How Hearkle measures your ear

What the numbers mean, how they're worked out, and where they stop.

Every Hearkle game ends with some kind of score. Before you read too much into one, here's the honest frame: each result is an informal, on-device estimate of how you did on that listening task, right then, on that hardware and in that room. It's a practice tool giving you feedback, not a clinical measurement of your hearing. Keep that in mind and the numbers become genuinely useful. Treat them as lab results and they'll mislead you.

The one caveat that changes everything: what you listen on

Start here, because it colours the rest. The tones Hearkle plays are generated live in your browser, but what actually reaches your ear depends on the whole playback chain. Tiny laptop speakers and phone speakers roll off the low end and colour the tone, so two pitches that are easy to tell apart on good headphones can smear together on a speaker. Volume matters too: set it too low and fine differences vanish under your own hearing floor; set it very loud and the sound can distort. And the room is part of the test. Background noise, a humming fan, or traffic outside all eat into the small differences the games ask you to hear.

On top of the gear, hearing itself is personal. It varies from one listener to the next, it shifts with age, and most people hear slightly differently in each ear. Hearkle can't see any of that, so it can't correct for it. What this means in practice: your absolute numbers aren't comparable across setups. A pitch difference you resolve cleanly on wired headphones in a quiet room might read worse through a laptop speaker in a café, even if your ear is the same. Don't compare your score to a friend's on different gear, and don't compare your own scores across a change of device or room. The fix is simple: use the same headphones in the same quiet spot and watch how you change over time.

What the games actually measure

Hearkle's games fall into a handful of families, each probing a different piece of pitch perception. They borrow their logic from tasks used in hearing and music-cognition research, adapted into something you can finish in about a minute.

• Detection (JND). Two tones play, and you say whether they differ, or which is higher. This estimates your just-noticeable difference: the smallest pitch gap you can reliably tell apart. Smaller is keener.
• Identification. You home in on a target tone over a few tries, as in the daily game. The game reads how quickly and how closely you close on the right pitch.
• Reproduction from memory. A tone or a short sequence plays, it stops, and you repeat it back. This leans on how well you hold a pitch once the reference is gone.
• Ordering. You arrange tones from low to high, testing whether you can rank pitch differences you might not be able to name.
• Interpolation. Given two anchor tones, you find the pitch that sits between them, like the note halfway between a low and a high one.
• Real-time control. You keep a matching tone locked onto a target as it drifts, so the score reflects how long you stayed on pitch rather than a single guess.

The units, and how scoring works in principle

Different families call for different yardsticks. A detection game reports a threshold: the smallest pitch difference between two tones that you handled reliably. That gap is measured in cents, where 100 cents make a semitone. Because pitch is a ratio between frequencies rather than a fixed number of hertz, a difference measured in raw Hz means something different low down than high up, whereas cents describe the same musical step at any pitch. That's why the games lean on cents, and quote frequencies in Hz only when naming the actual tones. Identification and reproduction games report closeness to a target, often as how far off you were in cents. Real-time games report time on pitch. And several games track a streak, the run of correct calls before your first clear miss.

The scoring idea underneath most of them is the same. The game nudges the difficulty toward the edge of what you can do, then reads off where that edge falls. In a detection game, it makes the two tones closer in pitch when you're right and further apart when you're wrong, homing in on the smallest gap you can still catch. That converged gap is your threshold for the session. In an identification or interpolation game, the score is just how near you landed to the correct pitch. There's no curve grading you against other players. It's you against the task.

Difficulty adapts to you

Most games don't sit at a fixed level. They watch how you're doing and adjust: get several right and the task tightens, slip and it eases back. That keeps you near the boundary of your ability, which is both where practice pays off most and where the measurement is most informative. It also means the game meets a first-timer and a trained musician at their own edges rather than boring one and crushing the other. The flip side is worth stating plainly. Because the difficulty moved to find your edge, the final number reflects that adaptive search, not a single fixed test everyone takes identically.

What a score does not mean

Being clear about the limits is the point of this page. A Hearkle result is not a diagnosis of anything, and it is not a hearing test or an audiogram. A clinical hearing test measures the quietest sounds you can detect across calibrated frequencies, on calibrated equipment, in a controlled room. Hearkle does none of that: it plays uncalibrated tones through whatever you happen to be listening on, and it measures pitch discrimination, not hearing sensitivity. If you're worried about your hearing, or you notice a real change in one ear, see an audiologist or a doctor rather than reading anything medical into a game score. A Hearkle number also isn't a fixed verdict on your ability. Any single session is noisy. Sleep, focus, the room, the volume, and your gear all push a result around. One low score doesn't mean you're bad at pitch, and one great score doesn't mean you've arrived. Your "true" ear is not a number you got on a Tuesday.

How to read repeated measurements

This is where the tool earns its keep. Don't fixate on today's number. Play on a regular schedule, keep the same headphones and the same quiet spot, and watch the trend across days and weeks. A single point tells you almost nothing; a line sloping the right way over a couple of weeks tells you a lot. Expect wobble from day to day even as you improve, and judge progress by the direction of the cloud of points, not by whether today beat yesterday. If you want the deeper theory behind the cents thresholds, the guide on pitch JND and how small a difference you can hear goes into it, and the guide on what cents are in music explains the unit itself.

More about the project

• Editorial policy: how guides are written, sourced, and kept modest.
• Corrections: how mistakes get reported and fixed.
• About the author: who builds and maintains Hearkle.
• About Hearkle: what the site is and how it works.
• All guides: the full library of pitch and ear-training articles.