🎛️ Part of the TizWildin Plugin Ecosystem — 20+ free audio plugins, creator tools, lists, deconstructed-loop routers, and release surfaces.

FreeEQ8 · FreeVox8 · XyloCore · Instrudio · Therum · BassMaid · SpaceMaid · GlueMaid · MixMaid · MultiMaid · MeterMaid · ChainMaid · PaintMask · WURP · AETHER · WhisperGate · RiftWave · FreeSampler · VF-PlexLab · PAP-Forge-Audio

🎧 SoundCloud: TizWildin on SoundCloud — original music, remixes, VIP mixes, experimental drops, and underground releases.

🎁 Free Packs & Samples — jump to free packs & samples

🎵 Awesome Audio — My (FREE) Awesome Audio Dev List

▶️ YouTube — music, visuals, demos, and releases
🌊 Voxel Audio — free RGB waveform visualizer and audio export tool
📘 Facebook Page — TizWildin / GareBearProductionz updates and Media

🎧 Awesome Audio Lists — even more audio lists, submission routes, and promotional directories

_{⭐ Featured / submitted on — full tracker →}

FreeEQ8 is a professional-grade EQ8 / EQ Eight inspired parametric equalizer, free and open-source 8-band parametric EQ plugin for macOS, Linux, and Windows. Linear phase, dynamic EQ, match EQ, per-band drive, band linking, M/S processing, oversampling, and a real-time spectrum analyzer — all in a single, zero-cost plugin. Built with JUCE for VST3 and AU.

ProEQ8 is one of the most advanced EQs available, competing directly with top tools like ZL EQ and TDR Nova.

Works with Reaper! Tested and built on Ableton 10+

🗂️ Release Vault — official monetized releases, distributed tracks, and catalog proof
🗂️ Release Vault Repo: https://github.com/GareBear99/TizWildin-Release-Vault

Is FreeEQ8/ProEQ8 the "Best"?

Yes, it is the best value EQ under $50, and arguably the best all-in-one creative EQ for budget-conscious producers. However, it is not the absolute best for pure surgical precision or workflow speed compared to industry standards "Best", Yet...

When factoring in the Pro version (ProEQ8 24Bands), the answer of quality shifts: FreeEQ8 is no longer the "best" for professional, unrestricted workflows, but the $20 upgrade makes the combined package arguably the best value EQ under $50, though it still trails paid industry standards in band count.

COMING SOON!!!

🔊 Trademark Sound

BaisMaed — play / download Collab sound with - YUME/GR!EF

💫 Thanks to our supporters

Final Verdict (June 2026)

Free Version: Currently Close But Not the absolute best yet. The 8-band limit and export cap make ZL Equalizer 2 the superior choice for zero-cost professional work. Pro Version ($20): The Best Value. By unlocking the 24-band SVF engine, it becomes the most capable "all-in-one" mixing tool for under $50. It is the best choice for producers who want to replace both their main EQ and a saturation plugin with a single $20 purchase. The Next engine will support a new form of quantum processing that will make it the 'Best' and that is why I put these here..

Is ProEQ8 ($20) the Best Value? Yes.

The ProEQ8 upgrade fundamentally changes the plugin's standing by unlocking the 24-band engine. 24-Band Architecture: The Pro version switches to a SIMD-vectorized State Variable Filter (SvfBandArray.h with AVX2/SSE2/Neon), matching the band count of FabFilter Pro-Q 4 and ZL Equalizer 2.

Unique Feature Set:

It is the only EQ under $50 combining:

• 24 Bands (Dynamic & Static)
• Linear Phase mode
• Per-Band Saturation (Tube, Tape, Transistor, Tanh)
• Match EQ, Band Linking, and Collision Detection

Value:

At $20, it offers a feature set that rivals plugins costing $199, making it the best budget all-in-one mixing tool (EQ + Saturation + Dynamics).

📰 Press & Coverage

studiorack plugin install garebear99/freeeq8

Worth being straight about

The two strongest signals here — the Audinux Fedora packaging and the Open Audio Stack registry entry — both came from third parties adopting the plugin on their own. Neither came from the submission campaign, which produced zero merges between 2026-04-23 and 2026-07-31.

Chinese-language coverage is the largest non-English share at four pickups (Midifan.com, AudioApp.cn, guaud.com, geekaifree/BookShelf). No Japanese or Korean coverage exists. FreeEQ8 is also not listed on KVR Audio, AlternativeTo, VST4Free, Plugins4Free or Audio Plugins For Free — those remain open opportunities rather than existing wins.

One further index exists at GareBear99 / awesome-audio-plugins-dev, but it is self-maintained and is deliberately excluded from the counts above.

Full evidence, per-item verification notes and the pending-submission tracker: docs/FEATURED_ON.md.

📄 Publications

🏫 DAFx26 Demo Track (MIT, September 2026) — Submitted May 29, 2026 (#109)

🎁 TizWildin Hub

Tracking our first 1,200 users — at 1,000 we'll run giveaways (ProEQ8 keys, sample packs, early access).

Full submission tracker: docs/FEATURED_ON.md · docs/SUBMISSIONS_2026.md

FreeEQ8 in Ableton Live

🆚 How FreeEQ8 Compares

FreeEQ8 is the only free EQ that combines linear phase + dynamic EQ + match EQ + per-band saturation + band linking in one plugin.

⚡ ProEQ8 — demo only in v2.3.1, not currently purchasable

Read this before trying ProEQ8 from the v2.3.1 download.

The ProEQ8 bundled in the v2.3.1 macOS DMG is a demo that cannot currently be unlocked, and it should not be used for real work. Specifically:

• There is no working purchase path. The configured checkout host does not resolve, so no license key can be obtained. Demo mode cannot be exited.
• Demo mode interrupts audio — 2 minutes of clean playback, then a 30-second mute window, repeating. Offline export is disabled entirely.
• Known shelf defect. ProEQ8's SVF shelves overshoot the set gain by up to 24 dB at high Q. A +12 dB shelf at Q 24 peaks near +36 dB. FreeEQ8's shelves were corrected in v2.3.1 and are bounded; ProEQ8's have not been yet.
• macOS only. ProEQ8 is built and validated on Windows and Linux but is not packaged there.

Use FreeEQ8 instead. It is unrestricted in real time, has no nag screens, and its shelf response is measured monotonic at every Q.

All four points are targeted for v2.3.2. Until then treat ProEQ8 in this release as a preview of the feature set, not a usable plugin.

ProEQ8 is the commercial counterpart to FreeEQ8 — same DSP engine, expanded feature set. The comparison below describes the intended product. It is not currently purchasable, and the demo bundled in v2.3.1 has the limitations listed above.

ProEQ8 is included in the macOS DMG download only. A license key is required to unlock it. As of v2.3.1 no key can be issued — see the notice at the top of this section. Without a license ProEQ8 runs in demo mode: 2 minutes of clean playback, then a 30-second mute window (repeats), with offline export disabled.

🧠 Smart EQ Layer (in development — not yet active)

The goal is a real-time decision layer on top of the existing 8-band engine — not to replace surgical control, but to reach a clean mix faster. This layer is not active in any current build. The pieces below exist as standalone, unit-testable components, but they are not yet connected to the audio path or the UI, so they have no effect on how the plugin sounds or behaves today.

Implemented as standalone components (not yet wired in):

• Source/DSP/ResonanceDetector.h — log-frequency peak finder producing up to 4 ranked suggestion bands with recommended frequency, cut-gain, Q, confidence score, and a semantic label ("mud", "boxiness", "harshness", "sibilance" …).
• Source/DSP/IntentMode.h — behavioural biasing curves: None / Vocal Clean / Drum Punch / Guitar Space / Master Polish, shifting detector scoring toward the frequency zones that matter for a given source.
• Source/DSP/FrequencyExplainer.h — static frequency → semantic description map intended to power an explain-on-hover UX.

Partially present:

• intent_mode APVTS parameter is registered and host-automatable, but nothing currently reads it. Automating it has no audible effect.

Still to build before this layer does anything:

• Instantiate the detector in PluginProcessor and feed it the existing SpectrumFIFO data
• Thread-safe suggestion hand-off from the analysis path to the editor
• Suggestion-node overlay on the response curve
• One-click apply into the next unused band via APVTS (undo-able)
• Explain-on-hover popup
• Read intent_mode and route it into the detector

See docs/SMART_EQ_LAYER.md for the algorithm design.

🔍 Code Audit

A full line-by-line audit of the DSP, threading, and restriction system is published at AUDIT.md. Summary: FreeEQ8 has zero feature locks and zero nag interruptions during real-time playback. Offline export is limited to 4 minutes 30 seconds (ProEQ8 removes this limit). All bug fixes through v2.2.5 are documented with root cause analysis.

📊 Benchmarks & RT-safety

FreeEQ8 / ProEQ8 v2.2.5 ships with a proven real-time-safe DSP engine:

• Zero heap allocation on the audio thread for any user action (pooled oversamplers).
• Canonical swap-chain triple-buffer for both the spectrum FIFO and the linear-phase FIR kernel — verified under concurrent stress (3 runs × 400 ms, ~600 M samples, 0 tears).
• Linear-phase FIR rebuild on a dedicated juce::Thread — no FFT work on the audio thread.
• MatchEQ chunking handles DAW blocks of any size (previously dropped on n > fftSize).

Full evidence and numbers:

• Milestone A report — math, benchmarks, stress-test results, sonic-impact analysis.
• v2.2.0 release readiness — what's verified green/yellow/red.
• Tests/AuditBench.cpp — reproducible micro-benchmarks (clang -O3 -DNDEBUG -pthread).
• Tests/AuditRegressionTest.cpp — concurrent-stress invariants for the triple-buffer + chunking patterns.

✨ Features

Core EQ

• 8 Independent Bands — full parametric control (frequency, Q, gain)
• 6 Filter Types per band: Bell, Low Shelf, High Shelf, High Pass, Low Pass, Bandpass
• Multiple Slopes — 12 / 24 / 48 dB/oct via cascaded biquad stages
• Per-Band Enable/Disable & Solo for A/B comparison and audition
• Parameter Smoothing (20ms linear interpolation, coefficients refreshed every 16 samples during parameter automation; every sample when Dynamic EQ is active)

Advanced Processing

• Linear Phase Mode — symmetric FIR from combined biquad magnitude, overlap-add FFT convolution (2048-sample latency when active)
• Dynamic EQ — per-band envelope follower with sidechain bandpass, threshold, ratio, attack & release
• Per-Band Saturation / Drive — gain-compensated tanh waveshaper (0–100%)
• Mid/Side Processing — M/S encode/decode with per-band channel routing (Both / L-Mid / R-Side)
• Oversampling — 1x / 2x / 4x / 8x using JUCE polyphase IIR half-band filters
• Band Linking — link groups A/B propagate frequency (ratio), gain & Q (delta) changes
• Match EQ — capture a reference spectrum, analyze current signal, compute & apply per-bin correction via FFT

Automatic compensation — two shipped layers

Both reduce manual correction, and both are active in v2.3.1. Neither analyses spectral content or makes recommendations — that is the separate Smart EQ layer, which is not yet wired in.

• Proportional Q (UI label: Adaptive Q) — Q scales with the absolute gain setting, Q_eff = Q * (1 + |gain_dB| * 0.12), clamped to [0.1, 24]. A Q of 1.0 at +12 dB becomes 2.44, so larger boosts and cuts automatically narrow. This matches the classic console behaviour usually called proportional Q. It responds to your gain knob, not to the audio.
• Auto Gain Compensation (UI label: Auto Gain) — measures input and output RMS per block, computes 20*log10(inRms/outRms) clamped to ±24 dB, and applies it through a one-pole smoother with a ~100 ms time constant. This one is signal-driven: it measures the actual audio, so bypassing stays level-neutral and A/B comparisons are not biased by loudness.

Visualization & UI

• Real-Time Spectrum Analyzer — 4096-point FFT, Hann window, pre/post EQ toggle
• Interactive Response Curve — composite + per-band colored curves with dB/frequency grid
• Draggable Band Nodes — click-drag for freq/gain, shift+drag for Q, right-click context menu
• Stereo Level Meter — peak hold + RMS display
• Selected-Band Paradigm — 8 colored band buttons, single set of controls rebound per selection
• Dark Theme — resizable UI (750×550 to 1400×900)

Global Controls

• Output Gain (-24 dB to +24 dB)
• Scale (0.1x to 2x) — scales all band gains simultaneously
• Preset System — save / load / delete, 30 factory presets
• Undo / Redo — integrated with JUCE UndoManager via APVTS
• State Save/Restore — all settings persist in your DAW project

DSP Specifications

• Stereo processing (or Mid/Side)
• Sample rates: 44.1 kHz to 192 kHz+
• Dual filter topology: RBJ TDF-II biquad (FreeEQ8) + Simper SVF via trapezoidal integration (ProEQ8) — modulation-stable Dynamic EQ
• Double-precision (64-bit) internal arithmetic, float I/O
• Zero latency in minimum-phase mode; linear phase adds 2048 samples
• 0.62% single-core CPU budget at 44.1 kHz (SVF 8-band stereo, 161× headroom)
• Low CPU usage (disable unused bands, lower oversampling to reduce load)

Compatibility

• macOS: 10.13 High Sierra and later (universal binary: Intel + Apple Silicon)
• Linux: Debian/Ubuntu 20.04+ (VST3 only; see build instructions for dependencies)
• Windows: 10 and later (64-bit)
• DAWs tested: Ableton Live 10+, Logic Pro, FL Studio, Bitwig, REAPER
• Formats: VST3, AU (macOS), Standalone (all platforms)

🚀 Quick Start

macOS

git clone --recursive https://github.com/GareBear99/FreeEQ8.git
cd FreeEQ8
./build_macos.sh

Linux (Debian/Ubuntu)

git clone --recursive https://github.com/GareBear99/FreeEQ8.git
cd FreeEQ8
./build_linux.sh          # installs deps via apt, then builds

Windows

git clone --recursive https://github.com/GareBear99/FreeEQ8.git
cd FreeEQ8
.\build_windows.ps1

Plugins will be automatically installed to your system plugin directories.

📋 Build Instructions

Prerequisites

macOS

• Xcode Command Line Tools: xcode-select --install
• CMake 3.15+: brew install cmake

Linux (Debian/Ubuntu)

• GCC 9+ or Clang 10+
• CMake 3.15+
• JUCE system dependencies (installed automatically by build_linux.sh): libasound2-dev libjack-jackd2-dev libfreetype6-dev libx11-dev libxcomposite-dev libxcursor-dev libxext-dev libxfixes-dev libxinerama-dev libxrandr-dev libxrender-dev libwebkit2gtk-4.0-dev

Windows

• Visual Studio 2019+ with C++ build tools
• CMake 3.15+

Detailed Build Steps

1. Clone with JUCE Submodule

git clone --recursive https://github.com/GareBear99/FreeEQ8.git
cd FreeEQ8

If you already cloned without --recursive:

git submodule update --init --recursive
cd JUCE && git checkout 7.0.12 && cd ..

2. Build

macOS:

chmod +x build_macos.sh
./build_macos.sh

Linux:

chmod +x build_linux.sh
./build_linux.sh

Windows:

.\build_windows.ps1

3. Plugin Installation

macOS (automatic):

• VST3: ~/Library/Audio/Plug-Ins/VST3/FreeEQ8.vst3
• AU: ~/Library/Audio/Plug-Ins/Components/FreeEQ8.component

Linux (manual):

• Copy build/FreeEQ8_artefacts/Release/VST3/FreeEQ8.vst3 to ~/.vst3/

Windows (manual):

• Copy build\FreeEQ8_artefacts\Release\VST3\FreeEQ8.vst3 to:
  • C:\Program Files\Common Files\VST3\

4. Rescan in Your DAW

• Ableton Live: Preferences → Plug-ins → Rescan
• Logic Pro: Automatic detection
• FL Studio: Options → Manage plugins → Find plugins

🎛️ Usage Guide

Parameter Ranges

Common EQ Techniques

Surgical EQ (Problem Frequency Removal)

Band: Bell filter
Q: 6-12 (narrow)
Gain: -6 to -12 dB

Musical EQ (Broad Tonal Shaping)

Band: Bell/Shelf filter
Q: 0.5-2 (wide)
Gain: ±3 to ±6 dB

High-Pass Filtering

Band: HighPass filter
Freq: 20-120 Hz (depends on source)
Q: 0.7 (standard)

Example Settings

Kick Drum:

• Band 1: Bell @ 60Hz, Q=1.5, +4dB (sub thump)
• Band 2: Bell @ 200Hz, Q=3, -3dB (cardboard removal)
• Band 3: Bell @ 3kHz, Q=2, +2dB (beater click)

Acoustic Guitar:

• Band 1: HighPass @ 80Hz (rumble removal)
• Band 2: Bell @ 200Hz, Q=1.5, -2dB (boominess)
• Band 3: Bell @ 3kHz, Q=1, +3dB (presence)
• Band 4: HighShelf @ 8kHz, +2dB (air)

Vocals:

• Band 1: HighPass @ 80Hz (rumble)
• Band 2: Bell @ 250Hz, Q=2, -3dB (muddiness)
• Band 3: Bell @ 1kHz, Q=1, +2dB (body)
• Band 4: Bell @ 5kHz, Q=2, +3dB (clarity)

🔧 Technical Details

Architecture

┌──────────────────────────────────────────────────────┐
│               FreeEQ8 Audio Processor                │
├──────────────────────────────────────────────────────┤
│  Input Buffer (Stereo)                               │
│          ↓                                           │
│  Spectrum FIFO (pre-EQ) ──→ UI spectrum display      │
│          ↓                                           │
│  ┌─── IF linear_phase ───┐  ┌── ELSE (min-phase) ──┐│
│  │ Build composite mag   │  │ Oversampling ↑ (opt.) ││
│  │ response from biquads │  │       ↓               ││
│  │       ↓               │  │ M/S Encode (optional) ││
│  │ FIR convolution       │  │       ↓               ││
│  │ (overlap-add FFT,     │  │ Per-band loop ×8:     ││
│  │  4096-tap, 8192 FFT,  │  │  ├ Dyn EQ envelope    ││
│  │  2048-sample latency) │  │  ├ Smooth + update    ││
│  │       ↓               │  │  │  coefficients      ││
│  │ Output Gain           │  │  ├ Cascaded biquads   ││
│  └───────────────────────┘  │  │  (1/2/4 stages)    ││
│                              │  └ Drive (tanh)       ││
│                              │       ↓               ││
│                              │ Output Gain & Scale   ││
│                              │       ↓               ││
│                              │ M/S Decode (optional) ││
│                              │       ↓               ││
│                              │ Oversampling ↓ (opt.) ││
│                              └───────────────────────┘│
│          ↓                                           │
│  Match EQ correction (FFT overlap-add, optional)     │
│          ↓                                           │
│  Spectrum FIFO (post-EQ) ──→ UI spectrum display     │
│          ↓                                           │
│  Output Metering (peak hold + RMS)                   │
│          ↓                                           │
│  Output Buffer (Stereo)                              │
└──────────────────────────────────────────────────────┘

DSP Implementation

• Filter Structure: Transposed Direct Form II biquad (Biquad.h) + Simper SVF (SvfBiquad.h) with tan() pre-warp
• Coefficient Calculation: RBJ Audio EQ Cookbook (FreeEQ8) / Cytomic SVF (ProEQ8)
• Smoothing: Linear interpolation over 20ms
• Update Rate: Coefficients refreshed every sample when Dynamic EQ is active; every 16 samples during parameter smoothing (amortised cost). Variable-cadence engine (v2.2.3) batches to 4-sample intervals during stable envelopes.
• Precision: Double-precision (64-bit) coefficients and internal state; float I/O
• Linear Phase: 4096-tap symmetric FIR, 8192-point FFT, overlap-add convolution
• Dynamic EQ: One-pole envelope follower with sidechain bandpass at band frequency
• Spectrum: 4096-point FFT, Hann window, lock-free SPSC FIFO
• SIMD Scaffold: SvfBandArray<8> with AVX2/SSE2/Neon dispatch (v2.2.4)

Project Structure

FreeEQ8/
├── Source/
│   ├── PluginProcessor.h/.cpp     # Main audio processor
│   ├── PluginEditor.h/.cpp        # UI editor & layout
│   ├── Config.h                   # Product config (FreeEQ8 vs ProEQ8)
│   ├── DSP/
│   │   ├── Biquad.h               # RBJ biquad filter (FreeEQ8 path)
│   │   ├── SvfBiquad.h            # Simper SVF filter (ProEQ8 path)
│   │   ├── SvfBandArray.h         # SIMD-vectorised SVF (AVX2/SSE2/Neon)
│   │   ├── EQBand.h               # EQ band with smoothing, drive & dynamic EQ
│   │   ├── SpectrumFIFO.h         # Lock-free SPSC triple-buffer FFT FIFO
│   │   ├── LinearPhaseEngine.h    # FIR-based linear-phase EQ engine
│   │   ├── NaturalPhaseEngine.h   # 256-tap natural-phase mode
│   │   ├── MatchEQ.h              # Reference capture & correction curve
│   │   ├── ResonanceDetector.h    # Allocation-free resonance peak finder
│   │   ├── IntentMode.h           # Vocal/Drum/Guitar/Master weighting
│   │   └── FrequencyExplainer.h   # Semantic frequency → label map
│   ├── UI/
│   │   ├── ResponseCurveComponent.h/.cpp  # EQ curve + spectrum + nodes
│   │   └── LevelMeter.h           # Stereo peak/RMS level meter
│   ├── Presets/
│   │   └── PresetManager.h/.cpp   # Preset save/load system
│   ├── UpdateChecker.h            # GitHub releases update checker
│   └── LicenseValidator.h         # License + demo + export limit
├── Tests/
│   ├── BiquadTest.cpp             # RBJ coefficient correctness
│   ├── SvfTest.cpp                # SVF correctness (11 assertions)
│   ├── AuditRegressionTest.cpp    # Triple-buffer + chunking stress tests
│   └── AuditBench.cpp             # RT-safety micro-benchmarks
├── server/
│   ├── stripe-webhook.js          # Cloudflare Worker for Stripe → license
│   ├── wrangler.toml              # Wrangler deployment config
│   └── package.json               # Server dependencies (wrangler)
├── PAPER.md                       # Technical paper (DAFx/AES submission)
├── BENCHMARK.md                   # Full benchmark results & methodology
├── AUDIT.md                       # Line-by-line code audit report
├── ROADMAP_2_5_PLUS.md            # Long-horizon roadmap through v2.3.0
├── STRIPE_SETUP.md                # ProEQ8 Stripe deployment guide
├── FeatureBench.cpp               # Standalone benchmark suite (no JUCE)
├── CMakeLists.txt                 # CMake config (FreeEQ8 + ProEQ8 targets)
├── docs/                          # Screenshots & assets
├── JUCE/                          # JUCE framework (submodule)
└── build/                         # Build output (ignored)

🛣️ Roadmap

v0.4.0

• Real-time spectrum analyzer
• Interactive frequency response curve display
• Draggable band nodes on curve
• Adaptive Q implementation
• Band solo/audition mode
• Preset management system

v0.5.0

• Multiple filter slopes (12/24/48 dB/oct) via cascaded biquads
• Mid/Side processing mode with M/S encode/decode
• Per-band channel routing (Both / L-Mid / R-Side)
• Oversampling options (1x, 2x, 4x, 8x)
• Output metering with peak hold and RMS
• Resizable UI (700×500 to 1400×900)

v1.0.0

• Linear phase mode (FIR convolution via overlap-add FFT)
• Dynamic EQ capabilities (per-band envelope follower with threshold/ratio/attack/release)
• Band linking (link groups A/B with delta-based freq/gain/Q propagation)
• Per-band saturation/drive (gain-compensated tanh waveshaper)
• Undo/Redo system (integrated with APVTS UndoManager)
• Match EQ functionality (capture reference spectrum, compute & apply correction)

v2.0.0

• Online license activation (2 devices per key) with Stripe checkout
• ProEQ8 commercial target (24 bands, 4 saturation modes, A/B, auto-gain)
• Cloudflare Worker license server + Resend email delivery
• Demo mode for unactivated ProEQ8 (2 min clean + 30 s mute cycle)

v2.1.0

• Standalone app included in all platform packages
• Hardened ProEQ8 license: device-bound activation, 7-day re-verify, 30-day offline grace
• Server /verify endpoint for periodic re-validation
• Obfuscated signing secret in binary
• Fixed JUCE 7.0.12 API compatibility

v2.2.0

• Real-time safety: zero heap allocation on the audio thread for any user action (Milestone A / A1)
• SpectrumFIFO + LinearPhaseEngine kernel on canonical swap-chain triple-buffer (A4 / A5)
• Linear-phase FIR rebuild moved to a dedicated juce::Thread worker (A5)
• MatchEQ::applyCorrection handles arbitrarily large DAW blocks instead of silent early-return (A3)
• Editor modal dialogs + license HTTP callbacks are lifetime-safe via juce::WeakReference (A2)
• Demo cadence: 2 minutes of clean playback + 30-second mute window
• getTailLengthSeconds reports the MatchEQ overlap-add tail for offline renders (A7)

v2.2.1–2.2.2

• Transistor saturation gain bug fixed (ProEQ8)
• Dynamic EQ zero-lag coefficient update when dynEnabled
• MatchEQ `correctionGain[]$ \text{precompute} (3 \times \text{throughput} \text{improvement})
• [\text{x}] \text{Simper} \text{SVF} \text{engine} ($SvfBiquad.h`) — modulation-stable SVF, trapezoidal integration

v2.2.3–2.2.4

• Variable-cadence Dynamic EQ (75% CPU savings on sustained signals)
• intent_mode APVTS parameter registered (host-automatable)
• Compact / mini-window mode (identical coordinate mapping)
• Oversampling crossfade buffer (128-sample, eliminates pop)
• SvfBandArray SIMD scaffold (AVX2/SSE2/Neon)
• NaturalPhaseEngine (256-tap, 128-sample latency)
• pluginval CI (strictness-level 10)

Corrected 2026-07-31. The following were previously listed here as complete but are not implemented. ResonanceDetector, IntentMode and FrequencyExplainer exist as standalone, unit-tested components, but nothing in PluginProcessor or PluginEditor references them, so the layer does not execute in the plugin. The intent_mode parameter is registered and host-automatable but is never read.

• intent_mode connected to an editor dropdown and routed into the detector
• ResonanceDetector instantiated and fed from SpectrumFIFO
• Thread-safe suggestion hand-off from analysis path to editor
• Suggestion overlay on the response curve
• One-click “apply suggestion” into next unused band
• Explain-on-hover popup on band + suggestion nodes
• Pre-ring warning overlay (DrumPunch + Linear Phase)

v2.2.5

• Transistor saturation invD*d no-op bug fixed in shipped code
• MatchEQ hot-path pow() eliminated — correctionGain[] precompute
• SvfTest.cpp correctness suite (11 assertions, all passing)
• FreeEQ8 export limit (4:30 offline render)
• Version consistency across all docs (no fake v2.4/v2.5/v3.0 refs)
• PAPER.md section numbering fixed (§6 Benchmarks, §8 Compact View, §9 Future Work)

v2.3.0 — Stable Release

• Version alignment (CMakeLists.txt + Config.h both 2.3.0)
• Get Pro button for FreeEQ8 (opens ProEQ8 checkout)
• Full codebase audit completed
• Smart EQ layer documentation expanded in PAPER.md
• CI/CD fixes (Linux webkit2gtk-4.1, Windows MSVC lambda capture, macOS retry)

v2.3.1 (Current Release) — Multi-platform + shelf correctness

• Shelf filter NaN fixed — RBJ slope parameter S constrained to its defined (0, 1] domain. Affected 11.3% of the parameter space from Q 3.80 upward in every release from v0.3.0 onward. Shelves now monotonic at every Q; Q <= 2 bit-identical to prior releases, so no factory preset changes
• Band-link propagation race fixed — atomic claim replaces a plain bool guard that failed under concurrent parameter changes (EDEADLK abort)
• Linux release job builds — ProEQ8 now links libcurl
• Windows release job builds — VST3_AUTO_MANIFEST disabled; was hanging six hours on juce_vst3_helper.exe. Restores Windows support last shipped in v1.0.0
• Linux pluginval runs under xvfb (headless editor tests)
• macOS AU validation — pluginval's auval test excluded; auval non-blocking
• macOS DMG packaging hardened against hdiutil "Resource busy"
• Two shelf regression tests in CI (34,774,500 parameter combinations)
• Five previously-unbuilt test sources wired into CMake and CI
• First release with macOS, Windows and Linux artifacts from one tag

v2.3.2 (Next) — make ProEQ8 usable

The ProEQ8 shipped in v2.3.1 is a demo that cannot be unlocked and has a known shelf defect. These four items are what make it a real product rather than a preview:

• Working checkout — the configured host (proeq8-checkout.tizwildin.workers.dev) does not resolve, and the Stripe products do not exist. Full verified assessment of the backend, auth and checkout state, including a live unauthenticated account-read issue that should be fixed first, is in docs/BACKEND_STATE.md
• Bounded SVF shelf response — ProEQ8 shelves overshoot the set gain by up to 24 dB at high Q (measured). Two bounded variants have been measured; neither applied. See docs/SHELF_RESPONSE.md
• ProEQ8 packaged for Windows and Linux — already built and validated on both, but discarded rather than bundled
• Rotate the license signing secret and move to asymmetric signing. The current HMAC secret is embedded in this public repository under XOR-0x5A obfuscation, so it must be treated as published and every key issued under it as forgeable

v2.4.0 (Planned)

• Smart EQ host integration (see corrected v2.2.3–2.2.4 list above)
• Zero-Lag auto-switch between linear-phase and minimum-phase modes
• Cross-instance ARC-Core spine

v2.4+ — linear phase accuracy (deliberately deferred)

The engine is frozen at 4096 taps for v2.3.2 and v2.4.0. The defect is disclosed rather than partly patched, because every candidate fix either changes latency for existing users or changes the linear-phase character while still leaving a multi-dB error. See Known Issues for the measurements.

• Selectable FIR quality tier — 16384 taps gives ~2.7 Hz resolution at ~186 ms latency instead of 46 ms, opt-in so the default never moves
• Per-band realisability check — flag bands narrower than the active kernel can represent, with the predicted error, instead of failing silently
• Ruled out: a gentler taper than the full-length Hann. Removing the window entirely recovers only 2.25 dB of an 8.73 dB error while doubling out-of-centre impulse energy, so kernel length is the real constraint

🤝 Contributing

Contributions are welcome! Here's how you can help:

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-feature
3. Commit your changes: git commit -m 'Add amazing feature'
4. Push to the branch: git push origin feature/amazing-feature
5. Open a Pull Request

Development Guidelines

• Follow existing code style
• Add comments for complex DSP algorithms
• Test on both macOS and Windows if possible
• Update documentation for new features

Areas for Contribution

• 🎨 UI/UX improvements
• 🔊 Additional filter types
• 🐛 Bug fixes and optimizations
• 📚 Documentation improvements
• 🧪 Unit tests

📝 Changelog

See CHANGELOG.md for the full version history with detailed per-file changes.

📄 License

This project is licensed under the GNU General Public License v3.0 - see the LICENSE file for details.

Note: JUCE has its own licensing requirements. For commercial use, you may need a JUCE license. See JUCE Licensing for details.

⚠️ Legal Notice

FreeEQ8 is an original implementation of a parametric EQ plugin. It is:

• NOT affiliated with, endorsed by, or derived from Ableton AG
• NOT a clone of Ableton's EQ Eight
• An independent, open-source project
• Built using public-domain DSP algorithms (RBJ Audio EQ Cookbook)

📊 Benchmarks

Every feature claim is backed by a reproducible standalone benchmark suite. All features pass at >10x CPU headroom at 44.1 kHz / 512-sample blocks.

Key numbers (g++ -O3, x86-64, median of 16 trials):

→ Full results and methodology: BENCHMARK.md

Build and run the benchmarks yourself (no JUCE required):

g++ -std=c++17 -O3 -DNDEBUG -pthread Tests/FeatureBench.cpp -o FeatureBench -I.
./FeatureBench

🐛 Known Issues

• Changing oversampling mid-playback may cause a brief click (IIR half-band filter state differs between orders; reset on switch)

• Linear phase mode adds 2048 samples of latency — 46.44 ms at 44.1 kHz, 42.67 ms at 48 kHz. This is reported to the DAW via setLatencySamples on toggle, on state restore, and in prepareToPlay, and getTailLengthSeconds() declares the FIR tail, so the host delay-compensates correctly and mixes and bounces stay aligned. It is nonetheless far past the ~10 ms where latency stops being noticeable, so linear phase is unsuitable for live monitoring or tracking. Use minimum phase while playing

• Match EQ capture is mono-summed; correction is applied per-channel

• Linear phase mode does not apply M/S, per-band drive, or dynamic EQ — those features require per-sample biquad state and are on the minimum-phase path only. The oversampling selector and the affected controls are greyed out in the UI when linear phase is active

• Linear phase cannot render narrow low-frequency bands accurately. The FIR is 4096 taps, which at 44.1 kHz gives roughly 10.8 Hz of frequency resolution. A bell at 100 Hz with Q = 16 is only about 6 Hz wide — below what the kernel can represent — so the achieved cut falls well short of the displayed curve.

• Raising the sample rate makes linear-phase accuracy worse, not better. firLength is a compile-time constant, so latency in samples is fixed while resolution is roughly fs / firLength. The two move in opposite directions, which is the opposite of what most users would assume. Measured for a −12.00 dB bell at 100 Hz, Q = 16:

  Sample rate	Latency	Resolution	Achieved	Error
  44.1 kHz	46.44 ms	10.77 Hz	−3.27 dB	+8.73 dB
  48 kHz	42.67 ms	11.72 Hz	−2.39 dB	+9.61 dB
  88.2 kHz	23.22 ms	21.53 Hz	−1.09 dB	+10.91 dB
  96 kHz	21.33 ms	23.44 Hz	−1.36 dB	+10.64 dB
  192 kHz	10.67 ms	46.88 Hz	−1.18 dB	+10.82 dB

  Moving a session from 44.1 kHz to 96 kHz roughly halves the latency and roughly doubles the error. For surgical low-frequency work use minimum phase at any sample rate.

  Measured, target vs achieved at the band centre (Bell, −12 dB, 44.1 kHz):

  Band	Target	Achieved	Error
  1 kHz, Q 1	−12.00 dB	−12.00 dB	0.00
  1 kHz, Q 8	−12.00 dB	−11.71 dB	+0.29
  5 kHz, Q 16	−12.00 dB	−11.94 dB	+0.06
  100 Hz, Q 8	−12.00 dB	−5.78 dB	+6.22
  100 Hz, Q 16	−12.00 dB	−3.27 dB	+8.73

  The full-length Hann window on the impulse response was investigated as a possible cause and ruled out. Measured error at 100 Hz / Q 16 by window choice:

  Window	Error	Impulse energy beyond ±256 samples
  Full Hann (shipped)	+8.73 dB	0.004%
  Tukey 25%	+7.40 dB	0.007%
  Tukey 10%	+6.85 dB	0.008%
  Tukey 5%	+6.71 dB	0.008%
  None at all	+6.48 dB	0.008%

  Removing the window entirely recovers only 2.25 dB of the 8.73 dB error, so the window is a minor contributor and kernel length is the actual constraint. The window is deliberately left as-is: changing it would alter linear-phase character for every user while still leaving a ~6.5 dB error.

  A genuine fix needs a longer FIR, which costs latency — 16384 taps would give about 2.7 Hz resolution at roughly 186 ms instead of 46 ms. That is a real trade-off rather than a free win, which is why it is a planned item and not a patch.

  Practical guidance: use linear phase for broad tonal shaping and mastering moves, where its phase behaviour is the point. For surgical low-frequency work, leave it off — the default minimum-phase path renders those bands accurately. This is a general property of FIR linear-phase EQ, not specific to FreeEQ8, but the numbers above are this implementation's.

Fixed in v2.2.1:

• Transistor saturation gain error (ProEQ8) — double-multiply bug caused ~d× level at high drive; fixed
• Dynamic EQ coefficient lag — up to 16-sample transient lag between envelope detection and filter response; now zero-lag when Dynamic EQ is enabled
• Linear phase + oversampling undocumented interaction — behaviour now explicitly documented in source

Report issues at: https://github.com/GareBear99/FreeEQ8/issues

💡 Tips & Tricks

Performance Optimization

• Disable unused bands to reduce CPU load
• Use wider Q values (lower numbers) for smoother processing
• Enable adaptive Q for automatic gain-dependent Q adjustment

Mixing Workflow

1. Start with subtractive EQ (cut problem frequencies)
2. Use narrow Q to identify resonances
3. Use wide Q for musical boosts
4. Check your EQ in mono to avoid phase issues

Sound Design

• Stack multiple bell filters at the same frequency with different Q values
• Automate the scale parameter for dramatic filter sweeps
• Use extreme Q values (>10) for creative resonances

🙏 Acknowledgments

• JUCE Framework - Cross-platform audio plugin framework
• Robert Bristow-Johnson - RBJ Audio EQ Cookbook
• Audio Plugin Development Community - For knowledge sharing
• Ableton - For inspiration (not affiliation)

💖 Support the Project

FreeEQ8 is free and open source. If it's useful to you, consider supporting development:

Other ways to help:

• ⭐ Star this repo — helps others find FreeEQ8
• 🐛 Report bugs — open an issue
• 🔀 Contribute — PRs are welcome
• 📣 Spread the word — tell a producer friend

📧 Contact

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Email: neovectr.inc@gmail.com

Other Projects by Me!

https://github.com/GareBear99/TizWildinEntertainmentHUB

https://github.com/GareBear99/awesome-audio-plugins-dev

https://github.com/GareBear99/PaintMask_Free-JUCE-Plugin

https://github.com/GareBear99/WURP_Toxic-Motion-Engine_JUCE

https://github.com/GareBear99/RiftWaveSuite_RiftSynth_WaveForm_Lite

https://github.com/GareBear99/AETHER_Choir-Atmosphere-Designer

https://github.com/GareBear99/WhisperGate_Free-JUCE-Plugin

https://github.com/GareBear99/Therum_JUCE-Plugin

https://github.com/GareBear99/MixMaid

https://github.com/GareBear99/BassMaid

https://github.com/GareBear99/SpaceMaid

https://github.com/GareBear99/GlueMaid

https://github.com/GareBear99/XyloCore

Built with ❤️ by Gary Doman (GareBear99/TizWildin)

"Great sound shouldn't cost anything"

TizWildin FREE sample packs

Related audio projects

• VF-PlexLab - VocalForge PersonaPlex Lab starter repo for a JUCE plugin + local backend + HTML tester around NVIDIA PersonaPlex.
• PAP-Forge-Audio - Procedural Autonomous Plugins runtime for generating, branching, validating, and restoring plugin projects from natural-language sound intent.

Tester Call

FreeEQ8 is looking for real-world DAW testers.

Please test the latest release in your DAW and report compatibility feedback here:

https://github.com/GareBear99/FreeEQ8/issues/new?template=tester-feedback.yml

Helpful feedback includes OS, DAW, plugin format, what worked, what broke, and steps to reproduce.

See the full tester call:

https://github.com/GareBear99/FreeEQ8/blob/main/docs/testing/TESTER_CALL.md