2 Burst results for "Voice Deepfakes"
"voice deepfakes" Discussed on ACG - The Best Gaming Podcast
"And I've tested those systems. They work quite well. There's a lot, but you can also do ones in audio. So what I see is that if you hire Jeremy to do audio in your game, it'll sound just like Jeremy, but underneath that, there will be a digital encoding in that audio that will say continually be an audio watermark that will be completely below any human's ability to hear. Like a hash key. Yeah, and it'll be like, this is an authorized use of my voice. And it'll just repeat all the time. No one will be able to hear it. There won't be any, you know, anything that's going on where somebody can see it, but you'll be able to quite easily and readily tell. And that's a great idea. And it needs to happen, because a person's voice is something. I get that deep fakes a lot of people think that's just like drawing on a peachy, your girlfriend in high school, but that's beyond, you know, now that it's so far, you do have to look at those and I think voice deepfakes are the next, you know, I saw some people who were posting some dead people in our Discord. They posted two dead people talking. And I was like, yeah, I'm not for that kind of weird. The person's dead, you don't get their voice. And this happened, by the way, with Charles Bronson after he died a company went back and put his face on some stuff and got in a bit of trouble, where his I think is a state said it was okay, but it's still the consumers were like, this is a voice from the dead. Dude, I'm just waiting for what the fuck is gonna happen with court systems. And just being able with lawyers defending their clients by saying, oh, this clearly AI, how they're gonna tell shit's getting so good right now. I don't know, man. I'm just waiting, man. I want to see what happens with that type of stuff. But you build have to be passed. It's going to be a dramatic shift. Speaking of a dramatic shift, let's shift over to unreal for a second because I do want to talk about unreal engine. So they showed a bunch of stuff. One of the things that I saw that they showed was metahumans. Did you guys see this one yet? Yeah. So that's their facial tracking. And I've got a video up for people to see this while we're talking about it. But basically, is that the right video? That might be the wrong video hang on. Let me make sure I'm doing humans. Don't put up the porn now. Scuffed podcast. Dude, I remember when Sasha Grey responded to me. I was like, this is the best day of my life. Yeah, so she was like, well, maybe not the best day. I was like, no, for sure, the best day. But metahumans are unreal's Unreal Engine's so I have a developer, I don't know if you guys do, but I've developer account within Nvidia and you have omniverse. Omniverse is like this suite of development tools that are ridiculous. And it what it's allowing is just everything to go into everything else. It's like cross play for blender. So you remember back in the day when it was like, oh, I've got Adobe they use their own template or whatever. And I can't do da Vinci and then somebody's like, well, I'll make a converter that converts so you can throw your project. You know, that kind of stuff. But what omniverse does is it takes audio or one of the parts of our universe and metahumans is it takes audio and it can really rig the lips to look right. But it's not only that man, it's that it's that they're
"voice deepfakes" Discussed on Firewalls Don't Stop Dragons Podcast
"I'll check that out myself and find some. I'll try to submit something. I think that's a great idea. All right next up this is this is really weird and creepy and we've talked about deepfakes on on the podcast before which are videos Using artificial intelligence machine learning to map one person's face on someone else's body There are other ways to do it. But that's that's the one that most people see there are other meanings to that word deepfake but basically it's fake video. It's it's taking real video footage and altering it and mostly notable ways to make it look like something that it wasn't well this one is now about voice deepfakes which i've talked about those before. They're back you know gosh. Two years ago maybe but there was a site that was doing fake trump and obama voices. That was mostly believable. That i said at the time. It's just gonna get better. Well it's gotten better. And i've got some recordings that will prove it But let's read this article. From the verge. It says. Recording advertisements and product endorsements can be lucrative work for celebrities and influencers. But it too much like hard work. that's what. Us firm vera. Tony betting today. The company is launching. A new platform called marvel dot. Ai that will let creators media figures and others generate deepfake clones of their voices to license as they wish. And this is a quote from ryan steele berg. Who's the president of our. Tony says quote. People want to do these deals. But they don't often have enough time to go into a studio and produce the content digital influencers athletes celebrities and actors. This is a huge asset. That's part of their brand quote with marvel. Ai he says anyone can create a realistic copy of their voice and deployed as they see fit while celebrity is sleeping. Their voice might be out about recording radio. Spots recording audiobooks and much more still burg says the platform will even be able to resurrect the voices of the dead using archive recordings to train. Ai models and he says again quote whoever has the copyright to those voices we will work with them to bring them to the marketplace. That will be up to the rights holder and what they feel is appropriate but hypothetically yes you could even have walter cronkite reading the nightly news again unquote speech. Synthesis has improved rapidly in recent years with machine learning techniques enabling the creation of ever more realistic voices. Just think about the difference between how apple's series sounded when it launched in two thousand eleven. And how it sounds now. Many big tech firms like amazon offer off the shelf text to speech models that generate voices at scale that are robotic but not unpleasant but new companies are also making boutique voice clones that sound like specific individuals and the results can be near indistinguishable from the real.