NISPS Playground

Neural Interactive Shaping of Parameter Spaces

Map a 2D joystick to 20 visual parameters or 126 synthesizer parameters through a neural network
Two learning modes: show examples of what you want, or give thumbs up/down feedback
The network learns to interpolate — move the joystick and the entire parameter space responds
You don't need to understand the parameters — just explore and teach

Immersive

Full-screen canvas with floating controls. The visualization IS the interface — everything else gets out of the way.

Try it →

Dashboard layout with a 2D heatmap showing what the network has learned. Built for understanding the mapping.

Try it →

Three phases — Explore, Teach, Perform — that dissolve as you master the mapping. The UI fades away after training.

Try it →