I wanted to give my humble voice AI a face. The only way to do that was to use Tavus or Lemonslice and pay a large amount of money per minute of usage.

I wanted some cheaper option that I can run locally, attach to my existing voice agent and get a talking face, so I built this.

Just select an image and you can create a realtime talking avatar with lipsynced video generated on your own GPU.

While it is still in very early stage, but the results look very promising.