Hearing Test
Find the highest frequency you can hear, with silent trials that catch a run you cannot trust — every tone is generated locally in your browser.
Read this before you read your result. This is not a hearing test in the medical sense. It is an indicative check of the top end of your hearing range, and the number it produces depends on your headphones, your volume setting, your sound card, the noise in your room and your browser far more than it depends on your ears.
Most consumer headphones roll off before human hearing does. A “failure” at 17 kHz very often means the headphones stopped, not the ears.
If you have a real concern — tinnitus, sudden hearing loss, hearing loss in one ear, a muffled feeling that will not clear, or trouble following speech in a noisy room — see an audiologist. Those symptoms deserve a calibrated audiogram from a person, not a web page. Sudden sensorineural hearing loss in particular is treated as a medical emergency and the window for effective treatment is measured in days, so do not spend a week testing yourself on websites first.
This browser cannot generate audio
Every tone on this page is synthesised with the Web Audio API (AudioContext), and this browser does not provide it. Nothing is broken on your machine — the page simply has no way to make a sound here.
It works in current Chrome, Edge, Firefox, Opera and Safari, including iOS Safari 14.5 and later. If you are on a locked-down corporate build or a very old browser, open this page somewhere else and the test will run.
Turn your volume down before you start — especially in headphones.
High-frequency tones are piercing, and a level that felt fine for music can be genuinely painful at 12 kHz. Drop your system volume to roughly a quarter, then bring it up with the slider on this page.
Audio starts at about 12% of full scale and this page will never send more than 40%. The slider below is the real output gain, not a cosmetic one.
Browsers require a click before any page can make sound, so nothing plays until you press this.
Play the reference and set the slider so it is clearly audible but quiet — about the level of a normal conversation in the next room. Leave it there for the whole test. This makes your runs comparable to each other; it is not a sound-pressure calibration, and this page has no way to know how loud the tone actually is at your eardrum.
1 kHz sits near the most sensitive part of human hearing. The test tones are sent at the same electrical level, so they will sound much quieter — that is physics, not a fault.
Each ear keeps its own result, so you can run all three and compare. A clear gap between your two ears is the one finding here worth taking to a professional.
Discrete tones from 8 kHz upward, about 2.5 seconds each with a ramped fade in and out. Answer honestly after every one. Some trials play nothing at all — they look exactly the same on screen, and they are there to check that the run is worth trusting. The test stops after two misses in a row.
Not started.
A single tone glides continuously, logarithmically, so every octave takes the same amount of time. Press Mark at the moment it disappears (high sweep) or appears (low sweep). The live frequency is deliberately hidden while the sweep runs — if you can watch the number climb, you stop reporting what you hear and start reporting what you expect.
Idle.
The low sweep is mostly a test of your equipment, not your ears: almost nothing you own reproduces 20–40 Hz, and hearing silence down there is the expected result. Do not raise the volume to chase it — that is how small drivers die.
Every tone here is synthesised on your machine by the Web Audio API. This page never requests your microphone, camera or screen — it only sends sound out. Nothing is recorded, and no result leaves your browser unless you copy it out yourself.
How to run it, and how to read the result
- Use wired headphones if you possibly can. Wired over-ear or in-ear beats Bluetooth, and both beat laptop, phone or monitor speakers by a mile. A Bluetooth link running a low-bitrate codec can discard everything above about 14 kHz before the sound ever reaches the driver, which caps your score at a number that says nothing about you.
- Get the room quiet. Fans, a mechanical drive, coil whine and traffic all sit on top of the faintest tones. If you can hear your PC, you are testing your PC.
- Set the level once, with the 1 kHz reference, and then leave it alone. Changing the volume mid-test invalidates the comparison between one step and the next — every step after the change is being played at a different level.
- Pick an ear, then answer honestly. Press I heard nothing when you are not sure. Some trials are silent by design, and pressing yes on those is what flags the run as unreliable.
- Read the highest tone you reliably heard — not the highest you thought you might have heard. If the run reports false alarms on the silent checks, throw it away and run it again; the number is not usable.
- Run each ear separately and compare. Two ears that agree within one step are unremarkable. A gap of several steps between them is the one result on this page that is genuinely worth mentioning to an audiologist, because asymmetry has causes that symmetric age-related loss does not.
What is actually being measured
An oscillator in your browser produces a mathematically perfect sine wave. Between that number stream and the hair cells in your cochlea sit a resampler, your sound card's digital-to-analogue converter, its reconstruction filter, an amplifier, a transducer, the seal between the earpad and your head, and the noise floor of the room. Every one of those attenuates the top octave more than it attenuates anything else. What this page measures is the whole chain, and your ears are only the last link in it.
Failure mode 1 — the headphones stopped, not you
This is by far the most common reason for a low result. Manufacturer specs that claim 20 Hz to 20 kHz are almost never accompanied by a tolerance, and a driver that is 12 dB down at 17 kHz still technically "reaches" it. Small in-ear drivers, laptop speakers and anything built into a monitor typically fall off a cliff between 15 and 18 kHz. If you switch to a different pair of headphones and your result jumps by two steps, you learned something about the headphones.
Failure mode 2 — the sample rate ceiling
Digital audio cannot represent any frequency at or above half its sample rate. A device running at 44,100 Hz has an absolute ceiling of 22,050 Hz, and the anti-imaging filter in front of the DAC is already pulling the level down hard a kilohertz or two below that. This page reads AudioContext.sampleRate, refuses to synthesise anything within 500 Hz of Nyquist, and flags every tone above 45% of the sample rate as compromised. A 22 kHz "miss" on a 44.1 kHz device is not a fact about your hearing — the tone was never really there.
Failure mode 3 — aliasing that makes you hear things that are not there
If the browser or the OS resamples between rates — 48 kHz content pushed into a 44.1 kHz device, for example — a tone near the top of the range can fold down and reappear as a completely different, much lower, easily audible frequency. That is why an occasional person reports hearing 21 kHz clearly and nothing at 19 kHz: they are hearing an artefact. If your result is oddly non-monotonic, suspect the signal chain, not a miracle.
Failure mode 4 — hearing the click instead of the tone
A tone that starts instantly has an infinitely steep edge, and a steep edge is broadband: it contains energy across the entire spectrum, including frequencies you hear perfectly well. Anyone can "hear" a 20 kHz tone that switches on abruptly, because what they heard was the click. Every envelope on this page ramps in over 150 ms and out over 250 ms for exactly that reason. Any online hearing test that clicks when the tone starts is overstating your range, and so is any test that lets you drag a slider through the frequencies in real time.
Failure mode 5 — constant voltage is not constant loudness
Human hearing is far less sensitive at 16 kHz than at 1 kHz, by a large margin, and the gap widens as you go up. Every tone here is sent at the same electrical amplitude, so the perceived loudness falls away steeply across the test even on perfect equipment. A properly calibrated audiometer compensates for this by raising the level at each frequency until you respond, and reports the level in decibels of hearing level. This page cannot do that, because it has no idea what your volume knob, amplifier and headphones do to a signal. That single limitation is why the output is a range check and not an audiogram.
What a bad result means, and what to do about it
- Nothing heard at all, even at 8 kHz: almost certainly output, not ears. Wrong playback device, muted app, volume at zero, or a Bluetooth link that connected in headset mode. Check the speaker test first.
- You stop somewhere between 14 and 17 kHz: completely ordinary for an adult, and equally consistent with ordinary headphones. It is not evidence of damage.
- False alarms on the silent checks: the run is not usable. Faint high tones are extremely suggestible, and if you have tinnitus you may genuinely be hearing something — it just was not coming from this page.
- One ear several steps behind the other: re-seat the headphones, swap them left-to-right and run it again to rule out the hardware. If the gap follows your ear rather than the earcup, get it looked at.
- Anything sudden. Age-related loss arrives over years. A change you noticed this week is a different thing entirely and belongs in front of a clinician now, not next month.
Note what this page does not test: the speech range. Between roughly 250 Hz and 4 kHz, where conversation lives, essentially any playback device can reproduce everything, so a browser test there would measure your volume slider and nothing else. Hearing loss that affects your daily life shows up in that region, and finding it needs a booth, a calibrated transducer and a person.
Why high frequencies go first — and why the mosquito tone works
Presbycusis is age-related hearing loss, and it starts at the top of the range for a structural reason. The cochlea is a coiled tube, and it is tonotopic: high frequencies are detected at the base, right where sound enters, and low frequencies deep at the apex. Every loud sound of your life passes over the high-frequency hair cells on its way in, so those cells accumulate the most mechanical stress and die first. The loss then creeps downward. Most people are already below 20 kHz by their late teens and below 17 kHz by their thirties, entirely normally.
That is what the mosquito tone exploits. Deterrent devices sold to discourage teenagers from gathering emit a continuous tone around 17 kHz: piercing and unbearable if you are fifteen, completely inaudible to the shop owner who installed it. The same trick turns up as a ringtone that students can hear and teachers cannot. It is not a party trick about superior ears — it is a demonstration of a slope that everyone slides down.
Noise-induced hearing loss is the other half of the story, and it is the half you control. Damage from loud sound is cumulative and permanent: hair cells in the human cochlea do not regenerate, so every exposure adds to a total that never goes back down. It is the combination of level and duration that matters, which is why a single concert and a daily commute at high volume can do comparable harm. The rough guide people usually quote is the 60/60 rule — no more than 60% of maximum volume, for no more than 60 minutes a day, then a real break. Ringing, fullness or a muffled feeling after listening is not a quirk; it is a temporary threshold shift, and it means the level was too high even if your hearing seems normal again by morning.
Tinnitus — ringing, hissing or buzzing with no external source — is a reason to see a professional, not a reason to run another web test. Persistent tinnitus, tinnitus in one ear, tinnitus that pulses in time with your heartbeat, or tinnitus arriving alongside hearing loss or dizziness all warrant an actual appointment. So does any sudden drop in hearing: that is treated urgently, and waiting costs you options.
FAQ
Is this a real hearing test?
No. This is an indicative check of the top end of your hearing range, not an audiogram. A clinical test uses calibrated transducers in a sound-treated booth and measures your threshold in decibels at each frequency. This page has no idea what sound pressure level actually reaches your ear, so it can only report the highest tone you said you heard, on this equipment, at this volume, in this room. If you have tinnitus, sudden hearing loss, hearing loss in one ear only, or a muffled feeling that will not clear, see an audiologist rather than a web page.
I could not hear 17 kHz. Does that mean my hearing is damaged?
Usually it means your headphones stopped, not your ears. Most consumer headphones and nearly all laptop, phone and monitor speakers roll off steeply somewhere between 15 and 18 kHz, and low-bitrate Bluetooth codecs throw that region away entirely. Try a wired pair, raise the volume a little, and run it in a quiet room before drawing any conclusion. Losing the top of the range with age is also completely normal, and it is not something a browser can diagnose.
What are the silent trials for?
Some trials in the step test play nothing at all. They look identical on screen to a real trial, so the only way to answer them correctly is to actually listen. If you report hearing a tone that was never played, the run is flagged as unreliable. Very high tones are faint and easy to imagine, and expectation alone makes people answer yes, so every honest hearing screen includes catch trials for exactly this reason.
Why does the test skip 21 or 22 kHz on my machine?
Because your sound card cannot reproduce them cleanly. Digital audio can only carry frequencies below half the sample rate, so a device running at 44100 Hz has a hard ceiling of 22050 Hz, and its reconstruction filter is already attenuating heavily well below that. This page reads your real output sample rate and refuses to play tones that would come out distorted or silent for reasons that have nothing to do with your ears.
Why can teenagers hear the mosquito tone and adults cannot?
Presbycusis. The hair cells that respond to the highest frequencies sit at the very start of the cochlear spiral, where every sound that enters the ear passes over them first, so they take the most punishment across a lifetime and they fail first. The loss creeps down from the top of the range and usually passes below 17 kHz somewhere in the twenties or thirties. Devices sold to move teenagers along exploit that by emitting around 17 kHz, which is unbearable at fifteen and inaudible at forty.
How loud is too loud, and what is the 60/60 rule?
Listen at no more than 60 percent of maximum volume, for no more than 60 minutes a day, then give your ears a proper break. Noise damage is cumulative and permanent — the hair cells do not grow back — and it is the combination of level and exposure time that matters, so a loud concert and a quiet but constant commute both add to the same total. Ringing or muffled hearing after listening means it was too loud, even if it fades by morning.