Gemini now has a face: Google’s AI speaks and changes expression in real time

Google has introduced Live Avatar, a novelty of Gemini 3.8 Live that pairs the assistant’s voice with an animated face, capable of responding in real time during the conversation. It’s a step that makes dialogue with AI much more like a video call, and which for now Google has decided to reserve for businesses.

Gemini 3.8 Live with Live Avatar: how it works

The avatar synchronizes lip movement with what it says and changes expression during the conversation, also managing speaking turns smoothly. According to Google, it can switch from one of the 97 languages supported without losing video quality and without image distortion: in one of the demos released by the company, the avatar speaks first in English and then in Japanese, with the mouth movement always aligned to the words.

While speaking, the avatar can also display information on screen. The tools run asynchronously: the system retrieves data in the background without interrupting the dialogue.

The novelty arrives about a week after the unveiling of the Gemini 3.8 Live model, which Google says processes visual inputs almost in real time.

For businesses only, including customized avatars

For now Live Avatar is available only in Gemini Enterprise, with access via the Google Cloud console and APIs. Among the uses cited by Google are customer support, interactive guided tours, and, as in an example shown, hotel check-in.

Companies can choose from a library of predefined avatars or create their own starting from reference images. The creation of customized avatars is however still limited to Enterprise customers included on a dedicated access list.

All content generated by Live Avatar includes SynthID, Google’s invisible watermark embedded directly in the audio and video, which helps make AI-generated content recognizable.

For all the latest news in the world of artificial intelligence, you can find our most recent articles in the AI section of GizChina.it.