Skip to content

EchoMimic

Animate a portrait from audio and facial landmarks

Visit official site

What EchoMimic is

The model is described as a portrait image animation model that generates realistic portrait videos from audio, selected facial landmarks, or both together. It uses a novel training strategy to address instability in audio-driven work and unnatural results from landmark-driven work.

What you can do with it

  • Generate realistic portrait videos from an audio signal
  • Animate a portrait from selected facial landmarks
  • Combine audio and facial landmarks for animation
  • View visual animation effects and access the source code

Application scenarios

  • Produce realistic facial expressions for a main character in video production
  • Add dynamic expressions and lip sync to game characters
  • Give a virtual teacher lively facial motions on an online education platform

Getting started

  • Visit the EchoMimic project page
  • Choose audio-driven or facial landmark-driven mode
  • Upload an audio file or mark facial landmarks
  • Select how audio and landmarks are combined

Pros & cons

✓ What we like

  • Audio and landmarks can drive animation separately or together
  • The training strategy targets instability and unnatural results
  • Source code is available to access and study

! What to watch out for

  • Animation needs an audio file or marked facial landmarks as input
  • The combined mode requires both audio and landmarks prepared

FAQ

What can drive the animation?

The material lists audio signals, selected facial landmarks, or a combination of both.

Which languages are listed for audio?

The feature list names multiple languages including Chinese and English.

What are the documented steps?

The steps cover visiting the project page, choosing a driving mode, uploading audio or marking landmarks, and selecting a combination.

How was it compared in the description?

The description reports comprehensive comparisons on public and self-collected datasets with quantitative and qualitative evaluation.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review