V-Express
tencent-ailab/V-ExpressReviewed by AI
Checked by RepoPicks Radar on
What is V-Express?
V-Express is research code from Tencent AI Lab that animates a portrait. You give it a square photo of a face (512x512 or larger), an audio clip and, if you want, a reference video that supplies the head movement; it writes out a video where the face speaks the audio. Everything runs from the command line: install Python packages, download several AI models from Hugging Face, then run inference.py with the file paths. A graphics card is needed, and the authors say results are better with English audio.
V-Express is free and can replace HeyGen, D-ID, and Synthesia.
How easy is it to install V-Express?
D · terminal, several tools or a specific graphics card · Windows
- Windows D
- Linux D
On Linux and Windows the only documented path is the terminal: pip install -r requirements.txt, git lfs to download the model files, move folders by hand, then run inference.py with several arguments. There is no interface and no installer, and a graphics card is needed (the README reports 7956 MiB of video memory for a 31-second clip).
Coming soon: install in 1 click with the RepoPicks app.
V-Express on GitHub
- Stars
- 2,354on Nov 10, 2026
- Downloads
- —all releases
- Growth
- —no gain in the last reading
Stars over time
Facts
- Runs on
- Windows, Linux
- You need to install
- Python, Graphics card, FFmpeg
- To build from source
- Python, Graphics card, FFmpeg
- Disk space
- 1.0 GB–4.6 GB · estimate
- Accounts or keys
- None
- Cost
- Free
- License
- No license (No license)
- Last change on GitHub
- Jan 24, 2025
V-Express risk signals (5)
- AttentionDependencies with known vulnerabilities
Pinned dependencies with known vulnerabilities (OSV): diffusers: GHSA-7wx4-6vff-v64p, GHSA-98h9-4798-4q5v, PYSEC-2026-2446; torch: GHSA-3749-ghw9-m3mg, GHSA-53q9-r3pm-6pq6, GHSA-5pcm-hx3q-hm94; transformers: GHSA-29pf-2h5f-8g72, GHSA-37mw-44qp-f5jm, GHSA-4w7r-h757-3r74
automatic check
- Attentionmodel weights non commercial only
README: "both manual-downloading and auto-downloading models from V-Express are for non-commercial research purposes. Our released checkpoints are also for research purposes only." The code itself is said to be free for commercial use, but without a license file the terms are only this paragraph.
AI review
- Attentionmakes video of a real persons face
The tool turns one photo of a real face plus any audio into a talking video ("Scenario 2 (A's picture and any talking audio)"), which can be used to make someone appear to say things they never said. The README puts the responsibility on the user: "they are obligated to comply with local laws and utilize it responsibly. The developers will not assume any responsibility for potential misuse by users."
AI review
- InfoNo license detected
No license detected by GitHub, the license file, the LICENSES folder or the winget manifest
automatic check
- Infopinned old dependencies
requirements.txt pins torch==2.0.1, diffusers==0.29.2, transformers==4.41.1 and xformers==0.0.22, versions from 2023-2024, which may not install on recent Python or recent graphics drivers.
AI review
Evidence is shown in its original language.
What Radar could not check about V-Express
- How many gigabytes the Hugging Face model files really take after the download
- How long one video takes on a normal home graphics card (the README's only number is 2617 seconds for 31 seconds of audio on a data-centre V100)
- Whether the installation works on Windows, since every command in the README is written for a Linux shell
- How good the lip movement is with languages other than English, which the authors say has not been tested in detail