For some reason the VibeVoice model from Microsoft (which is also able to clone voices and is also very good) has been deleted from GitHub 10 days ago even tough it was released under a MIT license. But this post shows that the cat is out of the bag for some time already now (post is from 2021) and we have to live with this technology.
there are many easy extant ways to do voice coding. many models are released without a “voice embedding” model but they are easy to recreate by passing the gradients through the soft prompt
Cloning the repo (running git clone on your computer) is enough because it makes a local copy. Forking merely makes a copy under your account on GitHub though which is not going to survive if they go on a deleting spree.
The way you wrote your comment/ the chosen example gives the vibe of "this is super dangerous"- and that it's actively happening, so everyone is pointing out this stuff is from a while ago.
For some reason the VibeVoice model from Microsoft (which is also able to clone voices and is also very good) has been deleted from GitHub 10 days ago even tough it was released under a MIT license. But this post shows that the cat is out of the bag for some time already now (post is from 2021) and we have to live with this technology.
Try Modelscope.
The reason is known: "we discovered instances where the tool was used in ways inconsistent with the stated intent"
The stated intend would be scamming people I guess? What would be the other ways inconsistent with that?
NSFW? That seems to be a bigger deal on the Internet today than scams, somehow.
Honest question - is NSFW just a code word for pornography now?
I had thought it would be anything that isn’t safe to open at work, including things with extreme profanity or gore, etc
To me, the word feel like it's almost a synonym for anime, pornographic or not, with a hint of negativity.
there are many easy extant ways to do voice coding. many models are released without a “voice embedding” model but they are easy to recreate by passing the gradients through the soft prompt
any links still up?
https://github.com/Enemyx-net/VibeVoice-ComfyUI
And that's why people need to clone these repos from big companies when their first released.
Cloning the repo isn't enough, because Microsoft/Github still control the platform, and can delete all copies they have control over.
Cloning the repo (running git clone on your computer) is enough because it makes a local copy. Forking merely makes a copy under your account on GitHub though which is not going to survive if they go on a deleting spree.
Yes, you are correct. I used the word "clone" when I should have used the word "fork" instead.
There's a community maintained fork.
https://github.com/vibevoice-community/VibeVoice
Does anyone know for sure if Voice ID has security measures to protect against AI voice cloning?
Atomic bomb level technological shifts happening, open sourced online. What a time to be alive!
(2021)
In fact, the YouTube video the GitHub repo links to is from 2019.
Papers are all from 2017-2018
Doesn't change the fact that it's really epic
The way you wrote your comment/ the chosen example gives the vibe of "this is super dangerous"- and that it's actively happening, so everyone is pointing out this stuff is from a while ago.
It looks rather complete, but indeed - the project has 3 commits for the past 3 years.