WAN 2.2 / SPEECH TO VIDEO

Give your image a voice.

One image. Your audio. A new performance — powered by the full open-source 14B model.

Full 14B model · 480p area · ZeroGPU 96 GB · Synthetic video

4 40
Clip length

Start with a clear image and a short WAV recording. The first generation may take longer. ZeroGPU xlarge uses 2× quota.

Try the upstream example
01 · Reference image 02 · Speech or singing 03 · Direct the performance

Model & license · Research · Generated media is synthetic. Only upload media you have permission to use.