Sound tolerance

Misophonia test where nothing plays until you say so

Rate 12 trigger situations — eating, breathing, clicking, ticking, one you only see — on a five-point scale, and get back a profile by family rather than a severity score. It is free, takes about six minutes, needs no account, and eleven of the twelve carry a synthesized 4.5-second sample that sounds only when you press that item’s own button. Answer the whole thing from the written descriptions instead and the profile comes out the same.

100% free
No signup
12 triggers, 5 families
Every sample opt-in
Works with sound off

Two ways through, and both produce the whole profile

12 trigger situations, each rated on a five-point scale, then six questions about what the reaction feels like and what it costs. 11 of the 12 have a 4.5-second synthesized sample attached. Nothing plays until you press a named button for that one specific sound, and the button says what the sample is made of before you press it.

You can switch to descriptions at any point without losing the ratings you have already given.

How to work through the trigger inventory

Choose whether sound is involved at all, then rate each situation once.

Pick the route, then set the level on hiss
The first screen offers samples or descriptions, and the descriptions route creates no AudioContext at all — there is nothing on the page that could make a noise. If you take the samples route, the level check plays pink noise, deliberately, because setting your volume against one of the triggers means being ambushed by a trigger to find out how loud the triggers are. Raise it until the hiss is clearly present and leave it well short of loud.
Read each situation, and play its sample only if you want to
Every item names the situation in a sentence and says what its sample is built from — the burst rate, the filter center, whether it is noise or a sine — before there is anything to press. One press plays one sample once, a Stop control sits at the top of the inventory the whole time, and switching to the descriptions route mid-way keeps every rating you have already given. The twelfth item has no sample by design: it is a bouncing knee, and animating one would be running the trigger at somebody who never chose to see it.
Answer the six questions that are not about intensity
Six checkboxes for what the reaction is made of — irritation, anger, anxiety, disgust, freezing, the urge to copy the sound — then three questions about whether the source matters, whether you have changed plans, and whether you carry something to cover it. Those three are reported back as themselves rather than added to anything, because how loud a feeling is and how much of your week it takes are different quantities and the published definition rests on the second.

Technical specifications

Items	12 situations across 5 families — 3 eating and mouth sounds, 3 nose and throat, 3 repetitive hand sounds, 2 steady background sounds, 1 seen rather than heard. Rated 0 to 4 on a labeled scale with no numeric midpoint offered as neutral
Samples	11 of the 12 carry a 4.5-second sample. Each is band-passed white noise in bursts on a jittered clock — 5 bursts a second at 3.2 kHz for the chip bag, one 15 ms click a second at 3.6 kHz with zero wander for the clock. The bass item is a pulsed 62 Hz sine, because filtered noise cannot produce a note that low
Consent model	No sound before a pink-noise level check; one press plays one sample once; a Stop control stays in the inventory header; the descriptions route never constructs an AudioContext. A sample that is playing when the component unmounts is faded out in 12 ms rather than following you to the next page
Scoring	Averages inside a family and never across them. The families hold 3, 3, 3, 2 and 1 item, so one total would silently weight eating sounds three times as heavily as the visual item
Heard against described	The profile counts how many of your ratings were given after playing the sample and how many came from the description alone, and prints both. A rating of an approximation you heard and a rating of a situation you imagined are not interchangeable
Prevalence figure	Not printed. Estimates in the literature sit far apart, and most of the distance between them is the criterion each study chose rather than the population it asked. The definition this page works from is Swedo et al. (2022), Consensus definition of misophonia: a Delphi study
Reported back	Five family averages, every item you rated 3 or 4 by name, the feelings you selected, and your three context answers in their own words. No total, no band, no percentile
What leaves the page	Nothing. The samples are generated by oscillators and noise buffers in this tab rather than fetched, so the inventory works with the network off, and the copy button writes to your own clipboard

Frequently asked questions

Will this page play a chewing sound at me before I am ready?

It cannot. There are three gates in front of every sample and all of them are yours: the opening screen where you choose samples or descriptions, the pink-noise level check that has to be passed before the inventory is drawn, and the individual button on the item itself, which names what the sample is made of before you press it. Nothing autoplays, nothing chains from one item to the next, and pressing Stop cuts the sound in about a hundredth of a second rather than at the end of the current burst. If you would rather not have the possibility on the page at all, the descriptions route never creates the audio context in the first place.

Is this the same thing as hyperacusis?

No — the two are separated by whether loudness is what matters. Hyperacusis is an intolerance of sound level: ordinary volumes are physically uncomfortable, and turning them down helps. Misophonia is an intolerance of specific sounds, usually quiet ones, and turning the volume down does not help at all, because the reaction is to what the sound is rather than to how much of it there is. That is why someone can be untroubled by a passing motorbike and undone by a person eating an apple a few feet away. Both sit under the audiological umbrella of decreased sound tolerance, along with phonophobia, and telling them apart is the first thing a clinician does.

Why are the samples synthesized rather than recordings of real people?

Because a recording carries the person in it and a rating of a recording is partly a rating of them. Their microphone, their room, their rhythm and how much you happen to like the sound of them all enter a figure that is supposed to be about the trigger. Filtered noise bursts at a chosen rate strip that out, so the item is the same item for every visitor and the parameters can be printed beside it. The cost is real and stated on each button: a burst train is an approximation of the acoustic shape of a trigger and not the trigger, so a mild rating for a sample is not evidence that the real thing is mild.

Why is there no severity score out of 100?

Because the two things that would have to be added together to make one are not the same quantity. How intense a sound feels and how much of your life it costs come apart constantly: people give the maximum rating on every eating item and have arranged their lives so it almost never comes up, and people give middling ratings and have stopped eating with their family. A single number hides exactly that difference, and the consensus definition of misophonia leans on the second half — distress and impairment — not on the first. So the page reports the intensity profile, the feeling mixture and the three cost questions as three separate readings and leaves them separate.

Why is one of the twelve items something you see rather than hear?

Because triggers are not confined to sound, and leaving the visual ones out would make the inventory quietly wrong for a lot of people. Seeing a knee bounce or a finger tap at the edge of vision produces the same rising reaction in many people who report sound triggers, and it has a name of its own — misokinesia — precisely because it needed one. The item is described rather than shown, which is the one asymmetry on this page: a sample you press to play is a choice, and a looping animation of a bouncing knee is not, since it would already be running by the time you decided you did not want it.

The sample sounded nothing like the real thing. Does my rating still count?

Yes, and the profile marks it. Every item is answerable from its written description, and the results panel prints how many of your ratings were given after listening and how many were not, because those are two different measurements sitting in one column. If a sample struck you as a poor imitation, rate the situation the sentence describes — that is the item. What you should not do is treat a low rating for a thin-sounding sample as evidence about the real sound, which is the specific error this split exists to make visible.

Does it mean anything that it is only my family who set me off?

It is a common enough pattern that the inventory asks about it directly. Reactions reported as much stronger when the person making the sound is a partner, a parent or a sibling are a recognized feature of the picture rather than an oddity of yours, and the usual reading is that misophonia is about meaning and context as well as acoustics — the same 40 dB chew from a stranger in a cafe and from someone across your own kitchen table are not the same event. The page reports your answer to that question on its own and draws no conclusion from it, because who a sound comes from is a fact about your circumstances rather than a measurement of you.

What the consensus definition says, and why the number stops there

Misophonia is a young word for an old complaint. It was coined in 2001 by audiologists working on decreased sound tolerance in tinnitus patients, spent two decades being argued about, and only reached an agreed description in 2022, when a Delphi panel of researchers and clinicians worked to a consensus definition: Swedo et al. (2022), Consensus definition of misophonia: a Delphi study. The definition that came out of it is worth knowing before you rate anything, because it is narrower than the popular usage. It describes a decreased tolerance of specific sounds — characteristically oral, nasal and repetitive ones — and of the sights associated with them; reactions of anger, irritation, disgust or distress that are out of proportion to the noise itself; and, crucially, consequences in the person’s life. Being annoyed by chewing is not the condition. Rearranging where you sit, what you eat and who you eat with is the part the definition is about, which is why this page keeps three questions about consequence away from the twelve about intensity and never adds them up.

The other thing the definition does not settle is how many people have it, and that is a gap rather than an oversight. Published estimates span a wide range because the field only recently agreed on a definition, and the figure depends heavily on whether the criterion requires significant distress or merely annoyance at trigger sounds. You can watch the mechanism directly in the published figures: a survey that asks whether specific sounds annoy you finds a large fraction of any population saying yes, and a survey that asks whether specific sounds have cost you relationships or jobs finds a small one, and both are honestly reported as the prevalence of misophonia. So no percentage appears anywhere on this page and no probability is attached to your profile. What the profile can do instead is show you the shape: which families of sound, how strongly, what the feeling is made of, and whether it has cost you anything — four readings that stay four readings. If you want the same question asked across the other senses rather than only this one, the sensory processing test covers sight, touch, smell, movement and body signals with sensitivity and seeking reported separately per sense.

Two design decisions here are worth explaining because most trigger tests make the opposite one. The first is that the samples are built rather than recorded: eleven burst trains of band-passed noise on jittered clocks, with the rate, the filter and the group structure printed on each button, so the stimulus is identical for every visitor and carries no particular person inside it. The second is that this page reports no time. Nothing here is measured in milliseconds and no reaction time is collected, so the timing floor that governs the speeded tasks on this site — the sixteen or so milliseconds of display and input hardware that sit under every reaction time — simply does not apply; the millisecond figures in the sample notes describe how a burst is shaped, not how fast anyone answered. That is a real difference from most of the neighborhood: the Simon task next door lives or dies on a difference of tens of milliseconds, and a self-report inventory that borrowed its apparatus would be dressing a questionnaire in a laboratory coat. For a neighbor that is also self-report but pairs it with something behavioral, the aphantasia test is the closest relative on the site.

This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish misophonia. Only a qualified professional, working with more than a browser, can make that judgment.

Where the sounds come from, and where your ratings go

Every number on this page is worked out by JavaScript running in the tab you are reading it in. Your answers, your reaction times and your score are never uploaded, logged or kept — which is also why the test carries on working after you disconnect from the network, and why nothing here can be held back behind an email address.

Nothing is recorded. This page asks the browser for the speakers and never for the microphone — the site requests no device permission of any kind — and every sample is generated from oscillators and a noise buffer inside this tab rather than downloaded, so the inventory works with the network disconnected and no request goes out when you press play. Your twelve ratings live in React state and are gone when the tab closes.