Listen
Voice or text enters the same channel independent API.
01 / Communication in motionBuilt for Zimbabwe
Let every
voice be seen.
One Python Marona agent, built to work across chat, events, and the channels still to come. Speak or type and see a live caption with an AI directed 3D avatar.
Open the studioLanguage lives in the movement of hands, the expression of a face, and the context between people.
Designed with community validation in mind02 / THE STUDIO
See the system take shape.
Speak or type a sentence. The same engine returns captions and avatar cues for chat or a live event.
AVATAR CANVAS LIVE RENDER
SIGN OUTPUT
Ready for your message
The agent directs the avatar in response to what you say.
AI directed visual accompaniment
CONVERSATION LABCONNECTING
About the avatar
The Marona agent directs these movements in real time. ZSL accuracy is being validated with the Deaf community; use the captions for the exact spoken words.
03 / THE ENGINE
One engine. Any stage.
The conversation interface is only the first channel. The Python API is designed to serve live events, mobile apps, and embedded screens.
Understand
A frontier model in the Python Marona agent interprets the message.
Move
The agent chooses motion cues that animate the avatar’s hands, body, and face.
Learn
Opt in ZSL recordings and Deaf review will make future translation measurable.