Voice agent presence

How it works

Give an assistant a face that breathes with its speech. Deep Current listens to the source, splits it into bass, mids, highs and transients, and drives the water surface from those bands every frame.

Setup

Agent audio → embedded canvas. Pick a scene, lock a palette, and it runs on its own. No timeline, no cues, no exports.

Good to know

Reduced-motion mode, custom palettes and session recording are available on every plan. Analysis stays on the device.