A memory for the note itself, and it is dated

Nobody agreed on what A was until 1939, and the spread of documented tuning pitches across Europe spans nearly a minor third. That essay’s conclusion was that absolute pitch level is a convention rather than a fact.

This one is about the people who have memorised it.

The evidence that it is a label #

The strongest single argument that absolute pitch is a memory for a convention rather than a perceptual capacity is that it can be wrong in a specific direction — and wrongness in a direction is the signature of a stored reference rather than of a sharper instrument.

Baroque pitch breaks it. A possessor trained at A=440 listening to a period-instrument performance at A=415 reports every note a semitone flat of its notation. They are not mishearing: the frequencies are exactly what they report. What has failed is the correspondence between the frequency and the name, and it has failed by precisely the amount the ensemble is transposed.

The experience is described consistently by possessors as disorienting rather than merely inconvenient, and some report being unable to read the score at the same time as listening. That is a strong hint about what the faculty is for: it is being used as a naming channel that runs in parallel with the score, and a mismatch between them is a conflict rather than an error.

It drifts with age. Longitudinal and cross-sectional studies both find that possessors’ labels go progressively sharp as they get older — a listener in their seventies typically names a tone a semitone higher than they would have at twenty. The usual explanation is a slow physical change in the cochlea’s frequency map. Whatever the mechanism, a faculty that drifts by a semitone over a lifetime and continues to feel exact to its owner is a faculty that is reading a reference, not measuring.

And it is a categorisation, so it inherits categorical behaviour. Possessors are quick and accurate at naming, and considerably less good at judging how far a tone is from the nearest name. The ability delivers a label, and a label is a box — which puts absolute pitch in the same family as ordinary interval perception rather than in a family of its own.

A listener naming a continuously varied tone identifies it categorically, and the categories are what a change of temperament moves through. The label survives a retuning and the frequency does not, which is the precise sense in which absolute pitch is a memory for a category rather than for a note.

How rare it is, and what “it” is #

The commonly quoted figure is one in ten thousand in Western populations, and it is close to meaningless without a definition, because the ability is not binary.

Three things get called absolute pitch and they have very different prevalences.

Naming any tone quickly and accurately, without a reference. This is the strict criterion, and it is genuinely rare in listeners raised in Western musical cultures.

Naming tones on one’s own instrument, in the familiar range, slowly. Much more common among trained musicians, and often described by possessors as “not real” absolute pitch. It is nevertheless the same faculty operating with less coverage.

Retaining a stable reference over days or weeks. Common enough to be unremarkable, and it is what makes it possible for a singer to start a piece in the right key from memory.

The prevalence also depends sharply on two things that are not about ears at all.

Language. Speakers of tone languages — Mandarin, Cantonese, Vietnamese — show much higher rates, in some studies by more than an order of magnitude. In a tone language, pitch is lexical: it carries meaning and must be learned as a category from infancy. The obvious reading is that absolute pitch is the ordinary outcome of learning to attach meanings to pitch categories early, and that Western musical training, which teaches relative pitch almost exclusively, teaches it out.

Age at which training began. Nearly every possessor began musical training before about six, and the rate falls steeply for later starts. This is the standard shape of a critical period, and it is why absolute pitch is one of the more frequently cited examples of one outside language acquisition.

That is the whole of the faculty stated as a picture: twelve labels and a direct route to each. Nothing in it is comparative, which is why absolute pitch survives having no reference note and why it is defeated by a change of standard that relative pitch does not even notice.

That figure also makes plain what absolute pitch does not have to do. It does not have to place a tone precisely; it has to place it in the right hundred-cent box. The precision required is roughly fifty cents, which is an order of magnitude coarser than what any listener can discriminate. The scarce resource is not resolution. It is a stable long-term reference and a set of labels attached to it.

The part almost nobody knows they have #

The rarity claim is about labelling. The underlying memory is close to universal.

Asked to sing a familiar recorded song from memory — a pop song they have heard many times — listeners with no musical training and no absolute pitch start it within a semitone of the recorded key a substantial fraction of the time, and within two semitones the great majority of the time. That is a stored absolute pitch reference of considerable precision, held by people who would deny having one.

The same shows up in recognition: a familiar recording played a semitone away from its original key is reported as sounding “wrong” or “off” by listeners who cannot say why and cannot name either version’s key.

What is rare, then, is not the memory for pitch. It is the labels — the trained correspondence between the remembered pitch and a name in a system. Absolute pitch is a naming skill built on a memory almost everybody has, and the reason it looks like a perceptual gift is that naming is the only part of it that is externally visible.

That reframing also explains the drift and the Baroque-pitch failure without any further machinery. Both are failures of a correspondence. Neither is a failure of the memory.

Where it comes from, and the one intervention that works #

If absolute pitch is a set of labels attached to a memory, the interesting question is why nearly everybody ends up with the memory and nearly nobody with the labels.

The best available account is that early musical training in the West teaches the wrong thing. A child learning solfège or intervals is being trained explicitly on relations and implicitly to discard the absolute value — a melody is the same melody in any key, and the whole pedagogy is organised around that being true. Learning to name absolute pitches requires the opposite habit, and requires it during the years when the categories are being formed.

The training programmes that produce something like absolute pitch in children do exactly that: they attach a name to a specific pitch, on a specific instrument, before the relative framework has taken over. Suzuki-style methods that start with a named reference note are frequently associated with higher rates, and while the causal claim is hard to establish — the families who choose such methods are not a random sample — the direction is consistent.

The structure relative pitch actually uses is a ring of relations in which no position is privileged, and this is what that costs to draw on a keyboard: the same interval appears twice and has to be recognised twice. Absolute pitch has the labels and no relations; relative pitch has the relations and no labels.

What it is good for, which is less than expected #

The practical value of absolute pitch is smaller than its reputation, and the reason is worth stating because it says something about what music is made of.

Almost all musical structure is relative. A melody is a shape that survives transposition; a chord progression is a sequence of relations; a key is a set of relations to a tonic. Every one of those is available to a listener with good relative pitch and nothing else, and none of them is better available to a possessor of absolute pitch.

What absolute pitch does buy is naming without a reference, useful for transcription and for tuning. What it costs, according to a consistent body of reports from possessors, is a susceptibility to transposition: a piece heard for years in one key and then played in another can be actively unpleasant, and following a transposing instrument’s written part while hearing sounding pitch requires deliberate suppression.

One more thing it buys is worth naming because it is genuinely useful and rarely mentioned: a possessor can detect that a recording has been sped up or slowed down, which nobody else can do without a reference. In an era when tape speed was a routine production variable and playback speed was unreliable, that was a practical skill. It is a small illustration of the general point — the ability is valuable exactly where an absolute reference has been lost, and worthless everywhere else.

There is also a real hazard in over-reading it. It is often described as though it were a sharper ear, and it is not: possessors’ discrimination thresholds — the smallest frequency change they can detect — are not better than trained non-possessors’. The limen is the limen. The difference is entirely in what is done with the result.

What it does to reading and to transposition #

There is a practical asymmetry here that musicians argue about and that follows directly from the account above.

A transposing instrument — a clarinet in B♭, a horn in F — is notated at a pitch different from the one it sounds. A player reads a written C and produces a sounding B♭. For a player with relative pitch this is a systematic relabelling learned once and then automatic. For a player with absolute pitch it is a permanent conflict between two naming systems, both of which are running, and the conflict is reported by many possessors as genuinely effortful.

The same asymmetry runs the other way for score reading. A possessor can look at a page and hear it; a relative-pitch reader can look at a page and hear it in some key, which for almost every purpose is the same thing. What the possessor gains is the specific key, and what music is made of is almost entirely relations.

Whose practice. These are claims about Western notated music with a stable pitch standard, which is a condition that has held for less than a century — the standard itself was only fixed in 1939. Before that, a possessor’s labels would have been wrong relative to half the instruments they encountered, and it is worth wondering whether the ability was as prized when the reference it depends on moved by a semitone between one town and the next.

Half the instruments, and it is nearly exactly half #

That last sentence is a guess dressed as a fact, and the pitch-standard table is right there to check it against. Take the seven documented standards the hero figure plots and ask, of each, how far it sits from A=440 and how much of that distance is left over after the nearest whole semitone is removed.

The residual is the number that matters, and it is not the semitone. A standard a whole semitone away is the benign case: every label is wrong, all of them by the same amount, and one act of mental transposition repairs the entire mapping. That is what 415 and 392 and 465 are, and it is why the Baroque-pitch story is the one possessors tell — it is annoying and it is solvable.

A residual near fifty is the case with no repair. The old Philharmonic pitch of the 1890s sits 48.1 cents above A=440, which is a quarter-tone: every note it plays lands on a category boundary, no transposition improves matters, and a possessor hearing it is not mis-naming so much as failing to name. That standard was in ordinary use in London until the 1890s, well inside the working life of musicians who then had to accommodate 435 and later 440.

Across the whole table the residuals average 14.8 cents, where standards falling anywhere at all would average 25 — so the historical standards do cluster near whole semitones, presumably because organs and forks were copied from each other and copies transpose. But run the same test on every pair of standards rather than on each against 440, which is the situation a travelling musician was actually in, and ten of the twenty-one pairs are more than a quarter of a semitone from a whole number of semitones apart. The hand-waved “half the instruments they encountered” turns out to be 48 per cent.

Which computation produced the numbers #

Very little here is computed, and the essay should be clear that it is the most citation-dependent of the perception ladder.

What is computed: the semitone. A=415 against A=440 is 1200·log₂(440/415) = 101 cents, which is why the Baroque-pitch mismatch is a semitone rather than “a bit flat” — the two standards happen to sit almost exactly one semitone apart, and that is the arithmetic of the pitch-standard figure rather than a coincidence anybody arranged.

What is quoted: the prevalence figures, the tone-language association, the critical-period age, the drift with ageing, and the song-key results. All are reported effects with published sources, all have contested effect sizes, and two of them — the prevalence and the tone-language association — are the subject of active argument about whether the samples compared are comparable.

The residual table is the same arithmetic run over the whole list rather than over one pair: cents from 440, the nearest whole semitone removed, and what is left. The pairwise version takes all twenty-one unordered pairs of the seven standards and counts how many have more than a quarter-semitone left over. Both are three lines and neither needs any data the hero figure does not already carry, which is the reason for including them — the sentence they check had stood as a rhetorical flourish, and the flourish happened to be right.

The site’s habit is that every claim gets a test it could fail. The test available here is the semitone arithmetic, and the figure asserts it: if the pitch-standard table were mistyped such that 415 and 440 were not within a few cents of a semitone, the check would fail and the essay’s central anecdote would stop being true.

What the picture cannot show #

Whether it is learned or innate, which is the question everybody asks. The critical-period evidence and the tone-language evidence both point towards learning; a genetic component is also reported, and family clustering is real. The honest answer is that it looks like a learnable ability with a developmental window and individual variation in how easily it is learned, which is not a satisfying answer and is the one the evidence supports.

Adults acquiring it. Attempts to train absolute pitch in adults produce real, measurable improvement and do not produce the fast automatic naming that early acquirers have. Whether the difference is one of degree or of kind is unresolved, and it is the question on which the whole critical-period reading turns.

How stable an individual’s reference is over short periods. The drift studies measure decades. What happens over a week, or after a day of listening to music at a different pitch standard, is much less studied and would say a good deal about whether the reference is being maintained or merely stored.

And the possessors’ own descriptions, which are the primary evidence for how it feels and are unavailable to any figure. Several possessors have described a note’s name as arriving with the same immediacy as a colour’s, which is a comparison worth taking seriously and impossible to test.

Where the ladder goes next #

This anchor — the pitch standard as a convention — now has two rungs: the convention itself, and the people who have memorised one. What it opens is the question of what happens when a listener’s categories are stretched rather than their labels being wrong, which is where category width turns out to be what makes temperament possible at all.

Sideways, the same word does different work in a different place. A listener with absolute pitch has attached names to frequencies. A listener with none has attached an entire hierarchy of expectations to scale degrees, which is a much larger structure learned by the same mechanism — exposure, over years, without instruction — and which almost everybody has.