psilobot.vercel.app/stellar The STELLAR-OSC V3.5 [SILK] is a multi-track additive/subtractive bassline synthesizer and 16-step matrix sequencer node. Engineered for aggressive sub-bass generation, drone layers, and high-resonance industrial transients, it combines real-time Web Audio API signal processing with live visual monitoring via an integrated oscilloscope canvas.Core DSP & Signal Flow Architecture[ 5-Track Matrix (Sine Oscs) ] ──> [ Master Gain ] ──> [ WaveShaper (Analog Heat) ] ──> [ Post-Dist Gain ] │ [ Oscilloscope / Destination ] <── [ Dynamics Compressor ] <── [ Delay Feedback Loop ] <───────┴──> [ Delay Node ] 1. Additive Frequency Architecture & Scale QuantisationThe sound engine uses a 5-track harmonic oscillator stack anchored around a sub-bass root frequency of $55\text{ Hz}$ ($A_1$). Each track is assigned a harmonic multiplier:$$f_{\text{base}} = 55\text{ Hz} \times (i + 1) \quad \text{for track } i \in \{0, 1, 2, 3, 4\}$$Track Breakdown:BASS_OSC: $55\text{ Hz}$ (Sub / Fundamental)SUB_DRONE: $110\text{ Hz}$ (Low Mid / First Harmonic)PULSE_OSC: $165\text{ Hz}$ (Mid Crunch / Second Harmonic)BELL_OSC: $220\text{ Hz}$ (Upper Mid / Third Harmonic)HIGH_REZ: $275\text{ Hz}$ (High Presence / Fourth Harmonic)When steps are triggered, pitch offsets are real-time quantised to four modal scales (Major, Minor, Phrygian, Lydian) using pitch-class index mapping:$$f_{\text{target}} = f_{\text{base}} \times 2^{\frac{\text{scale}[n] + (12 \times \text{octave})}{12}}$$2. Non-Linear Saturation Stage (makeDistortionCurve)Analog warmth and distortion are synthesized through an arc-tangent non-linear transfer function applied to a native WaveShaperNode.Transfer Curve Formula:$$y(x) = \frac{(\pi + k) \cdot x}{\pi + k \cdot \vert{}x\vert{}}$$Where $k$ represents the drive intensity ($0 \le k \le 50$).Harmonic Generation: As drive amount ($k$) scales, the transfer curve aggressively flattens input sine waves toward square-wave profiles, generating rich odd-order harmonics.Automatic Gain Compensation: To prevent clipping before the dynamic compressor stage, a dedicated postDistGain node applies dynamic compensation scaling ($1.0 + \frac{k}{30}$).3. Temporal Engine & Human Swing SchedulerThe sequencing clock calculates interval step times ($T_{\text{step}}$) based on standard 16th-note subdivisions:$$T_{\text{step}} = \frac{60}{\text{BPM}} \times \frac{1}{4}$$Swing Implementation: Odd-numbered steps (off-beats) are delayed using a variable offset fraction ($\Delta t = \text{swing} \times T_{\text{step}}$), pushing off-beats forward up to $50\%$ of a step duration without disrupting master clock sync.Enveloping & Void Gate: Tone triggers utilize extremely fast linear attack ramps ($5\text{ ms}$) paired with variable exponential decay curves controlled by the VOID_GATE parameter, enabling tight staccato bass plucks or elongated drone tails.4. Real-Time Oscilloscope Rendering LoopWaveform analysis is handled via a 2048-sample AnalyserNode in the time domain (getByteTimeDomainData). The HTML5 Canvas reads time-domain values frame-by-frame using requestAnimationFrame, rendering raw voltage fluctuations to visually verify analog saturation levels and gate lengths in real time.Key Specifications & PerformanceParameterSpecificationDSP ImplementationOscillator Core5-Voice Sine ArrayIndependent OscillatorNode instances per stepRoot Tuning$A_1 = 55\text{ Hz}$ MultipliersExtended across 2-octave pitch offset rangesDistortion EngineArc-Tangent Soft/Hard ClippingNon-linear lookup table via WaveShaperNodeDynamics ControlBrickwall Peak CompressorThreshold $-1\text{ dBFS}$, Ratio $20:1$, Attack $1\text{ ms}$Spatial EffectFeedforward DelayFixed $150\text{ ms}$ time with variable feedback gainExport CaptureClient-side MediaRecorderDirect WebM/WAV stream output via MediaStreamDestinationEngineering Strengths & Optimization OpportunitiesStrengthsAggressive Sub-Bass Stacking: Layering 5 fundamental frequency multipliers produces an enormous, full-spectrum bass output capable of rattling subwoofers while retaining mid-range bite.Integrated Dynamics Protection: A hard-knee compressor ($20:1$ ratio, $-1\text{ dBFS}$ threshold) is placed after the distortion stage, preventing digital clipping and speaker overload when all 5 tracks fire simultaneously.Probabilistic Randomization: The randomizePattern algorithm uses weighted probabilities for downbeats versus off-beats, generating instant industrial rhythms that fit natural club music structures.Recommended UpgradesOversampling on Distortion: The current WaveShaperNode lacks an explicit oversampling configuration. Adding distNode.oversample = '4x' would drastically reduce aliasing artifacts when pushing high distortion levels on high-register oscillators.Selectable Waveforms: Switching secondary oscillators (PULSE_OSC or HIGH_REZ) from sine to saw or square waves would produce aggressive 303-style filter sweeps when combined with lowpass filter controls.