging substantial similarity element by element, and testing every claim against prior art. That analysis is cited in the law reviews of the University of Richmond, Washington University, Rutgers, the University of Southern California, and Georgia State, and the written work appears in the United States federal court record. The method holds true on a case-by-case basis, and it can be checked against how those cases actually resolved.

If you are weighing a claim or you need a fast pre-release check, call. You will get a concise, independent read of where the music actually stands and a defensible path forward.

Regular turnaround is 72 hours.

Call (212) 217-9512. Email brianmcbrearty@gmail.com. Schedule a preliminary call at musicologize.com/contact. The initial call is free.

The musical analysis is mine. Legal advice is your attorney’s.

Frequently Asked Questions

What is the difference between a music expert witness and a music copyright expert witness?
A music expert witness can testify about many aspects of music. A music copyright expert witness focuses on copyright, plagiarism, infringement, and song-similarity disputes, combining technical musical analysis with the legal standard the court applies. The narrower focus is the point. Copyright cases turn on the line between protectable expression and common musical convention, and that line is the specialty.

How is similarity between two songs actually evaluated?
Through element-by-element analysis of melody, harmony, rhythm, and structure, followed by testing each shared element against prior art to separate protectable expression from common convention. Overall impression is not the test. The extrinsic test requires dissection, and an automated tool measures gross similarity rather than the legally meaningful kind.

When should an attorney bring in an expert?
As early as possible. An early viability read tells you whether a claim is musically real before you commit to it, on either side. Earlier engagement shapes the decision to proceed. Later engagement can only describe a decision that has already been made.

What qualifications should a music copyright expert witness have?
Command of music theory and analysis, plus real case experience, especially at the early stages where a clear independent opinion can shape outcomes long before trial. A published record that can be examined for method is a strength. An expert who cannot point to any documented basis for their framework is exposed under Rule 702.

What gets an expert’s testimony excluded?
Overclaiming into legal conclusions, substituting overall feel for analytical dissection, testifying beyond the written report, and thin or selective documentation of method and prior art. Reliability under Rule 702 and Daubert is about disciplined, documented method, not credentials.

Is it plagiarism, infringement, or copying?
People use the words interchangeably. Plagiarism is reputational, infringement is legal, copying is the act. The same analysis serves all three.

Music Expert Witness Services