Repo · Creative & generative UI

open-annie

Voice-driven 3D character in the browser: GPT-Live-1 talks with you over WebRTC while TypeSafe Jev reads the live transcript and picks every face,…

Open github.com ↗

From the repository

A voice-driven 3D character in the browser. The model that talks is not the model that moves: GPT-Live-1 talks, Jev moves her, three.js + VRM render her, a thin broker runs on fal.

avatarfallipsyncrealtimethree-vrmthreejsvoice-aivrm

How builders describe it

Voice-driven 3D character in the browser: GPT-Live-1 talks with you over WebRTC while TypeSafe Jev reads the live transcript and picks every face, gesture and dance (51 decisions at p50 118 ms and $0.0026 of Jev in the recorded session); lip-sync and rendering run in the browser behind a thin…

The decision Jev makes

Pick the next musical direction, visual option or UI component.

Where it fits

Using repeated small choices to steer something creative: music, pixel art, layouts and interfaces that assemble themselves from a fixed vocabulary. All 69 creative & generative ui projects →

Related projects