None of the five devices compared here — Lenire, Neosensory Duo, Levo, Neuromonics and Desyncra — has beaten a convincing sham; Lenire has the most evidence, and the one double-blind placebo-controlled trial of Desyncra's method found no difference from placebo.
How much a device's results mean depends on what it was compared against, and this page asks that of each one.
Why the comparison decides everything
Tinnitus scores often fall in studies whatever the treatment, as several comparison groups below show, so a before-and-after improvement says little on its own; what happens if you do nothing explains why. The 2018 Cochrane review of hearing aids, sound generators and combination devices found no data on its outcomes comparing any of them with a waiting list, a placebo or information alone, and too little to rank them against each other.
The 2024 VA/DoD tinnitus guideline lists Lenire, Levo and Neuromonics among the devices available as of May 2023, says inclusion is not an endorsement, and states that no evidence supports the superiority of one device over another, and that tinnitus-specific devices have not been proven superior to ordinary sound-generating devices. Reading a tinnitus study explains why control groups matter.
What "FDA-cleared" means here
Levo, Neuromonics and Desyncra were cleared through the FDA's 510(k) route under the regulation for a tinnitus masker, a class II device described as one intended to generate noise to mask ringing in the ears. A 510(k) clearance means the FDA agreed a device is substantially equivalent to one already legally on the market: a judgement of equivalence, not evidence that the new device works better.
Levo's two clearance summaries, from 2014 and August 2026, describe bench and software testing, not a clinical trial.
Lenire instead received a De Novo authorisation (DEN210033, March 2023), the route for a new kind of device, supported by the trial described below. Searches of the FDA's 510(k) and De Novo databases for this article found no clearance or authorisation under Neosensory's name.
Lenire: compared with its own first six weeks
Lenire pairs sound through headphones with mild electrical pulses on the tongue, up to an hour a day in two 30-minute sessions for 12 weeks. The Lenire article covers its pivotal trial in detail.
All 112 participants had six weeks of sound alone, then six weeks of sound plus tongue stimulation, always in that order. Among the 44 still at least moderately bothered at six weeks (a Tinnitus Handicap Inventory score of 38 or more), 58.6% responded during the bimodal weeks against 43.2% during the sound-only weeks; in the full cohort that advantage was not observed. The manufacturer, Neuromod Devices, sponsored the trial, and one author declares a consultancy with financial interests in the company.
The missing comparison is a sham. The authors write that a sham control was not possible because both stimuli are noticeable. With bimodal treatment always second, the design cannot separate the tongue stimulation from time or expectation.
Even so, it is the most tested device here. A 2026 meta-analysis of 26 randomised neuromodulation trials found bimodal stimulation the most consistent and durable of the methods it compared, while calling for adequately powered trials; bimodal neuromodulation devices covers it.
Neosensory Duo: compared with no wristband at all
The Duo was a wristband with four vibrating motors, paired with a phone app playing tones spread around the user's tinnitus pitch. Each tone set off a vibration at a point on the wrist matched to its frequency: bimodal stimulation, with the wrist in place of the tongue.
Its published evidence is one 2023 trial by three authors, all affiliated with Neosensory. It randomly assigned 45 adults, recruited online and scoring at least 25 on the Tinnitus Functional Index (TFI), to ten minutes a day for eight weeks of tones with the wristband or tones alone.
- With the wristband: TFI scores fell by 17.9 points on average (22 people).
- Tones alone: they fell by 7.5 points (23 people).
- Between the groups: the difference was statistically significant, at p = .04.
- Starting at 50 or above: the falls were 21.8 and 4.6 points.
The paper notes that a change smaller than 13 points is not considered clinically significant on the TFI.
The missing comparison is a sham wristband. The control group had no device at all, so the trial could not be blinded. The authors suggest the tones-only group's improvement may reflect a placebo effect, and a group handed a new device may have more reason for one. A wristband vibrating out of step with the tones would have separated the pairing from the novelty. Nobody was followed beyond eight weeks, and the paper's claim to match or exceed tongue stimulation compares figures from different trials.
Neosensory registered a 200-person trial, again against tones alone, in July 2024, estimating completion that November. Its record has not been updated since and shows no results. The Vestibular Disorders Association's article on bimodal devices carries an update saying the company ceased business operations in 2025 and the product is no longer available.
Levo: compared with other bedtime sound
Levo delivers sound therapy through earbuds — tinnitus pitch-matched tones, narrow-band noise centred on the tinnitus frequency or broadband noise, set by a clinician — and the VA/DoD guideline lists its use time as during sleep.
A randomised trial at the VA's National Center for Rehabilitative Auditory Research gave 60 people one of three sleep devices: a tinnitus-matched sound, a noise stimulus, or an ordinary bedside sound generator.
- Every group improved on average.
- The authors were at least 87% certain that the tinnitus-matched and noise options lowered TFI scores more than the bedside generator, by an estimated 4.5 to 5 points, and concluded the study offers some support for them.
- They were at least 95% certain the tinnitus-matched sound lowered loudness ratings more than both others, by about 0.75 points.
The missing comparison is no device, or a placebo. Every group had a sound device, so the trial cannot say what any of them added. Against the TFI developers' preliminary 13-point benchmark for meaningful change in a score, this gap between two improving groups is just over a third; how tinnitus is measured explains such benchmarks. On the TFI the tinnitus-matched and noise options shared the advantage; only on loudness did the tinnitus-matched sound stand apart.
The August 2026 clearance covers a second-generation Levo, used during sleep for up to eight hours, which adds an education programme informed by CBT and mindfulness. A PubMed search for this article found no published trial of it.
Neuromonics: compared with counselling, by the people who built it
Neuromonics uses music spectrally modified to each person's audiogram, played for at least two hours a day within a structured programme of counselling. It was cleared as a tinnitus masker in January 2005, and the 2024 VA/DoD guideline lists its Oasis Pro version at a minimum of two hours a day for six to eight months.
The favourable trials came from its developers:
- A 2007 trial randomised 35 people between two versions of the treatment. Both improved, and at six months 91% reported at least a 40% improvement in tinnitus disturbance. No group went without Neuromonics.
- A 2008 study compared it with counselling plus broadband noise and with counselling alone: 86% met its success criterion, against 47% and 23%. Its abstract does not say how people were allocated.
Paul Davis and Peter Hanley, authors of both, declared in a 2008 review a financial interest in Neuromonics Pty Limited, the company that commercialised the treatment.
The missing comparison is an ordinary sound generator, tested independently. The closest is a 2012 Cleveland Clinic study, not randomised, of 56 patients who had chosen either Neuromonics (33) or ear-level sound generators (23). Both groups improved on the Tinnitus Handicap Inventory, with no difference at six months, and the economic analysis favoured the sound generators. A nine-centre US study of 52 patients had no control group, and 14 withdrew or were lost to follow-up.
Desyncra: tested against a placebo, and it did not win
Desyncra delivers acoustic coordinated reset neuromodulation: sequences of tones just above and below the tinnitus pitch, intended to break up synchronised brain activity, for several hours a day. Its January 2016 clearance is for relief while using it, adding that regular use over months may bring relief when not.
- The first trial, in 2012, randomised 63 people between several doses and a placebo, single-blind. Treated groups improved and the placebo group did not change significantly, but the groups were not compared with each other on that measure, and the placebo group's 8.4-point fall on the Tinnitus Questionnaire sat inside the treated groups' range of 5.2 to 15.5.
- The double-blind trial, RESET2, tested the method on an earlier device (the T30 neurostimulator, from the manufacturer at the time). It gave 100 adults the same device, programmed with the treatment sequence or a placebo sequence avoiding the tinnitus frequency, for four to six hours a day over 12 weeks. The difference in tinnitus handicap was −0.45 points (95% CI −5.25 to 4.35; p = 0.85), no secondary outcome differed, and scores did not change significantly over a 24-week open-label extension.
RESET2 is the clearest example here of a sham done well: the same hardware and hours, playing a sequence that should not work. Run from 2012 to 2014, it posted its main results on ClinicalTrials.gov in June 2015, but the journal paper appeared only in 2022; the paper attributes the delay to an agreement between the sponsor and the funder, The Tinnitus Clinic, to embargo journal publication. A 2017 opinion piece co-written by RESET2's lead investigator and the chief executive of Desyncra Technologies, a company in the same group as that funder, argues the trial's pitch matching departed from the manufacturer's recommendations, a deviation the trial paper also records. It says commercial fitting has since used automated pitch matching.
A 2022 VA trial compared Desyncra with a short course of CBT in 61 people, unblinded, with Desyncra, Inc. as a collaborator. Tinnitus Questionnaire scores fell by 5 to 15 points across the arms, the gap between treatments was 1 to 2 points, and the authors concluded Desyncra was as effective as CBT or more so. With no placebo arm, it cannot show what the sequence itself adds; RESET2 tested that. A 2017 systematic review of eight studies and 329 patients judged the evidence insufficient for clinical use.
The five side by side
- Lenire — tongue stimulation with sound; De Novo, 2023. Tested against its own sound-only weeks. Missing: a sham.
- Neosensory Duo — wrist vibration with tones; no FDA record found; no longer sold. Tested by the company against no wristband. Missing: a sham wristband and an independent trial.
- Levo — sound through earbuds, tested during sleep; cleared as a masker, 2014 and 2026. Tested against other sleep devices. Missing: a no-device or placebo group.
- Neuromonics — customised music with counselling; cleared as a masker, 2005. Tested by its developers against counselling. Missing: an independent randomised trial against sound generators.
- Desyncra — coordinated reset tones; cleared as a masker, 2016. Its method tested double-blind, on an earlier device, against a placebo sequence: no difference.
Before paying for a device
- Have the tinnitus assessed first, particularly if it pulses, sits in one ear, or came with a sudden change in hearing; when to seek care sorts urgent from routine, and sudden hearing loss is where days matter.
- Start with what the guideline suggests: CBT by a trained provider for bothersome tinnitus, and hearing aids where there is hearing loss. The treatment comparison sets the options side by side.
- Test ordinary sound first. Free background sounds or a bedside sound machine can show whether sound helps you, and the VA/DoD guideline finds tinnitus-specific devices not proven better than ordinary sound generators. Sound therapy covers how to use it.
- Ask the seller what the device was compared against, whether there was a sham, who paid for the research, and what happens to the treatment if the company stops trading, as Neosensory did.