Compare
Browser 3D avatars vs video avatars
Jalpa draws a 3D avatar live in the visitor’s browser and lip-syncs it to the spoken answer, instead of streaming a generated video, so it does not need a video stream for every conversation.
Last updated 6 Nov 2026
How a browser 3D avatar works
The avatar model is loaded once, then the browser animates it and moves its lips in time with the voice. Only the answer text and voice audio travel over the network.
How video avatars work
A video avatar service generates video frames of a person speaking and streams them to the visitor, which needs more bandwidth and server time per conversation.
What to weigh up
- Look: video avatars can look like filmed people; a 3D avatar looks like a high-quality animated character.
- Speed: a browser 3D avatar starts speaking as soon as the voice is ready.
- Data: with Jalpa, everything runs in Azure Australia East.
See also Jalpa vs text chatbots and security and privacy.