DSP FX Visualizer
DSP FX Visualizer is a WIP Python desktop GUI application for visually analyzing audio effects. Users build a signal processing chain by wiring nodes together in a visual graph editor, and the application renders the result across three visualization domains in real time: time-domain waveform, FFT frequency spectrum, and impulse response.
Overview
Wire up effects in a directed acyclic graph using a drag-and-drop
flowchart interface. Auto-discovery means adding a new FX subclass
to fx_lib/ automatically registers it in the node
library — no boilerplate required.
A 3×3 grid of plots showing input, output, and difference across waveform, spectrum, and impulse response domains simultaneously. X-axes are linked across rows for synchronized zoom and pan.
Each FX is a class with typed parameter fields — name, range,
step, default — that auto-populate the node's UI controls and
bind directly to run(). You write the DSP; the UI
wires itself.
Generators
This is a "what you hear is what you see" audio effects sandbox — primarily designed for educational use but also helpful for DSP prototyping, filter experimentation, and distortion analysis.
Six built-in signal generators let you feed any FX chain with precisely controlled input: sine, enveloped sine, multi-sine (additive synthesis with up to 8 harmonics), white noise, a single-sample impulse, and frequency sweeps — linear or exponential chirp.
Each generator exposes its parameters directly in the node UI — frequency, phase, duration, envelope shape — so you can dial in exactly the signal you need without touching code.
Effects library
The effect library currently supports distortion, delay, and filter
nodes — enough to build meaningful signal chains out of the
box. The tool is fully extendable with no overhead: drop a new FX
subclass into fx_lib/ and it appears in the node
library automatically. No registration code, no boilerplate.
Live code editing lets you click any node, view and edit its Python source, and hot-swap at runtime with a 300 ms debounce. No restarts, no rebuilds, no config files.
The architecture and core DSP nodes are human-designed; agent-assisted development was used to accelerate the UI, auto-discovery system, and visualization pipeline using PyQt6.
