Please don't use the red bar in thumbnail as a clickbait

Bad actors will always find ways to misuse technology, and there are already decent voice cloning models publicly available like Eleven Labs. However, OpenAI should still release this Voice Engine model in a carefully controlled manner after implementing robust safeguards. The immense potential for restoring voices to patients and enabling accessibility makes responsible release extremely valuable, despite risks that must be mitigated.

@marcelcoetzee48

3 ай бұрын

🥱

Embarking on a journey of storytelling and video experimentation. VideoGPT gracefully entered the scene, effortlessly adding a layer of professionalism to my content.

Delving into storytelling and video experimentation lately. VideoGPT quietly became my secret weapon, enhancing the quality of my content with its professional touch.

If you use the Microsoft edge browser highlight text right click on the highlighted area and then select read aloud, I believe Microsoft has access to and is using the latest open AI voice model tech it's something different than everything else

They should release it. People already know you can't believe everything you hear.

Definitely need watermarks to help distinguish between Ai and real for security purposes.

@noone-ld7pt

3 ай бұрын

Take a look at pretty much any "AI watermarking" paper that came out the last 2 years, It's just not working. Anything we add can be removed, anything we alter can be altered back. And sure you can probably enforce it on the largest models from the mega-corps, but the open source models are just not that far behind on this and once everyone can run one locally the game is simply over. There is a reason why proxies of PirateBay are still up and running 21 years later despite going up against several of the most powerful industries on the planet. Once you give people the tech to do something amazing easily it's incredibly hard to then control it/limiy iy.

voices: string[] = ['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer']; But these voices are not available in the api, only these, I just tested.

I didn't get it. What's the point of this product if I cannot use it yet? Just to keep people interested about OpenAI products?

@budekins542

3 ай бұрын

Why should they release it only because you want them to?😂 They have to continue testing it out and work out how to deal with legal ramifications.

@helix8847

3 ай бұрын

@@budekins542 Which will take years.

This is the 'SORA' of voice generation😂

It seems AIs have an English accent as well when talking German hahaha :D This is not yet super-impressing, yet quite good.

The spanish sounds with a harsh english accent...

@edgarmedrano2562

3 ай бұрын

Perhaps they made that on purpose, it's not fair if the person is actually bilingual I know, so I guess in that case the person should use a different base recording in the target language to avoid that issue.

Yeah but the original voice sounds like an AI lol

It's worst than elevenlab lol😂

@JJ-vp3bd

3 ай бұрын

Is eleven lab good

@py_man

3 ай бұрын

@@JJ-vp3bd yeah , currently available best voice cloning software

@budekins542

3 ай бұрын

The first one is better.

@helix8847

3 ай бұрын

They trying to stay relevant. You know out of sight out of mind.