Hi everybody,
Using a picture of Ceara Lynch and the First order motion model method, I made this Deepfake preview. I was wondering what you people think about it?
youtu.be/RVrRtvy4Wxw
(This post was last modified: 02 Jan 2021, 15:04 by Like Ra.)
I like the spirals, the background music, and the pretty girl.
The voice needs additional work, though. It does not "fit" the girl 😊 What software/filters did you use to alter the voice? What can be done is to use TTS, and then speak with it to grab the mimics. Also, to look natural, the mimics must be exaggerated, just like for scenic lip syncing.
(This post was last modified: 19 Sep 2020, 23:40 by Like Ra.)
If you want to make it easier on yourself with the software pitch bending, try to actually modify your voice to sound more female to start with. It's something you can get better at with practice. I used to play video games with a bunch of guys that all thought I was a genetic female and I wasn't using --any-- software. Just look for trans female voice training videos on youtube. Doing at least part of the work yourself will make using software to correct the rest sound more natural.