Neon Synth

A futuristic, neon-styled virtual analog synthesizer built with JUCE and C++17, by Razvan Julian Petrescu / AudioDSP.

Features

• 4 Waveforms — Sine, Square, Triangle, Sawtooth
• Low-pass Filter — adjustable cutoff (20 Hz – 20 kHz) with on/off bypass toggle
• 8-voice Polyphony — up to 8 simultaneous notes with voice stealing
• Release tail — smooth ~300 ms exponential fade on note-off
• Post-filter waveform display — live output visualiser in the UI
• Custom neon GUI — background image, cyan/magenta controls
• MIDI — works with any MIDI controller or keyboard
• MIDI Logging — optional real-time logging of received MIDI messages to a file (--midilog <file>)
• DAW automation — all parameters are automatable in VST3 hosts

---

Download

Pre-built Windows installer: Latest release →

Run NeonSynth-x.y.z-Windows-Setup.exe. The wizard installs the VST3 to C:\Program Files\Common Files\VST3\ and optionally the standalone app to C:\Program Files\AudioDSP\Neon Synth\.

---

Building

Prerequisites

Tool	Version
CMake	3.21+
JUCE	8.x (download separately)
C++ compiler	MSVC 2022 / Clang / GCC with C++17

Steps

cmake -S . -B build \
      -DJUCE_DIR="/path/to/JUCE" \
      -DCMAKE_BUILD_TYPE=Release

cmake --build build --config Release

On Windows (PowerShell):

cmake -S . -B build `
      -DJUCE_DIR="C:/path/to/JUCE" `
      -DCMAKE_BUILD_TYPE=Release

cmake --build build --config Release

Build outputs

Format	Path
Standalone	build/BasicSynth_artefacts/Release/Standalone/BasicSynth.exe
VST3	build/BasicSynth_artefacts/Release/VST3/BasicSynth.vst3/

---

Usage

Standalone

Run BasicSynth.exe. Click Options → Audio/MIDI Settings to select your audio output and MIDI input device, then play from any connected MIDI keyboard.

You can also launch the standalone application with the --midilog <file> parameter to log real-time MIDI input:

BasicSynth.exe --midilog log.txt

VST3

Copy the BasicSynth.vst3 bundle to your DAW's VST3 folder (typically C:\Program Files\Common Files\VST3\). The plugin appears under the AudioDSP manufacturer in your DAW's instrument browser.

Controls

Control	Colour	Description
Waveform	Cyan	Oscillator shape — Sine / Square / Triangle / Sawtooth
Filter ON	Magenta	Toggle button — enables / bypasses the filter (dims the cutoff slider when off)
Cutoff	Magenta	Low-pass filter cutoff frequency (20 Hz – 20 kHz)
Output	Cyan	Live post-filter waveform display

---

Python Testing

The included test_midi.py script loads the VST3 directly from Python — no MIDI ports, loopMIDI, or DAW required.

Install dependencies

pip install pedalboard mido numpy sounddevice soundfile

Run all tests

python test_midi.py

Expected output:

Loading plugin …
Parameters: ['waveform', 'cutoff', 'filter_enabled', 'attack', 'decay', 'sustain', 'release', 'bypass']

  PASS  Single note             RMS=0.2747
  PASS  Polyphony (4ch)         single=0.2747  chord=0.6365
  PASS  Polyphony (8ch)         RMS=0.8984
  PASS  Voice stealing          RMS=0.8943
  PASS  Release tail            tail RMS=0.183607
  PASS  Filter cutoff           HFR@1.5kHz  200Hz=0.0000  18kHz=0.0409
  PASS  Waveforms               Sine=0.0000  Square=0.0331  Triangle=0.0001  Sawtooth=0.0446
  PASS  Velocity scaling        vel20=0.1256  vel120=0.3351

8/8 passed

Render and hear audio

The script can also render a demo arpeggio and play it through your speakers or save it to a WAV file — no MIDI keyboard or DAW needed.

# play through speakers
python test_midi.py --play

# save to WAV
python test_midi.py --save output.wav

# both at once
python test_midi.py --play --save output.wav

The demo renders a C major arpeggio with a sawtooth oscillator and a 1.1 kHz filter cutoff — warm and analog rather than chiptune.

MIDI Logging

You can enable logging of all received MIDI messages (with high-resolution timestamps in seconds relative to the start of processing) to a text file by passing the --midilog <file> option:

# Log MIDI events to a file during testing
python test_midi.py --midilog log.txt

Note

On Windows, when a relative path like log.txt is passed, the file is created relative to the VST3 plugin binary's location within the build folder (e.g. build/BasicSynth_artefacts/Release/VST3/BasicSynth.vst3/Contents/x86_64-win/log.txt). To write to a specific directory (like the repository root), provide an absolute path:

python test_midi.py --midilog C:\path\to\your\synth\log.txt

Example log output:

[0.025s] Note on C3 Velocity 127 Channel 1
[0.028s] All notes off Channel 1
[0.029s] Note on C3 Velocity 120 Channel 1
[0.118s] Note off C3 Velocity 0 Channel 1

What each test checks

Test	What it verifies
Single note	Basic audio output — RMS above noise floor
Polyphony (4ch)	4-voice chord is significantly louder than a single note
Polyphony (8ch)	All 8 voices fire simultaneously without dropouts
Voice stealing	9th simultaneous note steals a voice without crashing or silence
Release tail	Audio persists ~300 ms after note-off (exponential decay)
Filter cutoff	High-frequency energy is lower at 200 Hz cutoff than at 18 kHz
Waveforms	Sine, Square, Triangle, Sawtooth produce distinct harmonic content
Velocity scaling	Higher velocity produces proportionally louder output

Using the API directly

You can also drive the synth from your own Python scripts:

import pedalboard, mido, numpy as np

VST3 = (
    r"build\BasicSynth_artefacts\Release\VST3"
    r"\BasicSynth.vst3\Contents\x86_64-win\BasicSynth.vst3"
)

plugin = pedalboard.load_plugin(VST3)

# Set parameters (raw_value is VST3-normalised 0–1)
plugin.parameters["cutoff"].raw_value  = (500 - 20) / (20000 - 20)   # ~500 Hz
plugin.parameters["waveform"].raw_value = 1 / 3                        # Square

# Build MIDI messages (time field = offset in seconds)
msgs = [
    mido.Message("note_on",  note=60, velocity=100, time=0.0),
    mido.Message("note_off", note=60, velocity=0,   time=1.0),
]

# Render 2 seconds of audio → numpy array (channels, samples)
audio = plugin.process(msgs, 2.0, SR, reset=True)
print(audio.shape, audio.max())

Play a chord

chord = [60, 64, 67, 72]   # C major
msgs  = []
for note in chord:
    msgs.append(mido.Message("note_on",  note=note, velocity=90, time=0.0))
for note in chord:
    msgs.append(mido.Message("note_off", note=note, velocity=0,  time=1.5))

audio = plugin.process(msgs, 2.5, SR, reset=True)

Sweep the filter

import numpy as np

# Low cutoff → dark tone
plugin.parameters["cutoff"].raw_value = (200 - 20) / (20000 - 20)
audio_dark = plugin.process(msgs, 2.5, SR, reset=True)

# High cutoff → bright tone
plugin.parameters["cutoff"].raw_value = (16000 - 20) / (20000 - 20)
audio_bright = plugin.process(msgs, 2.5, SR, reset=True)

---

Project Structure

File	Purpose
BasicSynth.h	Voice and sound classes — oscillator, filter, envelope
BasicSynthProcessor.h/cpp	AudioProcessor — parameter management, MIDI, state
BasicSynthEditor.h	Neon UI — waveform selector, filter slider, output display
CMakeLists.txt	Build config — Standalone + VST3, JUCE path, binary resources
gui_background.png	Neon grid background image (embedded as binary data)
icon.png	Application / plugin icon
test_midi.py	Automated Python test suite
pyproject.toml	Ruff lint configuration
.github/workflows/build.yml	CI — CMake build on Ubuntu
.github/workflows/lint.yml	CI — Python ruff lint
.github/workflows/release.yml	CI — builds installer and publishes GitHub Release on tag push
installer.nsi	NSIS installer script (VST3 + Standalone)
LICENSE	MIT licence