Espectral started out as a personal tool. I needed a way to generate visual art for tracks on my own portfolio (briandukeslevert.com). I wanted something raw and expressive—heavily inspired by Basquiat's style—that built itself directly from audio inputs. But it’s so cool, I wanted to share it. Once the initial engine was working, the logical next step was making the artwork "dance" to the music. What’s currently built: 3 Engine Modes: Weave and Manual (running entirely on custom WebGL shader physics and math) plus AI Generated (for deep diffusion-backed art). 15+ Custom Renderers: Generating distinct visual styles driven by frequency analysis of uploaded tracks. Interactive & MIDI Controls: Twist parameters manually, map external MIDI controllers, or hit Auto Mode to let the engine drive. Export & Merch: Download WebGL video/audio recordings directly from the browser, save high-res prints, or turn generated pieces into physical posters and shirts. Try it with your own music: You can test your own tracks in the browser using the math/physics renderers for free. If you want to tap into the AI diffusion engine, pay-as-you-go packs and subscriptions are available. If you test out a track, I’d appreciate your feedback on the reactivity, the style options, and what features would make this a regular part of your creative workflow.

Espectral

Espectral — Real-time audio-reactive generative art.

Espectral — Real-time audio-reactive generative art.