1. Medical & Biophysical Overview
Tinnitus is the continuous perception of sound in the human auditory cortex entirely in the absence of an external acoustic stimulus. Commonly manifesting as an intractable high-pitched ringing, buzzing, or hiss, subjective tinnitus almost exclusively originates from peripheral hearing loss or mechanical deafferentation of the inner hair cells (IHCs) along the basilar membrane of the cochlea. When input from specific frequency bands is permanently lost due to acoustic trauma, ototoxic drugs, or presbycusis, the central auditory system attempts to radically compensate by altering the biochemical balance of neural excitation and inhibition along the auditory pathway.
In both the primary auditory cortex ($A1$) and the dorsal cochlear nucleus (DCN), this deafferentation induces a form of maladaptive neural plasticity. The physiological consequence is an environment characterized by hyper-synchronous spontaneous neural firing and the abnormal expansion of receptive fields into adjacent, functional tonotopic regions. This highly localized cortical hyperactivity, largely driven by the down-regulation of inhibitory $\gamma$-aminobutyric acid (GABA) and the up-regulation of excitatory glutamate, is experienced subjectively by the patient as a continuous, inescapable phantom sound defined precisely at their characteristic loss frequency $f_T$.
Tailored Notched Music Therapy (TMNMT) and custom-notched white noise acoustic therapies directly target this biological mechanism by exploiting the architecture of the auditory cortex. By presenting a continuous acoustic noise floor or musical spectrum from which a narrow frequency band—centered precisely around the patient's phantom tinnitus pitch $f_T$—has been surgically removed via digital signal processing, the surrounding functional neurons receive intense acoustic stimulation while the hyperactive target band receives absolutely zero external acoustic energy.
This profoundly asymmetric acoustic input effectively tricks the brain into engaging lateral GABAergic inhibitory interneurons. Because the surrounding tonotopic map is highly active, it naturally suppresses adjacent map regions. The un-stimulated region at $f_T$ therefore receives powerful lateral inhibition from its neighbors, actively down-regulating the spontaneous hyper-synchrony and inducing long-term synaptic depression. Over successive weeks of training, this leads to a measurable reduction in both subjective tinnitus volume and distress.
2. Interactive Visualizer & Controls Guide
This single-page interactive laboratory provides real-time dual-domain modeling of the therapeutic intervention. The visualizer is divided horizontally into two dynamic sectors: an upper Acoustic Spectrum & Filter Transfer Function graph tracing the audio modifications, and a lower Tonotopic Auditory Cortex ($A1$) Neural Field Visualizer showing real-time excitatory-inhibitory network reactions.
- Therapy Configuration Preset: Start your exploration by selecting pre-configured clinical profiles representing typical patient profiles. These include the standard $4,000\text{ Hz}$ tonal ringing, a broad high-frequency hiss typical of aging at $8,000\text{ Hz}$, or the sharp $6,500\text{ Hz}$ loss commonly associated with sudden acoustic trauma (e.g., gunfire or industrial noise).
- Phantom Tinnitus Pitch ($f_T$): This slider adjusts the central target frequency under investigation, spanning from $1,000\text{ Hz}$ to $12,000\text{ Hz}$. You can instantly match pitch by tapping or dragging directly anywhere on the visualizer canvas. On the upper acoustic canvas, $f_T$ is distinctly marked by a glowing crimson line. In the lower cortical map, you will visibly note that unmitigated tinnitus generates highly localized, fast-firing hyper-synchronous spikes around this precise location.
- Enable Acoustic Band-Stop Notch Filter: This toggle activates a digital biquad notch filter dynamically inserted into the Web Audio processing pipeline. Disabling the notch causes the auditory cortex to receive full broadband energy. Without the notch, the lateral inhibition cannot isolate the deafferented target region, leaving phantom hyper-active neurons uninhibited.
- Notch Quality Factor ($Q$): Modulates the selectivity, or structural sharpness, of the digital notch filter. High $Q$ values create ultra-narrow "surgical" notches, while lower $Q$ values broaden the spectral attenuation window, sacrificing precision to catch fluctuating phantom pitches across wider adjacent octaves.
- Masker Audio Spectrum: Toggles the baseline acoustic carrier spectrum between Pink Noise ($S(f) \propto 1/f$), White Noise ($S(f) = \text{constant}$), Brownian Noise ($S(f) \propto 1/f^2$), or a pure tone matching test to help the user identify their specific harmonic resonance point before therapy begins.
- Lateral Inhibition Gain ($\gamma$): Simulates the physiological strength of cortical GABAergic interneuron suppression. Increasing this slider mathematically deepens the suppression curve around the $f_T$ neighborhood.
- Audio & Demo Execution: Click the
#soundToggleBtn at the top of the controls to start the real-time Web Audio API synthesizer. For an automated, physiologically paced walkthrough of the phenomena, click Start Demo. The system will independently simulate acoustic pitch-matching, parameter optimization, and habituation.
3. Mathematical Foundations & Engineering Details
The underlying digital signal processing pipeline utilizes a second-order Infinite Impulse Response (IIR) biquad notch filter node. In the continuous-time analog domain, this filter's transfer function $H(s)$ is governed by the following ratio of polynomials:
$$H(s) = \frac{s^2 + \omega_0^2}{s^2 + \left(\frac{\omega_0}{Q}\right) s + \omega_0^2}$$
where $\omega_0 = 2\pi f_T$ represents the center rejection frequency in radians per second, and the parameter $Q$ is the quality factor determining the overall width of the attenuation bandwidth at the half-power points: $\Delta f = \frac{f_0}{Q}$. To process audio in real time within the browser, this continuous system is mathematically mapped into the discrete $Z$-domain utilizing a bilinear transformation, yielding the following recursive difference equation executed upon every audio sample:
$$y[n] = \frac{b_0 x[n] + b_1 x[n-1] + b_2 x[n-2] - a_1 y[n-1] - a_2 y[n-2]}{a_0}$$
Simultaneously, the simulation maps the physiological reaction of the primary auditory cortex. The cortex is computationally discretized into $N$ tonotopic cortical columns distributed logarithmically across frequency space $f(i) = f_{\text{min}} \cdot \left(\frac{f_{\text{max}}}{f_{\text{min}}}\right)^{i/N}$, matching human basilar membrane mapping mechanics. The local neural activation potential $u(x, t)$ within position $x$ follows a highly dynamic Wilson-Cowan lateral inhibition field model, described as:
$$\frac{\partial u(x,t)}{\partial t} = -u(x,t) + \int_{0}^{L} w(|x - x'|) \cdot \sigma\big(u(x',t)\big) \, dx' + I_{\text{acoustic}}(x,t) + \eta_{\text{phantom}}(x)$$
Crucially, the spatial connectivity kernel $w(y)$ defines a "Mexican-Hat" interaction profile, balancing short-range excitation and long-range inhibition via $w(y) = A_e e^{-y^2 / 2\sigma_e^2} - \gamma A_i e^{-y^2 / 2\sigma_i^2}$. When external acoustic input $I_{\text{acoustic}}(x,t)$ at spatial position $x \approx f_T$ is radically diminished to zero by our digital notch filter, while the surrounding regions $x \neq f_T$ continue receiving high broadband excitation, the negative inhibitory tail of the integral kernel $-\gamma A_i$ mathematically dominates at the target site $f_T$. This targeted influx of inhibitory pressure forcibly drives the local neural activity down toward its baseline resting potential, successfully silencing the $\eta_{\text{phantom}}$ spontaneous spiking.
4. Future Directions & Clinical Roadmap
While the current architecture accurately models single-tone subjective tinnitus and traditional custom-notched masking noise, future deployments of this interactive lab will introduce significantly more advanced clinical techniques. We are actively developing multi-frequency notch synthesis logic to simulate compound tinnitus profiles consisting of multiple harmonic resonance peaks or widespread cochlear dead regions.
Additionally, upcoming versions will feature true binaural phase-cancellation algorithms (utilizing customized interaural time and level differences) to investigate brainstem spatial tracking of tinnitus lateralization. Finally, we are laying the engineering groundwork for direct Web-Bluetooth (WebBLE) hardware integration. This will allow the simulator to synchronize the acoustic audio output with tactile bimodal stimulation devices—such as tongue, jaw, or cervical nerve stimulators—which have recently proven highly effective at disrupting the dorsal cochlear nucleus's maladaptive plasticity.
Related Native Environments on BioniChaos
-
Cochlear Simulation Framework
Explore a highly detailed biomechanical finite-element model of basilar membrane traveling waves and fluid-structure cochlear dynamics.
-
Cochlear Insertion Mechanical Simulator
Analyze the surgical path mechanics and scala tympani tissue resistance forces for multi-channel electrode arrays.
-
Real-Time Acoustic Spectrogram Lab
Utilize a high-resolution Fourier transform visualizer for dynamic sound spectrum analysis, ambient noise tracking, and harmonic isolation.
-
ICA PCA Brain EEG Signal Mixer
Experiment with blind source separation and independent component analysis on simulated neurological cocktail party multi-channel streams.