character3d

Character3D is a platform that gives your LLM AI endpoint a 3D face that speaks the response back to you!

By default, we have our Jane Austen bot. Ask her anything.

Swap out the API with your own endpoint URL.   The app sends the chat by posting a query parameter. And it expects a response that looks something like this:

{
    "query": " // nice to have (the original query the user sent)  ",
    "txtresponse": "// required ",
    "urlpolly: " // url to a MP3 file - optional, by default AWS Polly will be used to text to speech"
}

Only the txtresponse is required from your endpoint. The app takes care of the rest.

By default, if you omit an audio url endpoint, the app will use AWS Polly to do text to speech.

This is a half day hack, made from scratch, using Speech Graphics’s Unity SDK - which lets you turn any voice stream into emotional facial animations for your 3D characters.

Built in Unity, so it’s fast, at near-native speed for iOS, Android, Windows/Mac and WebGL.

Coming soon - OpenCV detecting the user's facial expressions and sending that to your endpoint as a mood parameter.

COME PLAY WITH THIS DEMO!