Hyperacusis Sound Therapy at Home — Low-Level Pink Noise, the Evidence, and the Mistakes That Make It Worse

Hyperacusis is reduced tolerance of ordinary sound: everyday noises that other people find unremarkable are experienced as too loud, uncomfortable or painful. It is not a hearing-test result; it is a judgement your auditory system makes about loudness, and that judgement has drifted.

The clearest short definition is from a tutorial by Henry and colleagues (Henry et al., 2022, American Journal of Audiology): hyperacusis “describes physical discomfort or pain when any sound reaches a certain level of loudness that would be tolerable for most people.” The words that matter are any sound and level: the problem is how loud something is, not what it is.

Clinically it is measured, imperfectly, with loudness discomfort levels: a tone is raised until you say it is uncomfortably loud. The largest published series (Sheldrake, Diehl and Schaette, 2015, Frontiers in Neurology) measured 381 patients whose main complaint was hyperacusis and found discomfort levels averaging around 85 dB HL across the frequency range, against roughly 100 dB HL in people with normal tolerance. Two details are worth keeping. More than a third of the ears tested had normal hearing thresholds, so hyperacusis is not a form of hearing loss. And discomfort levels were lowered at every frequency whatever the pattern of hearing loss, which the authors read as a generalised increase in gain in the central auditory system rather than a fault in the ear. That interpretation — the brain has turned the volume up — is why sound therapy exists at all.

A two-part review by Tyler, Pienkowski and colleagues (part I, part II, American Journal of Audiology, 2014) grouped the reactions people describe into four categories: excessive loudness, annoyance, fear, and pain. They matter because they call for different responses, and because several conditions with different mechanisms travel under similar names:

• Loudness hyperacusis is the one this page is about: moderate sounds are loud, and it is the form the sound-therapy trials were designed for.
• Pain hyperacusis is discomfort felt as pain, often in or around the ear, sometimes outlasting the sound. It is the least understood of the four and the one where pushing exposure can do harm — one reason the caution above says never at an uncomfortable level.
• Loudness recruitment comes with cochlear hearing loss: quiet sounds are inaudible, but once a sound crosses threshold its loudness grows abnormally fast, so the range between “cannot hear” and “too loud” is squeezed. Hyperacusis needs no hearing loss; recruitment requires it. They can coexist, and a hearing test tells them apart.
• Misophonia is a strong emotional reaction to particular sounds — chewing, sniffing, tapping — that does not depend on how loud they are; the Delphi consensus definition (Swedo et al., 2022, Frontiers in Neuroscience) calls it decreased tolerance to specific sounds or their associated stimuli. A quiet trigger is as bad as a loud one, the opposite of hyperacusis. See misophonia and trigger sounds.
• Phonophobia, in the audiological sense, is anticipatory fear of sound: avoiding situations because a sound might occur that will hurt or make tinnitus worse. Neurologists use the same word for sound aversion during migraine, a different thing. Fear can sit on top of any of the others and drives the avoidance that makes them worse.

Jastreboff and Jastreboff, who proposed the neurophysiological model behind tinnitus retraining therapy, draw the same line (Jastreboff and Jastreboff, 2015, Handbook of Clinical Neurology): hyperacusis is about the physical characteristics of sound, misophonia about its meaning. The distinction is practical: low-level broadband sound is a plausible response to a loudness problem, not to a chewing problem, and nobody should spend months on it for the wrong condition.

The idea behind sound therapy for hyperacusis is simple and has one good experiment behind it: the auditory system recalibrates its loudness scale to the sound it lives in. Everything after that experiment is thinner than the confident language of many clinics suggests.

The recalibration experiment. In 2003 Formby, Sherlock and Gold (Formby et al., 2003, Journal of the Acoustical Society of America) had adults with normal hearing wear either earplugs or low-level noise generators continuously for two weeks. Loudness judgements moved in opposite directions: after two weeks of attenuation, sounds became uncomfortable at lower levels; after two weeks of added low-level sound, tolerance moved up. Munro and Blount later showed that a week of a single earplug lowers the threshold of the middle-ear acoustic reflex in the plugged ear (Munro and Blount, 2009, JASA) — a change in brainstem gain, not just in what people call “uncomfortable” — and a follow-up found sound judged louder after a week of deprivation, with most of the change gone within a day of removing the plug (Munro, Turtle and Schaette, 2014, JASA). The mechanism claim is reasonably solid: your loudness scale is not fixed, and it tracks your recent sound diet.

Tinnitus retraining therapy. Jastreboff and Jastreboff built a treatment protocol on this idea — structured counselling plus all-day, just-audible broadband sound from ear-level generators — with hyperacusis as its own treatment category (Jastreboff and Jastreboff, 2000, Journal of the American Academy of Audiology). Their 2003 summary states that TRT “can provide cure for decreased sound tolerance” (Jastreboff and Jastreboff, 2003, Otolaryngologic Clinics of North America); read that as the originators reporting their own clinic, not a controlled trial. The Cochrane review of TRT (Phillips and McFerran, 2010) found one low-quality trial of 123 participants, for tinnitus, not hyperacusis. There is no Cochrane review of any intervention for hyperacusis.

The one placebo-controlled trial. The strongest test of the sound-therapy component is Formby and colleagues’ randomised trial (Formby et al., 2015, Seminars in Hearing). Be precise about who was in it: 36 adults with sensorineural hearing loss and a restricted dynamic range — people who could not get on with hearing aids because amplified sound was too loud — not people with primary hyperacusis and normal hearing. They were randomised, nine per group, to counselling plus real sound generators, counselling plus placebo generators, real generators alone, or placebo generators alone. The full treatment raised judgements of “uncomfortably loud” by about 12 dB across 500 to 4,000 Hz, more than any other group, with most of the gain arriving within three months. Two findings are less often quoted: counselling with placebo devices was statistically indistinguishable from the control condition in most analyses, and sound therapy alone did some of the work but clearly less than the two together. Both components were needed for the best result, and the effect, about 12 dB, is real and modest.

The recent field trial. The same group’s 2024 report (Formby et al., 2024, Journal of Speech, Language, and Hearing Research) is the first from this programme in people with debilitating primary hyperacusis and near-normal hearing. Twelve adults wore ear-level devices delivering low-level broadband sound, with counselling, for six months; the devices could also limit loud sound early on, a limit raised step by step at monthly visits according to measured loudness judgements. Eleven of twelve reached the improvement criterion and relied less on ear protection. The effect looks large. It is also twelve people, no control group and a patented device — a study that justifies a trial rather than replaces one.

The other treatment with a randomised trial is not sound at all. Jüris and colleagues randomised 60 patients with hyperacusis to cognitive behaviour therapy or a waiting list (Jüris et al., 2014, Behaviour Research and Therapy). Loudness discomfort levels improved with moderate effect sizes (Cohen’s d of 0.67 and 0.69 for the two ears), most questionnaire measures improved, the anxiety subscale did not, and gains were largely held at twelve months. The two lines of evidence converge: what changes tolerance is graded, sustained contact with sound plus a change in how it is appraised, whether the vehicle is a generator or a therapist.

What the reviews say. A scoping review from the Nottingham hearing group (Fackrell et al., 2017, BioMed Research International) found 43 records on interventions for adult hyperacusis, a third of which did not define the condition, and most of which studied people for whom hyperacusis was a secondary complaint. A scoping review of sound therapy specifically (Kalsoom et al., 2024, Brain Sciences) collected 31 studies and found that in all but two the effect of sound could not be separated from the counselling delivered alongside it. A systematic review published this year (Hamid et al., 2026, Acta Otorrinolaringológica Española) narrowed the field to nine quantitative studies, three of them randomised, and reported that every one showed improvement — which reads as good news until you notice that a literature in which every small study is positive is usually one short of well-controlled negatives. When the James Lind Alliance asked patients, carers and clinicians what they most needed to know (Fackrell et al., 2019, BMJ Open), none of the 85 questions they distilled had already been answered by research.

The honest summary: the mechanism is demonstrated in healthy ears; counselling plus low-level sound has one small placebo-controlled trial behind it, in a population that is not quite hyperacusis, with a real but modest effect; and the field’s own reviewers keep asking for the trial that would settle it. That is enough to make clinician-led sound therapy reasonable to try, and not enough to make an unsupervised web page a treatment.

Most of the harm in hyperacusis is done by sensible-seeming self-protection. Four habits in particular; the first has the best evidence against it.

1. Earplugs in quiet rooms. If loud sound hurts, wearing plugs everywhere feels rational. The recalibration experiments say it is the most reliable way to make tolerance worse. Two weeks of earplugs in normal-hearing adults lowered the level at which sound became uncomfortable (Formby et al., 2003); a single earplug for a week measurably raised gain in the brainstem (Munro and Blount, 2009) and made a given sound level feel louder (Munro et al., 2014). Schaette, Turtle and Munro went further: 18 volunteers with normal hearing wore one silicone earplug continuously for seven days, and 14 of them developed phantom sounds — tinnitus — pitched in the frequency range the plug attenuated most; every case disappeared when the plug came out (Schaette et al., 2012, PLoS One). Deprivation does not rest the auditory system; it turns the gain up, which hyperacusis already has too much of. Protection has a place — a concert, power tools, a fire alarm — and no place in a kitchen, an office or a quiet street. The TRT protocol makes “do not block your ears” an explicit instruction, and the 2024 field trial counted reduced reliance on protection as an outcome.

2. Going too loud, too soon. The clinical protocols run their sound just audible, for most of the day, for months. The temptation at home is to treat it like exercise and push the level up for faster progress. That inverts the mechanism: a sound that is uncomfortable is one your system flags as a threat, and repeated threat is how avoidance and fear get trained in — and how pain hyperacusis, which is poorly understood, can be aggravated. The comfort ceiling on the player above exists so that “a bit louder each day” runs out of road early. Progress in the trials was measured in months and in decibels of discomfort level, not in how loud a session felt.

3. Testing yourself against triggers. Deliberately standing next to the hand dryer to see whether it still hurts is not exposure therapy; it is a loudness discomfort test with no control over the level, and a failed test teaches your nervous system that the dryer is dangerous. Graded exposure in the CBT trial (Jüris et al., 2014) was planned, stepped and supervised; the point of a low-level sound bed is to change the baseline against which everyday sound is judged, not to schedule confrontations.

4. Treating it as a hearing-loss problem. Hyperacusis is often assumed to mean the ears are damaged and fragile. In the largest series, more than a third of tested ears had normal thresholds and discomfort levels were lowered at every frequency whatever the hearing pattern (Sheldrake et al., 2015); recruitment, the loudness problem that does come from cochlear damage, is a different condition that a hearing test identifies. So “resting” the ears, avoiding sound and turning everything down are not protecting a damaged organ; they are feeding a central gain problem. The exception bears repeating: sudden onset, a loud event, pain, one side, or any change in hearing is a medical question first, and the right move is a doctor.

The clinical programmes share a shape: quiet sound, most of the day, for months, at a level someone has measured, revisited on a schedule. Each part of that sentence does work, and each is where a home version usually fails.

Quiet. In TRT the ear-level generators are set so the noise is just audible — loud enough that you would notice if it stopped, quiet enough that you stop noticing it is on. The player above starts at 15 on a scale the site’s other players run to 100 on and cannot pass 45 — a proxy for “just audible” on an unknown device, not a calibrated level. If you cannot hear it with your device at its normal volume, raise the slider a little; if you can, leave it.

Most of the day. The protocol described in the sound-therapy scoping review calls for more than eight hours a day, plus a tabletop generator at night. The free web tool stops after twenty minutes because every free tool here does, and because a browser tab is a poor way to deliver eight hours of anything. That limit is also an honest statement of what this page is: a way to hear what just-audible broadband sound means and try a short spell of it, not the delivery mechanism for a programme. The noise generator runs to ninety minutes and the mobile app has no session cap; neither is a prescribed device either.

Months. In the 2015 trial most of the change in loudness judgements arrived within three months; the 2024 field trial ran six monthly visits. Nothing in this literature works in a week, which is about how long an unsupervised attempt usually lasts.

At a level someone has measured. This is the part a web page cannot do. An audiologist measures your discomfort levels before starting, which says how far below your tolerance the sound should sit and gives a number to compare against later. They can fit devices that sit in the ear all day, check for hearing loss and recruitment, and in the newer protocols set and then gradually raise a limit on loud sound so the world is turned back up in steps. A browser cannot know how loud your headphones are; two people at the same slider position are not doing the same thing.

What to track. If you are doing this with a clinician, they will tell you. If you are trying a short home experiment first, keep a plain daily note rather than a score: hours with some low-level sound around you; whether you used protection and where; one line on how ordinary sounds felt; and anything new — pain, a change in tinnitus, a change in hearing — which should end the experiment and start an appointment. Do not test yourself against triggers to generate data; the useful signal is slow, and shows up as fewer situations you plan your day around, not as a number.

Hyperacusis and tinnitus travel together, and the sound therapies for each are near relatives. In a Swedish population study of 3,645 people, hyperacusis was strongly associated with tinnitus, and the more severe the tinnitus the more likely it was present (Cederroth et al., 2020, Journal of Clinical Medicine); in the London hyperacusis series most patients also reported tinnitus (Sheldrake et al., 2015). The same paper argues the two probably do not share one mechanism, a reminder not to assume that what helps one helps the other. What sound-based approaches can and cannot do for tinnitus is covered on binaural beats and tinnitus; the Cochrane review of sound generators and hearing aids for tinnitus (Sereda et al., 2018) found eight trials and no usable data for any of its main comparisons. The tinnitus frequency matcher lets you hear roughly where your own tinnitus sits — an experiment, not a measurement.

What the tools on this site are. The player on this page and the noise generator make white, pink, brown, green and violet noise in your browser from a four-second loop; nothing is streamed and nothing is stored. Pink noise is the default here because its energy falls with frequency in roughly the way hearing sensitivity rises, so it sounds even rather than hissy at low levels, and because broadband sound is what the protocols use. The main generator is a different instrument: binaural, isochronic and pure-tone modes over a noise bed, capped at twenty minutes. Binaural beats need headphones; noise does not, and for a low-level sound bed a speaker across the room is usually the better choice. The mobile app adds 44 presets, 34 ambient sounds and the Frequency Lab, with no session cap.

What they are not. Binaural beats have no evidence in hyperacusis and no plausible reason to have any; this site does not suggest otherwise. None of the tools here is a prescribed sound generator, a hearing-aid programme or a substitute for a supervised programme. They are a way to hear what low-level broadband sound is, at no cost and storing nothing, while you arrange the appointment that matters.