---
title: InfiniteTalk
emoji: 🎬
colorFrom: indigo
colorTo: blue
sdk: docker
pinned: false
license: mit
short_description: Drive a portrait from an audio track — lip sync and head...
---

# InfiniteTalk

Drive a portrait from an audio track — lip sync and head motion.

Runs [`wavespeed-ai/infinitetalk`](https://wavespeed.ai/models/wavespeed-ai/infinitetalk) through the
[WaveSpeed](https://wavespeed.ai) v3 API.

## Bring your own API key

This Space does not ship an API key. Paste your own in the field at the top —
create one in the [WaveSpeed dashboard](https://wavespeed.ai/dashboard).

Your key is sent only to `api.wavespeed.ai` to run the model. It is not stored,
not logged, and not written to disk. It is masked in the browser, stripped out
of error messages before they are displayed, and the generate event is marked
private so it never appears on Gradio's public API page — which would otherwise
publish client snippets containing every input, the key among them.
Generations are billed to your own account.

## Development

Source of truth is `_shared/` in the working repo, not this copy:

    python3 _shared/build_apps.py