Wav2Lip
Original open lip-sync model that matches any face video to new speech; non-commercial use only.
Sync mouths in any face video to new audio; personal and research use only.
Wav2Lip is the paper-code release of "A Lip Sync Expert Is All You Need" that matches the lips in a video of any identity to a new audio track. It suits researchers and hobbyists, and its terms exclude commercial use.
What it is
The README gives inference.py to run on a face video or image plus an audio file, pretrained checkpoints to download manually, and Python 3.6 with ffmpeg as requirements. A Colab notebook and a Docker image are linked.
Who it's for
Suits students, researchers and hobbyists experimenting with lip-sync. It does not suit anyone selling the output: the README restricts it to personal, research and non-commercial use and points commercial users to Sync Labs.
Key features
Lip-sync expert
Trained with a lip-sync discriminator for accurate mouth sync.
Any identity
Works on arbitrary faces in video or a still image.
Command-line inference
Run inference.py with a face file and an audio file.
Colab notebook
A notebook runs it without local setup.
Pretrained checkpoints
Manual downloads of sync-expert models.
Training scripts
Scripts to train the expert and the generator.
Up and running in 3 steps
Our honest verdict
What's great
- Long-standing reference implementation with many community forks and tutorials.
- Works from a single still image or any video.
- Colab notebook allows a no-install trial.
- Training code is included.
Where it falls short
- Licence is non-commercial only; commercial use must go through Sync Labs.
- Python 3.6 era dependencies make a modern install awkward.
- Checkpoints are downloaded by hand and the README warns the training data is research-only.
- Output face region is low resolution compared with newer models.
Pricing & license
Free for personal, research and non-commercial use. The README advertises a hosted commercial lip-sync API from Sync Labs, which it describes as higher quality; no price is shown on the README itself.
Open remote AI-training roles
870 open remote AI-training roles, updated November 14, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Can I use Wav2Lip commercially?
No. The README says the repository can only be used for personal, research and non-commercial purposes, and refers commercial requests to Sync Labs.
Is Wav2Lip still maintained?
Its latest commit on the default branch is from June 2025, and the README now leads with Sync Labs' commercial product.
Does it need a GPU?
The README gives a standard PyTorch setup; a GPU is the normal route for speed, but check the README options before relying on CPU.
