Early attempts at Speech and Expression sync.
Using Eleven labs API for text to speech—at least for testing— we generate viseme timestamps that we can then pass to code and then map in Rive.
I plan to add other means of bodily expressions, and obviously the UI, but for a first draft, looks promising.
I made this one as an experiment without opening a single design app. Basically, I just kept telling Codex what I wanted it to do.
It’s not a magic button. It took around 100 prompts and quite a few hours. But the fact that you can make something like this just by talking, without touching a keyboard or mouse, is still pretty wild.
And the whole thing runs and renders right in the browser.