Masking protocol simulator — from cochlear deafferentation and central gain to mixing-point titration and habituation
Subjective tinnitus — the perception of sound without an external source — begins peripherally but is sustained centrally. Noise exposure or age-related hair cell loss reduces afferent output from the cochlea; the central auditory system responds by turning up its own gain, and this maladaptive homeostatic plasticity, not the ear itself, is what generates and maintains the phantom percept.
Cochlear hair cells — particularly outer hair cells tuned to 3–8 kHz, the frequencies most vulnerable to acoustic trauma and presbycusis — die or lose synaptic contact with auditory nerve fibers (cochlear synaptopathy, sometimes called "hidden hearing loss" because standard audiograms can appear near-normal). The reduction in afferent drive removes a normally inhibitory, calibrating signal from the central auditory pathway.
Just as visual cortex increases gain when deprived of retinal input, the central auditory system responds to reduced cochlear output with compensatory gain increase at multiple relay stations: cochlear nucleus, inferior colliculus, medial geniculate body, and auditory cortex. This is homeostatic plasticity — a mechanism that normally keeps neural firing rates within a useful dynamic range — but when it amplifies noise in a now-silent channel, the amplified spontaneous activity is perceived as sound. This is the central gain theory of tinnitus.
The dorsal cochlear nucleus (DCN), the first central relay station after the auditory nerve, shows some of the clearest physiological correlates of tinnitus in animal models. Following acoustic trauma, DCN fusiform cells show:
• Increased spontaneous firing rate — often 2–3× baseline in the frequency region corresponding to the hearing loss edge • Increased burst firing and neural synchrony across neighboring units, which may be more perceptually salient than rate increases alone • Reduced synaptic inhibition, partly from loss of glycinergic and GABAergic inputs that normally dampen spontaneous activity • Altered somatosensory input from the dorsal column and trigeminal system, which converges on the DCN and can modulate tinnitus loudness with jaw clenching or head movement in some patients
This hyperactivity then propagates upstream to the inferior colliculus and auditory cortex, where it is further amplified and where it acquires spectral and temporal structure resembling a perceived tone or noise band.
Auditory cortex is normally organized tonotopically — a smooth map from low to high frequencies across the cortical sheet. Hearing loss at a specific frequency causes cortical neurons that previously responded to that frequency to become responsive to adjacent, still-intact frequencies — map reorganization that concentrates neural resources at the edge of the lesion, precisely the frequency region most tinnitus patients match their percept to.
Pawel Jastreboff's neurophysiological model (1990) reframed tinnitus as a three-system problem, not purely an auditory one: (1) the auditory pathway generates an abnormal neural signal, (2) the limbic system (amygdala, anterior cingulate cortex, insula) attaches emotional valence and threat salience to that signal, and (3) the autonomic nervous system produces the associated stress response. Critically, Jastreboff argued that most people with measurable tinnitus-like neural activity never perceive it consciously or find it bothersome — it is the limbic/autonomic reaction, learned through classical conditioning, that converts a neural signal into suffering. This model underlies Tinnitus Retraining Therapy: habituation targets the limbic reaction, not the cochlea.
In Jastreboff and Hazell's original framework, an estimated 80–85% of adults have detectable subclinical tinnitus-like neural activity if tested in a silent chamber, yet only 10–15% ever consciously perceive tinnitus and just 1–3% find it severely distressing — evidence that the limbic/attentional filter, not the peripheral signal itself, is the decisive variable.
Before any sound therapy is prescribed, a structured psychoacoustic workup characterizes the tinnitus percept itself. Pure tone audiometry establishes the hearing loss contour, pitch matching narrows the tinnitus frequency, loudness matching quantifies its perceived intensity relative to threshold, and minimum masking level (MML) testing determines how much external noise is required to render the tinnitus inaudible.
Air-conduction thresholds are measured at octave intervals from 250 Hz to 8 kHz (often extended to 12–16 kHz for suspected noise damage), with bone-conduction thresholds obtained where a conductive component is suspected. The resulting audiogram typically shows a high-frequency sloping loss in noise-induced and presbycusic tinnitus, with the steepest drop — the "edge frequency" — frequently coinciding with the tinnitus pitch match.
A notch-shaped dip centered at 4 kHz (the classic "noise notch") is a hallmark of occupational or recreational noise exposure and strongly predicts a tinnitus pitch match in the 3–6 kHz range. Audiometric findings directly inform which frequency region a notched-music or targeted masking protocol should center on.
Pitch matching presents sequential pure tones (or narrow bands of noise) and asks the patient to select the closest match to their tinnitus percept, typically using a two-alternative forced-choice or bracketing procedure across octaves and then half-octaves. Because tinnitus is often broadband or has poorly defined pitch, results can show octave confusion — patients frequently select a tone one octave removed from their true percept, so clinicians confirm convergence across at least two independent trials.
Loudness matching then presents the matched-frequency tone at increasing intensity in 1 dB steps until the patient reports equal loudness to their tinnitus, expressed in dB sensation level (dB SL) above their threshold at that frequency. Despite tinnitus often being described by patients as "unbearably loud," objective loudness matches are typically only 5–10 dB SL — underscoring that tinnitus-related distress is driven far more by emotional salience and attentional capture (the limbic component) than by acoustic intensity.
Minimum masking level is determined by presenting broadband noise (typically white or speech-shaped noise) to the ear, increasing intensity in small steps from below threshold until the patient reports the tinnitus is just completely masked — inaudible in the presence of the external sound. MML is recorded in dB SL and is a key benchmark distinguishing patients who mask easily (MML close to loudness match, often good sound-therapy candidates) from those with a large gap between loudness match and MML (harder to mask, sometimes indicating a stronger central/limbic component).
Immediately following a masking period, residual inhibition (RI) is assessed: does the tinnitus remain suppressed, reduced, or unchanged for a period after the masking noise is switched off? A positive RI response (temporary suppression lasting seconds to minutes, occasionally longer) is both diagnostically informative and prognostically favorable — it demonstrates that the auditory pathway is capable of a period of quiescence, and predicts a more favorable response to sustained sound therapy.
Three complementary sound-based strategies dominate clinical practice, each exploiting a different mechanism: broadband noise generators simply cover the tinnitus acoustically, notched music therapy exploits lateral inhibition to suppress activity at the tinnitus frequency, and Tinnitus Retraining Therapy combines low-level noise with structured counseling to drive long-term habituation rather than momentary relief.
The simplest approach delivers steady-state broadband noise — white noise (flat power spectral density), pink noise (equal energy per octave, perceived as more natural), or speech-shaped noise — via combination hearing aids with integrated noise generators, dedicated wearable sound generators, bedside sound machines/pillows, or smartphone apps. For patients with concurrent hearing loss (the majority), combination devices are preferred: amplification restores audibility and can itself reduce tinnitus salience by restoring competing environmental sound, while the integrated masker adds a controllable layer of noise.
Commercial platforms such as the Widex Zen program add "fractal tones" — algorithmically generated, non-repeating, chime-like tone sequences designed to be relaxing and to avoid the habituation-resistant monotony of static noise. These are typically used adjunctively for relaxation-focused sessions rather than as the primary masking stimulus throughout the day.
Pantev and colleagues (2010, 2012) proposed a fundamentally different mechanism: rather than masking tinnitus with noise, remove a 1-octave frequency band centered on the individually determined tinnitus frequency from the patient's own preferred music, then have them listen for at least one hour per day over 12 months. The theory rests on lateral inhibition — auditory cortex neurons tuned to a given frequency are normally suppressed by activity in neighboring frequency channels. By eliminating energy at the tinnitus frequency while preserving normal stimulation in flanking frequencies, notched music is hypothesized to enhance lateral inhibitory input onto the hyperactive, deafferented cortical zone, reducing its excitability.
Pantev's original trial reported significant reductions in both subjective tinnitus loudness and objective auditory-evoked cortical activity (slow cortical potentials) at the tinnitus frequency after 12 months of notched-music listening, with the effect being frequency-specific — non-tailored, randomly notched music did not produce equivalent benefit. Subsequent replications have shown more mixed effect sizes, and current evidence positions notched music as a promising, low-risk adjunct rather than a stand-alone cure — most effective in tonal tinnitus with a well-defined pitch match and no more than mild-to-moderate hearing loss at the notch frequency.
Tinnitus Retraining Therapy (TRT), developed by Pawel Jastreboff and Jonathan Hazell, is not primarily a masking technique — it is a habituation protocol built on the neurophysiological model. TRT has two inseparable components:
1. Directive counseling: structured education explaining the neurophysiological model to the patient — that tinnitus is a real neural signal but one that can be reclassified by the brain as neutral, non-threatening background information, breaking the conditioned reflex arc linking the tinnitus signal to the limbic/autonomic stress response.
2. Sound therapy at the mixing point: low-level, broadband, ear-level noise generators (or environmental enrichment sound) are worn most waking hours, set not to mask the tinnitus but to partially reduce the perceptual contrast between tinnitus and background — enriching the auditory environment so the auditory pathway is never in complete silence, which is itself thought to slow the deafferentation-driven gain increase.
TRT is a long protocol by design: Jastreboff's published series report treatment courses of 12–24 months, reflecting the time believed necessary for genuine synaptic and perceptual habituation, as opposed to the immediate but temporary relief produced by simple masking.
The mixing point is the single most important titration parameter in TRT: the sound level at which the tinnitus percept and the background masking noise become perceptually blended into a single auditory object, without the noise being loud enough to fully cover — mask — the tinnitus. Setting the level too high (full masking) removes the tinnitus from awareness but, per the neurophysiological model, prevents the habituation process the therapy depends on.
It is tempting to simply raise noise level until the tinnitus disappears entirely — this is trivially achievable in most patients above their minimum masking level. But the Jastreboff model predicts that continuous full masking removes the opportunity for the auditory and limbic systems to reprocess the tinnitus signal as neutral: if the brain never has to encounter and reclassify the tinnitus-associated neural activity because it is acoustically obliterated, no extinction of the conditioned limbic response occurs, and tinnitus perception typically rebounds as soon as the masking noise is removed.
Instead, TRT protocol sets the sound generator level so that the tinnitus remains audible but blends with the background noise — patients often describe this as the tinnitus becoming "part of" the noise rather than standing out against silence. This partial-masking, sub-threshold titration is the mixing point, and it is re-checked and adjusted over the course of therapy as habituation progresses and tolerance for a lower noise floor increases.
Mixing point titration is performed with the same broadband stimulus intended for daily use (typically speech-shaped or white noise from ear-level generators), starting well below the patient's hearing threshold and increasing in ~1 dB steps while the patient reports the relationship between tinnitus and noise: initially the tinnitus is clearly distinguishable "in front of" the noise; as level rises, the two percepts begin to overlap spectrally and spatially; the mixing point is marked at the lowest level where the patient reports they can no longer clearly separate the two sounds, while the tinnitus is still faintly detectable.
Jastreboff's original clinical protocol stratifies patients into categories (0 through IV) based on the presence of hyperacusis, significant hearing loss, and severity of the reaction, which determines whether treatment leads with sound generators, hearing aid amplification, or a combined approach, and how aggressively the mixing point is approached in early sessions versus built up gradually.
Effective masking noise is not necessarily flat white noise. Because masking efficiency depends on spectral overlap between the masker and the tinnitus percept's frequency content (the "critical band" around the tinnitus pitch contributes disproportionately to perceived masking), many modern sound generators allow spectral shaping — tilting energy toward high frequencies for the common high-pitched noise-induced tinnitus profile, or notching out frequencies where residual hearing is poor and amplification would otherwise cause discomfort or feedback.
A masking stimulus with energy concentrated near the tinnitus frequency achieves adequate masking at a lower overall sound pressure level than flat-spectrum noise of the same loudness elsewhere in the spectrum — an important consideration for patients with hyperacusis or reduced dynamic range, where excessive broadband loudness itself becomes aversive.
Formby, Sherlock, and colleagues (2003, 2015) showed that simply increasing a tinnitus patient's daily exposure to soft background sound — even without any structured TRT counseling — measurably expanded their loudness discomfort level (dynamic range) within weeks, supporting the principle that auditory deprivation itself (silence) drives central gain upward, and that any consistent sound enrichment, properly titrated, works partly by reversing that process.
Sound therapy outcomes are tracked with validated, standardized questionnaires rather than relying solely on subjective loudness reports, because the disability caused by tinnitus is driven mainly by its emotional and functional impact, not its acoustic intensity. The Tinnitus Handicap Inventory and Tinnitus Functional Index are the two most widely used instruments in both clinical practice and trials.
Developed by Newman, Jacobson, and Spitzer (1996), the THI is a 25-item self-report questionnaire scored 0–100, with items grouped into functional, emotional, and catastrophic subscales. Each item is answered "yes" (4 points), "sometimes" (2 points), or "no" (0 points), and total scores are graded into severity bands: slight (0–16), mild (18–36), moderate (38–56), severe (58–76), and catastrophic (78–100).
Baseline THI in patients presenting to specialty tinnitus clinics commonly falls in the severe range (58–76), reflecting a referral population with significant functional and emotional impact — concentration difficulty, sleep disruption, irritability, and in the catastrophic band, despair and inability to function. Longitudinal TRT and structured sound-therapy series report mean THI reductions bringing group scores into the mild range (below 18–36) after 12–18 months of consistent treatment, though the pace of improvement varies widely by baseline severity, hyperacusis status, and psychiatric comorbidity.
The Tinnitus Functional Index (Meikle et al. 2012) was designed specifically to be sensitive to treatment-related change, a property the THI was not optimized for. Its 25 items span eight subscales — intrusiveness, sense of control, cognitive interference, sleep, auditory (listening) difficulties, relaxation, quality of life, and emotional distress — each scored 0–10 and averaged into a 0–100 composite. A change of roughly 13 points on the TFI is generally considered the threshold for a clinically meaningful treatment response, compared with roughly 7 points on the THI.
Across published TRT case series and controlled comparisons (Jastreboff and Jastreboff, Henry et al.), 60–80% of patients completing a full 12–24 month course report meaningful subjective improvement, most commonly a marked reduction in tinnitus-related annoyance and improved sleep and concentration rather than elimination of the percept itself — consistent with the model's prediction that the goal is reclassification and habituation of the signal, not its removal.
Henry, Schechter, and colleagues' prospective TRT outcome studies found that patients categorized with concurrent hyperacusis (Jastreboff category II/III) required visibly longer courses to reach comparable THI reductions than those with tinnitus alone — underscoring why individualized categorization and mixing-point titration, not a one-size-fits-all noise level, drive real-world response rates.
Beyond questionnaire scores, repeated residual inhibition (RI) testing over the therapy timeline provides an objective, session-level marker of the auditory pathway's changing responsiveness. Early in treatment, RI duration after a masking trial is often brief (seconds) or absent; as central gain normalizes and habituation progresses, many patients show progressively longer RI durations, and some eventually report spontaneous, unprompted quiet periods even without recent masking exposure — an encouraging sign correlated with, though not a substitute for, formal THI/TFI improvement.
The endpoint of successful sound therapy is not silence but indifference: the tinnitus signal continues to exist at the level of the auditory pathway, but perceptual and emotional habituation mean it no longer reaches conscious attention or triggers a limbic stress response under most circumstances. Maintaining this state over years requires planned tapering, relapse-triggers education, and periodic reinforcement.
As habituation consolidates — typically indicated by sustained THI/TFI improvement and lengthening residual inhibition — the reliance on a dedicated ear-level sound generator is gradually tapered in favor of general environmental sound enrichment (fans, soft music, nature sound apps used only as needed) rather than structured all-day device wear. This mirrors the underlying goal: it was never the noise itself that was therapeutic in the TRT model, but the sustained avoidance of silence combined with counseling-driven reclassification of the tinnitus signal as unimportant.
Patients are counseled that intermittent tinnitus awareness after discharge is expected and not a treatment failure — the clinical target is a return to a state functionally indistinguishable from pre-tinnitus life, tolerating brief awareness (for example in a very quiet room at night) without renewed distress or catastrophizing.
Even after successful habituation, tinnitus loudness and salience commonly fluctuate with systemic stress, poor sleep, upper respiratory infections and associated eustachian tube dysfunction, new noise exposure, ototoxic medication use, and psychological stressors. Relapse-prevention counseling explicitly reviews these triggers in advance so that a temporary flare is recognized as an expected, self-limited fluctuation rather than a sign that therapy has failed — patients who reinterpret a flare catastrophically are at highest risk of re-establishing the original limbic conditioning that TRT worked to extinguish.
Practical guidance typically includes maintaining hearing protection in loud environments (a flare after unprotected noise exposure is common and reinforces the deafferentation-central-gain mechanism from Stage 1), resuming brief sound-enrichment use during high-stress periods, and having a clear, low-friction pathway back to clinical review rather than defaulting to alarm.
Longitudinal follow-up of TRT cohorts (Jastreboff and Jastreboff; Bauer and Brozoski for the tonotopic/behavioral analogue in animal models) reports that a substantial majority — commonly cited above 80% of initial responders — maintain their improvement at three to five years post-discharge, particularly when periodic maintenance check-ins (roughly every 6–12 months) are scheduled to reinforce counseling messages and catch early relapse triggers before they consolidate into renewed distress.
Where relapse does occur, a shortened "booster" course — typically a few sessions of counseling reinforcement plus reintroduction of ear-level sound enrichment at a re-titrated mixing point — is usually sufficient to restore habituation, since the neural and behavioral groundwork from the original course is not lost but rather requires reactivation.