This is not the final voice by any means! This is just testing a TTS engine and the viseme data generated from that. I created a quick script to translate the viseme mark data into morphs for the model.
There is a problem with the time marks for the viseme data getting translated into a keyframe for the animation. There's a rounding error I'm not accounting for. So I'll end up needing to apply a smoothing function so that things can't jump so dramatically between frames. The problem is VERY clear if there's no other motion in the scene. So I wanted to test how it looks when there is motion in the scene, including body and other facial movements.
Links:
https://driggsart.sfo3.digitaloceanspaces.com/private-animations/maddy_talking.mp4
https://driggsart.sfo3.cdn.digitaloceanspaces.com/private-animations/maddy_talking.mp4
SoftSafeVoreLover
2022-04-28 22:11:57 +0000 UTC