SadTalker
Make a single photo talk from an audio file.
- Stars
- 14.1K
- Forks
- 2.7K
- Open issues
- 664
- Last commit
- 2024-06-26
Live from the GitHub API · updated 2026-10-15
What is SadTalker?
SadTalker generates talking-head video from one portrait and a speech recording, predicting head pose and expressions from the audio. It has a Gradio UI and a Stable Diffusion WebUI extension.
Key features
- Audio-driven talking heads from one image
- Still, crop and full-body modes
- Optional GFPGAN face enhancement
System requirements
- Platforms
- Windows · Linux · macOS
- GPU
- NVIDIA recommended
- Python
- 3.8
- Notes
- Requires ffmpeg.
How to install SadTalker
Summarized from the project’s README. Projects change quickly — check the GitHub page for the latest instructions.
Python
- 1Clone and install.
git clone https://github.com/OpenTalker/SadTalker.git cd SadTalker conda create -n sadtalker python=3.8 conda activate sadtalker pip install torch torchvision torchaudio conda install ffmpeg pip install -r requirements.txt - 2Download models and start the UI.
bash scripts/download_models.sh python app_sadtalker.py
Strengths
- Only needs one photo and audio
- Easy demo
Limitations
- Lip sync quality is dated
- Older dependency stack
SadTalker alternatives compared
| Software | Stars | License | Platforms | VRAM |
|---|---|---|---|---|
| SadTalker Make a single photo talk from an audio file. | 14.1K | Apache-2.0 | — | |
| LivePortrait Animate a portrait photo with the expressions of a driving video. | 19.2K | MIT | — | |
| Wan 2.2 Alibaba’s open-weight video generation models (text, image and speech to video). | 17.9K | Apache-2.0 | 24 GB for TI2V-5B at 720p; 80 GB-class GPUs (or offloading) for A14B models | |
| HunyuanVideo Tencent’s 13B-parameter open video foundation model. | 12.6K | Tencent Hunyuan Community License | 45 GB minimum, 60–80 GB recommended for 720p (per README); less with community quantizations | |
| CogVideoX Open text-to-video and image-to-video models from Zhipu AI / Tsinghua. | 13.1K | Apache-2.0 | ~12–24 GB depending on model and optimizations (see the README table) | |
| LTX-Video Fast, real-time-capable open video generation from Lightricks. | 11.1K | Apache-2.0 | 12 GB+ for smaller variants; more for 13B models |
FAQ
Is SadTalker free?
Yes — SadTalker is open-source software released under the Apache-2.0 license. Check the license (and any model weight licenses) before commercial use.
Does SadTalker run on Windows, Mac and Linux?
SadTalker supports: Windows, Linux, macOS.
What hardware do I need for SadTalker?
NVIDIA recommended
How do I install SadTalker?
1. Clone and install. 2. Download models and start the UI.
What are the best SadTalker alternatives?
Popular alternatives include LivePortrait, Wan 2.2, HunyuanVideo, CogVideoX. See the comparison table on this page.
Can I use SadTalker online without installing it?
If you don’t have a large GPU, AiLoft runs Veo 3.1 and Kling 2.5 in the cloud — generate and compare clips in the browser.
No 24 GB GPU? Generate AI video online
Run Veo 3.1 and Kling 2.5 in the cloud and compare clips side by side — pay per clip.