ComfyUI-NoiseGen

Noise synthesis and audio processing for ComfyUI. 18 nodes across generation, processing, and analysis.

Overview

ComfyUI-NoiseGen provides tools for experimental audio, harsh noise, and sound design within the ComfyUI environment. All nodes use standard ComfyUI audio format and integrate with existing extensions.

Node Categories:

• Generate (4 nodes): Noise sources and chaos systems
• Process (10 nodes): Filters, distortion, granular, effects
• Utility (5 nodes): Mixing, analysis, output

Installation

cd ComfyUI/custom_nodes
git clone https://github.com/eg0pr0xy/comfyui_noisegen.git
pip install -r comfyui_noisegen/requirements.txt

Node Reference

Generate

NoiseGenerator - Seven noise types with configurable spectral characteristics. White, pink, brown, blue, violet, perlin, and band-limited generation.

PerlinNoise - Organic noise textures using Perlin noise algorithms. Multi-octave generation with persistence control.

BandLimitedNoise - Frequency-constrained noise generation with precise high/low cutoffs.

TrueChaos - Mathematical chaos systems: Lorenz, Chua, Rössler, Henon, Duffing attractors with 4th-order Runge-Kutta integration.

Process

FeedbackProcessor - Delay-based feedback with filtering and modulation. Six modes: simple, filtered, saturated, modulated, complex, runaway.

HarshFilter - Extreme filtering with self-oscillation capabilities. Eight filter types including comb, morph, and chaos modes.

MultiDistortion - Twelve distortion algorithms with multi-stage processing. Tube, transistor, digital, bitcrush, waveshaper, foldback, ring mod, chaos.

SpectralProcessor - FFT-based frequency domain manipulation. Ten modes: enhance, suppress, shift, morph, gate, compress, chaos, phase, vocoder, freeze.

GranularProcessor - Granular synthesis engine with professional controls. Five grain envelopes, positioning modes, pitch manipulation, stereo spread.

GranularSequencer - Pattern-based granular control with step sequencing. Euclidean rhythms, probability gates, swing timing.

MicrosoundSculptor - Extreme granular manipulation for harsh noise and microsound art. Destruction modes with spectral processing.

ModulationMatrix - 8x8 parameter routing system. LFOs, envelope, audio follower, chaos, random, step sequencer sources.

ConvolutionReverb - Impulse response processing with eight reverb types. Hall, room, cathedral, plate, spring, chamber, ambient spaces.

Utility

AudioMixer - Four-channel mixer with gain and pan controls. Professional mixing with master gain.

ChaosNoiseMix - Experimental mixing algorithms. Nine mix modes including XOR, modulo, ring mod, chaos operations.

AudioSave - Enhanced audio output with waveform visualization and metadata. WAV and FLAC export.

AudioAnalyzer - Comprehensive audio analysis with RMS, peak, spectral centroid, frequency band analysis.

SpectrumAnalyzer - Real-time FFT spectrum analysis with magnitude, power, log, mel, bark display modes.

Usage

Basic harsh noise wall:

NoiseGenerator (white) → FeedbackProcessor (complex) → HarshFilter (comb) → MultiDistortion (destruction) → AudioSave

Granular processing:

[Audio Input] → GranularProcessor → MicrosoundSculptor → ConvolutionReverb → AudioSave

External audio processing:

VHS_LoadAudio → SpectralProcessor (chaos) → MultiDistortion → AudioAnalyzer → AudioSave

Technical Notes

• All nodes process audio at sample-by-sample level for real-time feedback
• Compatible with ComfyUI audio format: {'waveform': torch.tensor, 'sample_rate': int}
• Safety limiting prevents system damage from extreme parameter settings
• FFT processing uses configurable window sizes (512-8192) and overlap factors
• Chaos systems use numerical integration with adaptive step sizing

External Dependencies

Works with VideoHelperSuite (VHS_LoadAudio), ACE-Step sequencers, and other ComfyUI audio extensions. No format conversion required.

License

MIT