ArchestackBest of Open-Source AI: List on GitHub ↗ (opens in a new tab)

#6464th of 79 in Image and videoFor image, video and design, video and animation

echomimic

Audio-driven talking-head video from a portrait, with editable landmarks

Live demo ↗ (opens in a new tab)Website ↗ (opens in a new tab)Repository on GitHub ↗ (opens in a new tab)
Stars
4.3k
License
Apache-2.0
Last commit
Apr 2026
Language
Python

Missed the latest GitHub check; these facts may be older.

Overview

EchoMimic is the research code and checkpoints for an AAAI 2025 paper on animating a still portrait from audio, facial landmarks, or both. It is a diffusion pipeline built on Stable Diffusion components (sd-vae-ft-mse, sd-image-variations-diffusers) with Whisper tiny for audio features. Inference runs through Python scripts or a Gradio UI; weights are on Hugging Face and ModelScope.

Who it is for: Researchers and hobbyists animating portraits from speech on a CUDA GPU

Strengths

  • Supports audio-only, landmark-only and audio plus selected-landmark driving modes
  • Accelerated pipelines cut 240 frames from about 7 minutes to about 50 seconds on a V100
  • Hosted Hugging Face and ModelScope demos plus a local Gradio UI (webgui.py)
  • Tested on 16 GB V100, 24 GB RTX 4090D and 80 GB A100 cards

Weaknesses

  • Needs manual download of several checkpoints and a static ffmpeg build with FFMPEG_PATH set
  • Stated as academic research only; no Docker image or packaged installer
  • Portrait-only; semi-body animation is in the separate EchoMimicV2 repo
  • Better singing model and Chinese talking-head dataset are still listed as TBD

What it needs

  • GPU required
  • Needs CUDA >= 11.7, ffmpeg 4.4 static build, git-lfs, conda
  • Models: EchoMimic checkpoints (denoising_unet, reference_unet, motion_module, face_locator), sd-vae-ft-mse, sd-image-variations-diffusers, Whisper tiny
  • port 3000

Also in Image and video

See all 79
Also in Image and video
RankProjectScore
1frigateLocal NVR with realtime object detection for IP cameras37.4k stars+923 (+2.5%) in 30 daysMIT93 out of 100
2MoneyPrinterTurboGenerates short videos from a topic, with script, footage, voiceover and subtitles142.8k starsMIT88 out of 100
3toonflow-appInfinite-canvas tool for AI short drama and comic video production21.5k starsMIT87 out of 100
#4ComfyUINode-graph editor for running diffusion image, video, audio and 3D workflows146.5k starsGPL-3.086 out of 100
#5InvokeAIWeb UI for diffusion image generation with inpainting canvas and node workflows29.1k starsApache-2.086 out of 100