NOMN

A sound in space is never spectrally uniform. Different frequencies arrive from different directions, reflect off different surfaces, decay at different rates. The perception of spatial life in audio — the sense that a sound is in a place rather than merely from a speaker — arises from these spectral asymmetries. In acoustic environments they happen automatically. Convolution and spatialization tools reproduce and model them. CHAKRABOOST lets you compose with them — not toward realism, but as raw material for effects that have no acoustic equivalent.

Multiband phasers have existed since the 1970s. Split a signal into frequency bands, phase-shift each one, recombine. The topology is old. What hasn't been done is treating each band as a complete processing channel — its own phaser, its own compressor, its own position in space — all driven by the same modulation source so the whole thing breathes as a single organism but behaves differently across the spectrum.

CHAKRABOOST takes this to a curious domain. 8 frequency bands, each with a vactrol-modeled phaser, an adaptive parallel compressor, and a spatial panner with LFO-driven orbits. Your low end can sit still while your upper mids sweep left, your air frequencies orbit overhead, and the whole thing pulses with a single unified rhythm. Or each band moves independently.

The modulation engine is not a periodic LFO. It's a small neural network — a learned prosodic-trajectory model from NOMN's microtiming research, repurposed here as a modulation source — generating continuous, non-repeating curves that move the way speech and song do. Phaser sweeps, pan orbits, and compression timing all run on these trajectories. All 8 bands share the same underlying motion, with a small amount of per-band noise added so the bands are correlated but not identical.

This is an effect, with a useful subtle setting. It will not clean up your mix. It will make a static pad develop internal motion and 3D position. It will turn a drum bus into something that phases and breathes across the spectrum. It will make a vocal sound like it's being transmitted through eight different rooms simultaneously. At low dry/wet settings (15–30%) with Pan Depth at zero it adds subtle multiband phaser character without moving the stereo image. At high settings it's a full reimagining of the source material.

A note on mono: with Pan Depth at zero on every band, CHAKRABOOST preserves the stereo image end-to-end and folds cleanly to mono. Once you enable per-band pan modulation, you're moving frequency content across the stereo field at LFO rate — which means phase cancellation in mono is part of the deal. If your output needs to survive a mono fold, either keep Pan Depth at zero or use the plugin on sends, not across a master bus.
--
## How It Works

Audio enters via a true stereo crossover that splits each channel into 8 phase-coherent frequency bands across 20 Hz to 20 kHz. Two crossover topologies are available, switched by the global Phase Mode parameter:

Zero Latency (default): IIR Linkwitz-Riley 4th-order in a balanced binary tree with allpass phase compensation. Sums flat magnitude. Zero added latency.

Linear Phase: 255-tap windowed-sinc FIR with Blackman window. Sums perfectly flat in both magnitude and phase. ~127 samples auto-compensated latency. Use this when you want the vactrol phaser to be the only source of phase manipulation in the signal — every phase relationship in the output is then a direct consequence of the phaser's settings, not the band split.

Each band passes through three processing stages in series:

Phase modulation. A 4-stage all-pass phaser where each stage is controlled by an independent vactrol model. The vactrol's asymmetric response (fast ~3.5 ms attack, slow 150–1500 ms release with thermal memory) creates the organic, breathing sweep that defined classic analog phasers like the Ibanez FP-777 Flying Pan. Soft-clipped tanh feedback for resonance without runaway.

Adaptive compression. Parallel (NY-style) compression with crest-factor-driven timing. The Adaptivity knob (0–1 per band) controls how much the timing range scales with program material — at 0, attack and release are fixed; at 1, they vary 4× based on the signal's dynamic character. This keeps each band's level stable as the phaser reshapes it, without the compression fighting the modulation.

Spatial output. A global Spatial Mode parameter selects how each band reaches the output:

- Stereo: per-band constant-power panning with X-axis modulation depth.
- Surround: VBAP-style positioning across Quad / 5.1 / 7.1 with 2D X/Y orbital modulation per band, an Orbit Angle that rotates the modulation path, and configurable LFE routing.
- Object3D: per-band binaural rendering with X/Y/Z position. Each band becomes a point in 3D space, sounding best on headphones. Mono-sums per band by design — objects are points, not stereo sources.

The modulation source for all of these is the NeuralLFO — a single learned prosodic-trajectory model running once per buffer, with all 8 bands reading the shared output. Per-band Rhythm Strategy and Articulation knobs steer where each band sits in the model's z-space; per-band Temperature controls how much each band wobbles independently of the shared trajectory.

All bands recombine and a global dry/wet mix blends the processed result with the original signal. An optional output soft-clip (default on) catches over-the-top peaks via tanh saturation; an Output Gain trim (-24 to +24 dB) lets you compensate for level changes from the per-band processing.
--
## When To Use It

On a synth pad or drone. Slow LFO rates, moderate Neural Depth, wide spatial orbits. The pad develops internal motion, different harmonics drifting to different positions. Feels three-dimensional without being obviously effected.

On drums or percussion. Faster LFO, low feedback, subtle spatial depth. Adds a rotating, breathing quality to the transient tail. The adaptive compressor (low Adaptivity for predictable timing) keeps the attack intact while the phaser works the sustain.

On a vocal (committed, not live). Gentle phaser depth on the mid and upper bands only, with light spatial movement. Creates a diffuse, spatialized quality without the obvious swoosh of a traditional phaser. Linear Phase mode is worth a try here — the vocal's phase character becomes purely a function of the phaser's shaping, not the crossover.

On a full mix bus (with caution). Linear Phase mode, dry/wet at 10–20%, Pan Depth at zero, slow modulation, low Adaptivity, low per-band phaser mix. Adds micro-variation across the spectrum without smearing the image or breaking mono. Use the Output Gain to compensate for the per-band processing level.

For headphone-first delivery. Object3D mode places each band at its own X/Y/Z position. Bands move in 3D space along the same shared rhythmic trajectory. Sounds dramatic on headphones; collapses to a narrowed stereo image on speakers (it's mono-summing per band by design).

As a sound design tool. Crank everything. Fast modulation, high feedback, wide orbits, aggressive compression. Turn a field recording into something unrecognizable. The per-band control depth is the point — most phasers don't give you separate phaser, compressor, and 3D position per frequency range.
--
## The Engine

True stereo processing in Stereo mode: independent crossover banks for left and right channels, independent L/R phasers driven by a shared LFO, stereo-linked compression detection (gain reduction computed from `max(|L|, |R|)` per band and applied identically to both channels). 8 parallel processing chains with per-band solo, mute, and full DAW automation across every parameter on every band.

Latency: 0 samples in Zero Latency Phase Mode, ~127 samples (~2.6 ms at 48 kHz) in Linear Phase mode — auto-compensated by your DAW.

The visualization is an 8-lane scrolling waveform display, one lane per band. Real post-processing amplitude envelope, time-aligned, with pan position as vertical displacement within the lane. Click any lane to select that band for editing. The chakra-aligned color palette is colorblind-safe — a warm-to-cool gradient varying both hue and luminance.

A built-in preset system saves and loads named state files (`.chakrapreset` JSON) to your user library; the dropdown at the top of the window selects an existing preset or saves the current state.
--
## Parameters
Global: Input Gain, Output Gain, Dry/Wet, Bypass, Soft Clip, Phase Mode, Spatial Mode, Speaker Layout, LFE Channel + LFE Level.

Per-Band Phaser (×8): Depth, Feedback, Mix, Rate.

Per-Band Spatial (×8): Pan X, Pan Y, Pan Z, Pan Depth (X / Y), Orbit Angle.

Per-Band Compression (×8): Threshold, Ratio, Attack, Release, Comp Mix, Adaptivity.

Per-Band Modulation (×8): Rhythm Strategy, Articulation, Neural Depth, Temperature.

Per-Band Utility (×8): Gain, Solo, Mute.
--
## Features & Formats
Key Features:
- 8-Band Spatial Processing: full-spectrum frequency splitting with per-band phaser, compressor, and spatial panner.
- Two Crossover Topologies: zero-latency IIR LR4 (default) or linear-phase FIR (~127-sample auto-compensated latency, perfectly flat phase response).
- Vactrol Phaser Modeling: authentic LED/LDR optocoupler simulation with asymmetric attack/release, thermal memory, and logarithmic resistance curves.
- NeuralLFO Modulation: shared learned prosodic-trajectory model from NOMN drives all 8 bands. Continuous, non-repeating curves with the timing character of speech rather than periodic LFO.
- Adaptive Parallel Compression: crest-factor-driven timing with a user-controllable Adaptivity knob per band. NY-style parallel mix.
- Three Spatial Modes: Stereo, Surround (Quad / 5.1 / 7.1 with VBAP and orbital modulation), and Object3D (binaural with per-band X/Y/Z position).
- True Stereo Throughout (Stereo Mode): independent L/R crossovers and L/R phasers; stereo image preserved end-to-end with Pan Depth at zero.
- Built-in Preset System: save and load named state files; quick-access dropdown at the top of the window.
- Full DAW Automation: every parameter across all 8 bands.

Compatibility:
Formats: VST3, Audio Unit (AU), Standalone.
OS: macOS 10.15+ (Apple Silicon and Intel).