Eight J-SONIC audio plugins for dynamics, de-essing, stereo width, overdrive, transformer coloration, phase alignment, and adaptive reverb/delay. Built with JUCE and C++17, with VST3 and Standalone targets on Windows, macOS, and Linux, plus Audio Units on macOS.
Collection version: 0.5.0. Hover over any plugin's title to see its developer
(J-Sonic Team), release year (2026), and version. The shared release
version is defined in projects/PluginVersion.cmake.
The editors support 50%–120% resizing, one-sentence hover hints, host automation,
and session recall. Each plugin is an independent project under projects/;
the JUCE framework is included in this repository.
Plugin collection · Measured examples · Build and install · Detailed build guide
| Plugin | Project / build target | Purpose | Main features |
|---|---|---|---|
| Yinyang Compressor | YinyangCompressor |
Dynamics | Four compressor models, Yin/Yang operation, sidechain shaping, transformer color |
| Shun De-Esser | ShunDeEsser |
Sibilance control | Four attenuation modes, relative detection, separate detector and attenuator tuning |
| Kuo Stereo | KuoStereo |
Stereo widening | Adaptive Haas delay, Velvet convolution, spectral transient protection, three-band sound-field meters |
| Zao Drive | ZaoDrive |
Overdrive | Four circuit models, Smart harmonic shaping, Clear contour, waveform preview, Auto compensation |
| Ren Transformer | RenTransformer |
Transformer coloration | Yinyang's transformer stage in a compact four-knob interface |
| Kuo Velvet | KuoVelvet |
Convolution widening | Kuo's Velvet stage with Width, Transient, Attack, and Output controls |
| Xuan Phase | XuanPhase |
Phase rotation | Independent low/high rotation, 25 ms lookahead, spectral transient alignment and Auto Depth |
| Lan Reverb | LanReverb |
Reverb / delay | Long stereo tails, live feedback-tap scattering, automatic resonance EQ and phase dispersion |
Choose FET, VCA, Variable-mu, or Optical behavior to shape attacks, sustain, and recovery. Yang reduces levels above threshold; Yin raises quieter material toward it.
- Tune Threshold, Ratio, Knee, Attack, and Release, with VariTransient to adapt timing as downward gain reduction increases.
- Shape the detector with a high-pass filter and SC High Tilt; use stereo linking and Mid/Side operation to control the stereo response.
- Add optional oversampled transformer coloration with Core Drive and DC Bias, while monitoring the gain meter and scrolling history.
- Process the complete signal path at 4× with one shared converter and a fixed 64-sample host latency.
Controls and signal path · Measurements
Control sibilance with separate tuning for what the plugin detects and what it attenuates. The response plot shows both curves over a live input spectrum beside the gain-reduction display.
- Wideband reduces the full signal; Split-band concentrates attenuation around the selected band.
- Gold bands show the smart attenuator’s active detail processing in Brilliant and Smooth.
- Brilliant restores detected spectral detail after smart attenuation; Smooth reverses that restoration to suppress peaks further.
- Select Absolute detection for a fixed threshold, Listen to audition the detector band, or Relative to compare its energy with the full signal.
Controls and modes · Detector and attenuator measurements
Combine content-adaptive timing with convolution texture, then inspect the resulting stereo field across three frequency bands.
- Smart Haas chooses signed delay times in 0.2 ms steps within Center Delay ± Variability, scoring the mid spectrum and useful side energy. Max Rate limits delay movement; zero Variability switches the search off to save CPU.
- Dynamic shortens Haas delay around detected transients. Velvet Width adds sparse convolution texture, with spectral Transient protection and adjustable Attack to keep onsets focused.
- Three goniometers show low, mid, and high stereo activity. Their adjacent 0%–200% sliders set how much of each band feeds Velvet.
Controls and processing · Audio examples and measurements
Explore BJT common-emitter, op-amp feedback, Schottky diode shunt, and series-diode crossover overdrive. The first knob follows the circuit: Bias for BJT, Input for op-amp, and Knee for the two passive diode modes.
- Smart analyzes generated harmonics against the input using a 12-TET dissonance model, then filters selected wet components. It updates at up to 20 Hz, with 50 ms sigmoid crossfades between filter banks.
- Clear applies a linear-phase 80-phon equal-loudness contour shifted down an octave and lowered so it adds no positive gain. Tone and Mix shape the separate driver-minus-reference wet residual.
- Expected Output renders a 1 kHz triangle through the current processing. Auto derives fixed RMS compensation from that reference; Output adds or subtracts gain from the matched level. An optional Yinyang transformer adds another stage of coloration.
Circuit models · Smart signal path · Auto measurements
Use Yinyang's transformer independently: Core Drive sets saturation, DC Bias introduces asymmetry, Mix blends coloration, and Output sets the level entering the transformer. The shared stage uses 4× oversampling, magnetic hysteresis, and a frequency-dependent transformer response.
Controls and signal path · Processor checks
Add Kuo's convolution width with four controls: Width, Transient, Attack, and Output. Width scales the added side signal while preserving the input mid before output gain. The shared spectral onset detector protects attacks; this compact version omits Haas search, meters, and band controls.
Controls and processing · Processor checks
Rotate the low and high frequencies independently with a combined ±150° cap and a smooth adjustable crossover, 25 ms correction lookahead, and 50 ms total latency. Auto detects clear transients and uses their spectrogram phase and phase slopes to align narrow frequency bands; signed Auto Depth scales or reverses that correction independently of the manual settings, and Transient sets its release time. Stereo Link blends from shared correction through independent channel processing to opposing transient shifts. The side spectrogram moves left to right, colouring early bands green and late bands red, with white left-channel and gold right-channel traces for applied phase.
Controls and signal path · Phase and transient measurements · Auto alignment performance
Create long, wide tails at high wet settings with an eight-line stereo feedback network. Scatter moves from regular delay repeats at zero to dense stereo reverb at full; Time, Decay, and Diffusion shape the result; Tap Shift crossfades feedback read positions while audio is ringing. Auto protects input-note peaks while tracking persistent ringing, applies EQ and phase correction only to feedback taps, and leaves the wet pickup unfiltered. Tame and Phase set their strengths. Drag the spectrum's L/H shelf handles to color the feedback, and add a Shimmer octave layer through a 2× oversampled pitch shifter. The tamer uses narrow cuts within a shared 6 dB budget. The dry path has zero added latency, and Mix reaches 100% wet for sends.
The shelves can boost feedback without compensation for sustained tails. A linked feedback soft limiter controls runaway growth while leaving low-level feedback unchanged; Auto reduces detected ringing.
Controls and signal path · Measurements
At the same threshold, ratio, knee, and timing settings, the four models produce different transfer and gain-reduction curves. These measurements use 48 kHz, a −18 dBFS threshold, 4:1 ratio, and 6 dB knee, with the detector HPF and transformer off. The measurement report also covers attack, recovery, upward compression, and transformer harmonics.
The dry mono signal has identical left and right channels, forming a diagonal line. With Velvet at 100%, its added side signal spreads the stereo field. Haas is removed in this test, isolating the convolution's contribution. The feature report includes transient tests, adaptive Haas measurements, and reproducible audio examples.
The models are developed for creative audio processing. Individual project documents describe their equations, assumptions, and measurement conditions.
Use CMake 3.22+, a C++17 compiler, and the platform's JUCE dependencies. The source includes JUCE; no separate framework checkout is needed. Start from the repository root:
git clone --branch master https://github.com/javelinmaxx/Multi-Compressor.git
cd Multi-CompressormacOS — build all eight plugins as VST3 and AU: install full Xcode and CMake, then run:
./build-macos-all.shUse --arch universal for Apple Silicon and Intel, or add --install to copy
the completed bundles into your user plugin folders. The script builds without
installing by default. See ./build-macos-all.sh --help for all options.
Windows — build VST3: the manual
All Plugins - Windows VST3 workflow
packages the previous seven plugins, including Xuan Phase; build Lan Reverb with its per-project CMake commands. For a local Yinyang build, use
build-windows-vst3.ps1 with Visual Studio 2022's C++
tools. The build guide covers local commands,
workflow artifacts, and installation.
One project — CMake: for example, build Zao Drive's native VST3:
cmake -S projects/ZaoDrive -B build/zao -DCMAKE_BUILD_TYPE=Release -DZAO_DRIVE_BUILD_TESTS=OFF
cmake --build build/zao --config Release --target ZaoDrive_VST3 --parallel 4The complete bundle is under build/zao/ZaoDrive_artefacts/Release/VST3/.
Use each project's own target name when building another plugin. Linux builds
require the JUCE system dependencies.
| Platform | Plugin installation folder |
|---|---|
| Windows VST3 | C:\Program Files\Common Files\VST3\ |
| macOS VST3 | ~/Library/Audio/Plug-Ins/VST3/ |
| macOS AU | ~/Library/Audio/Plug-Ins/Components/ |
| Linux VST3 | ~/.vst3/ |
Copy the entire .vst3 or .component bundle, then rescan plugins in your DAW.
Build a binary for the host's operating system and architecture.
Full build and installation instructions include architecture options, individual project targets, manual CMake commands, and AU validation. Each project's README lists its processor tests, measurement tools, and editor preview commands.
Built with JUCE. JUCE modules are available under AGPLv3 or commercial terms; see LICENSE.md for the framework license and dependency details.









