Skip to content

MEMO

An open-weight model for audio-driven talking video

Visit official site

What MEMO is

The model is described as an open-weight system that generates audio-driven talking video with identity consistency and expressive faces.

What you can do with it

  • Generate a talking video from an image and audio
  • Keep identity steady over a long clip
  • Detect emotion in audio to refine expressions
  • Shift or offset emotion in the video

Application scenarios

  • Make a portrait speak with a film audio clip
  • Generate a singing video from a portrait
  • Combine a sculpture image with a speech track

Pros & cons

✓ What we like

  • Portrait, sculpture, digital art and animation styles are listed
  • Speech, singing and rap audio types are listed

! What to watch out for

  • An audio file and a reference image are both required
  • Lip sync is refined through model parameter tuning

FAQ

Which languages are listed?

English, Mandarin, Spanish, Japanese, Korean and Cantonese are listed.

What does the audio module do?

The material states it detects emotion in audio to refine facial expressions.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review