Tinnitus Frequency Test — Match the Pitch of Your Ringing, and What the Number Can and Cannot Tell You

It is a way of putting a number on a sound only you can hear, so that you can talk about it. It is not a way of finding out what is wrong. Those are different jobs, and the second one belongs to a clinic.

Tinnitus is a symptom rather than a disease, and the guideline most clinicians in the United States work from — the American Academy of Otolaryngology–Head and Neck Surgery’s clinical practice guideline, Tunkel and colleagues (2014) — is mostly about sorting: bothersome from not bothersome, recent from persistent, and the ordinary kind from the kind that needs investigating. It recommends a prompt, comprehensive hearing evaluation when tinnitus is in one ear, has lasted six months or more, or comes with hearing difficulty. It reserves imaging for tinnitus that localises to one ear, pulses, or arrives with neurological signs or an asymmetric hearing loss — which is the source of the list in the caution above. It recommends a hearing-aid evaluation when there is documented hearing loss, and cognitive behavioural therapy when the tinnitus is persistent and bothersome. Sound therapy appears as an option the clinician may recommend, not a recommendation; supplements, routine medication and magnetic stimulation are recommended against. Nothing in that list can be done by a web page.

The reason the caution says today is sudden hearing loss. The companion guideline, Chandrasekhar and colleagues (2019), asks clinicians to confirm a sudden sensorineural loss with audiometry within fourteen days of onset, and lists corticosteroids as an option within two weeks. Tinnitus frequently arrives with it. If your hearing changed abruptly, the useful thing to do with this evening is to arrange to be seen, not to find the pitch.

For everyone whose tinnitus is the ordinary, persistent, both-ears kind, the number this page produces has three honest uses and one thing it is often mistaken for. The uses are covered below: describing the sound to an audiologist, knowing roughly where on your audiogram to look, and feeding a notch filter. The mistake is treating it as a severity score. In the review that still frames how tinnitus is measured, Henry and Meikle (2000) record the consensus that neither loudness nor the other psychoacoustic measures of tinnitus bear a consistent relation to how severe it is or how loud it feels to the person living with it. A high pitch is not a worse pitch. The number tells you where; it tells you nothing about how much.

Pitch matching is one of the four classic psychoacoustic measures of tinnitus, and the one this page borrows. Henry and Meikle date the set — pitch, loudness, maskability and residual inhibition — to a 1981 consensus meeting, and note that standardising how each is measured has been an unfinished project ever since. Pitch is the most intuitive of the four and the least stable.

The procedure sounds simple. The clinician plays a tone, the patient says higher or lower, and the tone is adjusted until it sits as close to the tinnitus as an external sound can. There are several ways to run that loop. In the method of adjustment the patient turns the dial. In the two-alternative forced choice the clinician plays two tones and the patient picks the nearer one, and the pair narrows each round. In likeness rating the patient hears a spread of frequencies and scores each for how much it resembles the tinnitus, which yields a spectrum rather than a single number. Neff and colleagues (2019) compared all three in 59 people with chronic tinnitus over five runs each and found good reliability for each, with forced choice showing the widest within-person scatter and likeness rating taking the longest; everyone got faster and more confident with practice. The slider on this page is a method of adjustment. The octave step is borrowed from the forced-choice tradition.

Octave confusion is the reason for that step. A tone one octave above or below another shares most of its harmonic structure and can sound, to someone concentrating on a phantom sound, like the same note. Matches land an octave off often enough that clinicians test for it explicitly, by playing the match and the tone an octave away and asking which is closer. In one of the earliest method comparisons, Tyler and Conrad-Armes (1983) found that many of their ten subjects produced matches spanning a full octave across seven repetitions, that one of the three methods produced more octave confusions than the others, and that some people were unreliable enough that they recommended repeating the match seven to nine times. In a web-based comparison, Mahboubi and colleagues (2012) saw five of twenty patients land an octave off with self-directed matching, all corrected by the octave challenge. And in the largest self-administered comparison, Kim and colleagues (2017) found octave confusion was the factor that most explained disagreement between 82 patients’ own slider matches and an audiologist’s. That is why this page does not treat a match as finished until it has offered you the octave either side.

Then there is the variability that no octave check removes. Penner (1983) had people with noise-induced hearing loss match a tone to their tinnitus repeatedly over twenty days and, as a control, match an external tone to another external tone; the tinnitus matches were, in her words, extremely variable relative to the matches for real sounds. Henry and colleagues (2004) collected fourteen matches from each of 42 people across three methods and found that about half of them spread their matches over a range of two and a third octaves, which led the authors to suggest reporting a range of pitch matches rather than a single one. Hoare and colleagues (2014) found that a computer-estimated dominant pitch agreed acceptably when sessions were two weeks apart, but that at three months only the group average held up, not the individual estimates — a finding they flag as a problem for any therapy prescribed from one person’s pitch. And Hébert (2018), retesting 31 people a month apart with experienced clinicians, found the standard forced-choice method gave the same final pitch in only seven of them, against 26 of 31 for whom likeness rating found at least one dominant frequency in common.

Read together, those studies say something useful: a tinnitus pitch is a region, not a point. If you match at 5.8 kHz tonight and 4.9 kHz on Thursday, neither reading was wrong and nothing has changed; that is what the measurement does. The way to use this page is to repeat it on different days and keep the range.

A pure tone is also not the only thing you can match against, and for many people it is the wrong thing. Clinicians sometimes match with narrow-band noise — a hiss centred on a frequency — rather than a sine wave. Korth and colleagues (2021) ran both in twenty people with chronic tinnitus over three days and found narrow-band noise gave smaller run-to-run differences; when the same people were asked to set the bandwidth of the comparison sound, none of them consistently chose a pure tone. This page plays a sine wave because that is what the shared oscillator produces. If your tinnitus is a hiss or a rush rather than a whistle, expect the match to feel approximate, and treat the frequency as the centre of a band.

The oscillator is exact to a fraction of a hertz; everything after it is unknown. That is the limitation the site’s hearing range sweep carries, and it applies here in a slightly different shape.

1. Your headphones have a frequency response, and you do not know it. Most consumer headphones and nearly all phone and laptop speakers are uneven above a few kilohertz and fall away steeply near the top of this page’s range. A tone that seems to change character as you drag the slider may be your headphones, not the tone. Matches above about 12 kHz are the least trustworthy for this reason alone.
2. Nothing is calibrated. A clinic delivers tones at known sound-pressure levels through equipment checked against a standard. Here the level depends on your volume setting, your device’s amplifier and the headphones’ efficiency, and the page cannot know any of them.
3. Loudness changes the comparison. A comparison tone that is much louder than your tinnitus is hard to judge against it, and a high sine tone can be uncomfortably loud before you register it as loud. The volume control starts low for both reasons. Raise it until the tone sits just above the ringing, no further.
4. Your room is not quiet, and your attention is not fixed. A fan, a fridge, traffic, tiredness and the effort of listening for a phantom sound all move the answer. The clinical studies that found large day-to-day variation were run in sound-treated rooms; a bedroom at midnight adds its own.
5. There is a session cap. Like every free tool on this site, the tone stops after twenty minutes. You should need a small fraction of that; listening to a sine wave near your own tinnitus pitch for long stretches is not something anyone recommends.

Against all that, there is a modest reassurance from the literature: self-administered matching is not hopeless. Mahboubi’s twenty subjects produced a median match of 6,000 Hz on an audiometer in an anechoic chamber and 5,925 Hz on a web slider, and the authors called the two protocols comparably accurate provided an octave challenge is included. Kim’s 82 patients averaged 6.3 kHz on their own slider against 7.0 kHz with the audiologist, with 57 of the 82 within half an octave. Wunderlich and colleagues (2015) put a forced-choice procedure on an iPod for seventeen patients and found reliability comparable with the clinic’s, and that people found the device easier. The catch is that every one of those studies handed the participant known headphones in a quiet room with a clinician nearby. What you have is the same idea with the controlled parts removed. The result is worth having as a description; it is not worth arguing with an audiogram over.

Three things, and each of them is a conversation — with a clinician, with your audiogram, or with a filter.

Describing it to an audiologist. “A high whistle” is what most people say. “Somewhere around 6 kHz, both ears, steady, closer to a hiss than a note, and it moved between 5 and 7 over three evenings” is a description a clinician can work with. They will redo the match properly — and will very likely get a different number, for all the reasons above — but yours shortens the conversation and shows that you have been paying attention to the sound rather than only to the distress. Note too whether it is in one ear or both, whether it changes with jaw movement or neck position, and whether anything covers it. Those observations matter more to the examination than the frequency does.

Knowing where to look on your audiogram. Tinnitus pitch tends to sit inside the region where hearing is reduced, rather than at its edge. Sereda and colleagues (2011) measured 67 people with chronic tinnitus up to 16 kHz and found the pitch generally fell within the area of hearing loss, with no relationship to the edge frequency across the whole group. Schecklmann and colleagues (2012) found the same in 286 patients: pitch tracked the frequency of maximum hearing loss, not the edge. Roberts and colleagues (2008) showed in 90 people that the whole tinnitus spectrum, and the sounds that briefly suppress it afterwards, cover the region of threshold shift — while also finding that reduced hearing on its own was not enough to cause tinnitus. So a match near 4 kHz is a reasonable prompt to ask about hearing at 4 kHz, where McBride and Williams (2001) describe the audiometric notch that is a well-established sign of noise damage. It is not a substitute for the audiogram, and if there is hearing loss the guideline’s advice is a hearing-aid evaluation — which the tinnitus page on this site is blunt about: nothing you can stream through headphones replaces amplification.

Feeding a notch filter. Notched sound therapy removes a band of frequencies around the tinnitus pitch from music or noise, on the theory that starving the overactive region of input lets its neighbours inhibit it. The number from this page is exactly the parameter that approach needs, which is why the summary panel links to the site’s notched noise player with your frequency already filled in. Be clear-eyed about what that buys you. Okamoto and colleagues (2010) reported reduced tinnitus loudness and reduced tinnitus-related cortical activity after twelve months of tailor-made notched music in a small study. The same group’s properly powered follow-up, Stein and colleagues (2016), randomised 100 people to notched or placebo music for two hours a day over three months and found no effect on any primary outcome; loudness in the notched group was lower than control only at the one-month follow-up. A 2024 meta-analysis by Alfonso and colleagues pooled three trials — 99 people on notched music against 109 on ordinary music — and found no significant difference in tinnitus handicap. And Hoare’s reliability paper makes the awkward point that a therapy prescribed from an individual’s pitch inherits that pitch’s instability. A notch is a reasonable thing to try because it is free and harmless at a sensible volume, not because the evidence says it works better than the same sound without the notch.

What it is not. It is not a diagnosis; the same pitch can come from noise exposure, age, an ear infection, a medication, or nothing anyone can find. It is not a severity score, as Henry and Meikle’s review makes plain. It is not a tracker of whether something is working, because the measurement moves more between sessions than most interventions move it. And it is not a hearing test: the site’s frequency sweep explains at length why a browser cannot produce one, and every word of that applies here.

Two things, honestly: cover the sound, and help you settle. Neither is a treatment, and the evidence for the first is weaker than most sound-therapy pages admit.

Start with the Cochrane reviews, because they are the least forgiving. Hobson and colleagues (2012) reviewed six trials of masking devices and hearing aids in 553 adults and found no significant change in tinnitus loudness or severity compared with education, relaxation, counselling or retraining — while adding, carefully, that the absence of conclusive evidence is not evidence that sound therapy does nothing, and that the trials were too varied and too weak to settle it. Sereda and colleagues (2018), updating the question with eight trials and 590 participants, found no evidence that sound therapy beats a waiting list, a placebo or information alone, no evidence that hearing aids, sound generators or combination devices differ from each other, and rated the evidence low quality throughout. Hoare and colleagues (2014) could find one randomised trial of hearing aids specifically for tinnitus, in 91 people, and concluded there is no evidence to support or refute their routine use for it.

Against that, the broadest recent synthesis is warmer. Chen and colleagues (2026) pulled together 44 systematic reviews and found cognitive behavioural therapy, hearing aids, tinnitus retraining therapy and sound or music therapy all consistently improved tinnitus outcomes, with neuromodulation and acupuncture modest and inconsistent. Both readings can be true at once: sound as part of a structured programme, alongside counselling, is associated with people doing better; sound on its own, tested rigorously, has not been shown to outperform the alternatives. The tinnitus page on this site walks through that ordering, and it puts CBT and a hearing test first for a reason.

Binaural beats specifically add nothing to that picture. Munro and Searchfield (2019) played twenty adults with tinnitus recorded ocean sound with and without an 8 Hz binaural beat and found small improvements in tinnitus ratings with sound, and no significant group benefit from adding the beat. Two 2025 trials from one group — Prakash and Konadath in the Journal of Otology and the same authors in Auris Nasus Larynx — found binaural-beat groups improving more than a standard masker in people with normal hearing, and then wrote that the benefit cannot be definitively attributed to entrainment and may reflect the effects of sound therapy in general. If a page tells you a beat frequency treats tinnitus, it is ahead of its evidence.

So here is what the free web tools are for. The noise generator plays white, pink, brown, green and violet noise in the browser, with a volume control and an auto-stop timer, and needs no headphones. For a high-pitched tinnitus, pink or white noise has energy up where the ringing is and can blunt it in a quiet room; brown has far less up there and is the wrong choice for covering a whistle, though a reasonable one for sleep if you want something lower. The main generator layers a binaural, isochronic or pure tone over a noise bed, capped at twenty minutes a session — binaural mode needs headphones, isochronic and pure modes do not — and its relaxation presets are there for the part of tinnitus that is about arousal rather than acoustics. Set any of these so the ringing is blunted rather than buried; the point is to make it easier to ignore, not to add a louder sound on top. The mobile app extends the same toolkit with 44 presets, 34 ambient sounds and the Frequency Lab, and has a tinnitus-relief preset among them; it is the same idea in more shapes, not a stronger medicine.

And this page is the one tool here that is not about listening at all. It is about a number you can carry into a room where someone can actually help.