Mrkomiljon
commented
1 month ago @SWAPv1 you can use audio2video in real-time but there is a little bit latency. https://github.com/Mrkomiljon/Webcam_Live_Portrait please see my repo. I am actively updating and improving this repository.
I just want to thank you for open sourcing such a amazing and fun tool!
Most impressive thing is how fast the model runs and still outputs the best quality I have seen. What is the key to your models impressive performance? Also is this meant to be run in real time? and in this case what kind of hardware is required?
Regarding feature request, I think people would love to see a audio2video feature so we don't have to rely on existing video footage. And maybe a couple presets of genres like narrative or singing which would control expressiveness.
Anyways thank you for the tool and keep developing and sharing!