Device test · Audio

Perfect Pitch Test

Two modes, answering two different questions. Name a tone with no reference to hear how your absolute pitch holds up — or run the staircase, which narrows the gap between two tones until it finds the smallest difference you can reliably tell apart, and reports it in cents. A semitone is 100 cents.

Modenote naming Trials0 Result Web Audio

Start with the system volume at a comfortable level. If you are wearing headphones and are not sure how loud it is set, hold them slightly away from your ears for the first tone.

Mode
The note was

Gap now 120 cents · reversals 0 of 8 · seed 51721

How this test works

Both modes synthesize sine tones in your browser with the Web Audio API. Nothing is downloaded, nothing is recorded, and no audio file is involved — each tone is generated at the moment it plays, at a fixed modest level, with a short fade at each end so that starting and stopping cannot themselves be mistaken for part of what you are listening to. In naming mode a single tone plays somewhere between A3 and A5 and you pick its name from the twelve; the table records what played, what you said, and how many semitones apart those were.

Resolution mode is the one that produces a measurement rather than a score. Two tones play in sequence, separated by a gap in cents, and you say whether the second was higher. Two correct answers in a row narrow the gap; one wrong answer widens it. That two-down-one-up rule is the standard adaptive method because it settles at the point where you are right about 71% of the time — comfortably above the 50% you would get by guessing, and well below the ceiling where every trial is obvious. A fixed list of intervals cannot do this: it measures whoever chose the list.

The number is withheld until the staircase has reversed direction eight times, and it is the geometric mean of the last six reversals rather than the arithmetic mean, because the steps are multiplicative and an arithmetic average would be dragged upward by the wide early ones. Simulated against listeners with known thresholds of 5, 10, 20, 40 and 80 cents, it converged in every run and tracked each of them, settling around three quarters of the true value — which is what a 71% target is supposed to do. What it cannot correct for is your equipment or your room, so the figure describes this run rather than a fixed property of you.

Frequently asked questions

What does this perfect pitch test measure?
Two different things, in two modes. Note naming plays a tone with no reference and asks you to name it — that is what absolute pitch means, and it is the mode this page's title describes. Cents resolution plays two tones and asks which was higher; the gap narrows when you are right and widens when you are wrong, converging on the smallest difference you can reliably hear. The first gives you a score out of however many trials you take. The second gives you a number in cents, which is the more useful of the two because it does not depend on how many notes you happen to guess.
Is this a test for tone deafness?
No. This page measures how finely you can tell two pitches apart and how often you can name one. It is not a test for tone deafness, it cannot detect amusia, and it is not a hearing test of any kind. Those are clinical questions and a browser page is the wrong instrument for them; a qualified professional is the right one. What you get here is a measurement of one run, under whatever headphones you are wearing, today.
What is a cent, and what threshold is normal?
A cent is one hundredth of a semitone, so the gap between two adjacent piano keys is 100 cents. Published averages for untrained listeners commonly land somewhere in the tens of cents, and trained musicians often reach single figures — but those figures come from quiet rooms and calibrated equipment, and this page has neither. Treat your number as a measurement of this run rather than a position in a ranking, and compare it against your own later runs rather than against someone else's.
Why does the gap change after every answer?
Because a fixed list of intervals measures whoever wrote the list. This uses an adaptive staircase: two correct answers narrow the gap, one wrong answer widens it. That rule settles at the point where you are right about 71% of the time, which is well clear of the 50% you would get by guessing. The trade is that early trials are the staircase travelling rather than measuring, which is why no number is reported until it has reversed direction eight times.
If I score well, do I have perfect pitch?
A good run is not a certificate. Naming twelve notes gives you about a one-in-twelve chance of being right by luck on any single trial, so a handful of correct answers means very little and this page says so under eight trials. More importantly, absolute pitch as musicians use the term means naming notes reliably across a whole instrument's range, on unfamiliar timbres, without a reference — far more than a short run of sine tones in a browser can establish. What a strong result here honestly supports is that your pitch memory is worth testing properly.

Related tools