OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Image & Video AI

Talking Face Avatar

Talking Face Avatar is a Jupyter Notebook-based application for generating talking-head videos from portrait and audio inputs. Its README describes using Leonardo.ai for portrait image generation and ElevenLabs for text-to-speech, and provides SadTalker-based local setup and inference instructions.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Talking Face Avatar — OpenViST project profile in Image & Video AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you have or can generate a character portrait and an audio track, and want to explore a SadTalker-based workflow for producing a talking-head clip.

Who it is for

  • Creators prototyping animated character portraits.
  • Game-asset makers preparing talking character clips.
  • Developers exploring portrait-to-video workflows with generated speech.

Documented uses

  • Turn a portrait image and speech audio into a talking-head video.
  • Generate speech audio from text through the described ElevenLabs workflow.
  • Create a full-body video variant using the documented still and full-preprocessing options.
Before you choose

Requirements & limits

The project is published as a public GitHub repository under the MIT license.

  • The README documents a local Linux setup with Python 3.8, Conda, FFmpeg, listed CUDA 11.3 PyTorch packages, and downloaded model files.
  • Leonardo.ai and ElevenLabs are named as external image-generation and text-to-speech services.
  • Manual inference accepts audio and image inputs and saves MP4 output; the README also provides a distinct full-body still-mode example.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up the Linux environment and model files

Prepare the documented Python environment and download trained models before using the inference workflow.

Before you start

  • Linux
  • Anaconda, Python, and Git
  • CUDA 11.3-compatible PyTorch packages as listed in the README

Steps

  1. Clone the repository and change to the project directory.

  2. Create and activate a Conda environment with Python 3.8.

  3. Install the listed PyTorch, torchvision, and torchaudio packages, then install FFmpeg and the requirements file.

  4. Run the model-download script to place the trained models in the expected folders.

Expected outcome

A configured environment with the project requirements and downloaded model files.

Guide limitations

  • The README does not provide a complete local runtime specification beyond the listed setup instructions.
  • The optional TTS package is described as optional for the Gradio demo; the setup steps do not explain how to connect external API credentials.
Documentation 1 ↗
Source-based practical guide

Generate a talking-head video from image and audio files

Use the documented manual inference command to create an MP4 from a source image and a driving audio file.

Before you start

  • The configured project environment and downloaded model files
  • A driving audio file and a source image

Steps

  1. Provide the audio file with the --driven_audio argument.

  2. Provide the portrait image with the --source_image argument.

  3. Optionally add the documented GFPGAN enhancer argument and run inference.py.

Expected outcome

The README says results are saved as MP4 files under a timestamped results directory.

Guide limitations

  • The README documents a separate still-mode command for full-body output; that mode requires additional options.
  • The instructions do not specify API-key setup or a complete configuration procedure for Leonardo.ai or ElevenLabs.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
Yazdi9/Talking_Face_Avatar · Official homepage title · retained in its source language
How we review sourcesSuggest a correction