Company Lets Celebrities Digitally Clone Their Voice So They Can Loan It Out | Hacker News Reader

Company Lets Celebrities Digitally Clone Their Voice So They Can Loan It Out

pcmag.com

21 pointspulse715 comments

Imo the real reason this company exists is asserting that deepfakes have economic value and suing unauthorized creators of same. You're not going to get a deepfakes of Beyonce advertising a local car lot at any price. But you can bet she wants to protect her likeness (including her voice, and digital recreations of it) and will give a company the authority to go after anyone using it.

Curious now ... what if doppelgänger takes issue with, for example, Beyonce and counters that they are the original and she is infringing on their voice, digital recreations and likeness

Maybe it’ll be like trademarks or patents. First to file or something like that. So people like Nissan.com can win over someone like Nissan Motors.

What if impressionists file first?

I'm no expert in this subject matter. Best I can do for now is:

  https://en.wikipedia.org/wiki/Personality_rights

The nissan.com story for me is legendary and absolute win for the old internet.

Is it just me or does the voiceover in the video on this page seem low quality? It sounds like was recorded in a real low bit rate. I would expect much more from a company claiming to focus on high quality synthetic voice. I also hate those "AI", "Data lake" type videos, but the voice thing is an immediate distrust factor for me.

Yeah, I agree. It seems to clip between phonemes and it has weird diction because it doesn't get the stress patterns of the words right.

I entered "Hello, this is a test of the emergency broadcast system. This is only a test." and ran through the 5 voices they have highlighted in the demo about halfway down their page on https://www.veritone.com/applications/marvelai

It sounds, to me, like the voice synthesisers that come on the OS, and I can't tell any actual difference? All of them sound fake and generated and if I focus on it to think about why, the parts that always seem wrong to me are the space connecting the words, word to word there seems to be no connection in the inflection, such that there is no consistent flow in it. Maybe I don't have the right words to describe this.

Google's Wavenet Voices are pretty good.. https://cloud.google.com/text-to-speech/docs/wavenet

I can’t see the benefit of this outweighing the potential future catastrophes it will create, not to mention the future inherent distrust of audio we will develop.

We’re going to start missing real things.

Then you clearly don't have any background in video or audio production :)

This is going to be such a step change in media production, especially combined with video. As with any tech there will be negative use; but in 5 years I think a very high percentage of online video/audio will be synthetic because it's just so much easier. Just like every photo is edited to some degree and most videos use a green screen.

Moving in the direction of The Congress https://en.wikipedia.org/wiki/The_Congress_(2013_film)

Marketing tagline: Own your clone.

Reply on news.ycombinator.com