Machine Learning Engineers Use AI To Create Near-Perfect AI Simulation Of Joe Rogan's Voice

May 17, 2019

deep-learning-joe-rogan-voice.jpg

This is a video (audio with still images) of an artificial intelligence voice synthesizer created by machine learning engineers Hashiam Kadhim, Joe Palermo and Rayhane Mama using Joe Rogan's voice as a model. Everything you hear in the clip is the result of a deep-learning system they developed called RealTalk, which converts any text typed into the model's authentic sounding voice (it even does some tongue-twisters). Some pros and cons of the technology as imagined by Medium:

Pros:


- Talking to a voice assistant in a way that feels as natural as talking to a friend
- Customized voice applications -- for instance, a workout app that contains a personalized pre-workout pep talk from Arnold Schwarzenegger
- Improved accessibility options for people that communicate through text-to-speech devices, for example, people with Lou Gehrig's disease
-Automating voice dubbing for any media and in any language

Con artists:

- Spam callers impersonating your mother or spouse to obtain personal information
- Impersonating someone for the purposes of bullying or harassment
- Gaining entrance to high security clearance areas by impersonating a government official
- An 'audio deepfake' of a politician being used to manipulate election results or cause a social uprising

I didn't even think about all that, all I could think about is how awesome my cell phone voice mail greeting is about to get. "Even better than the current one that says you're probably in the bathroom and then you hear a toilet flush then a bunch of cursing and screaming?" Haha, my legs fell asleep and I crashed into the towel bar trying to stand up. But you're right -- equal but different.

Keep going for the video.

Thanks to Andy TU, who agrees the future nears, and the future is terrifying.

  • You could use a real voice as a type of voice pitch frame work and then synth the voice of your choice as a skin. Imagine a tribute music group who loves Nirvana making new material and then skinning Kurt Cobain's voice over the lyrics.

  • Javier Arreola

    Joe "wait until the chimps start using DeepFake and RealTalk while on DMT" Rogan

  • xvl260

    We need a Joe Rogan Vocaloid.

  • KungFuTreachery

    Just wait until the Russians start using this to post radical political shit on twitter and FB bc people are fucking idiots and will believe anything.

  • Bling Nye

    People can't distinguish between even amateur photoshop images and real, FFS, now we've got DeepFake and RealTalk... nobody will believe anything, and it'll drive them further into the arms of their "trusted" echo chamber information sources.

    Humanity is fuuuuuuuuuuuuuuuuucked.

  • Jon

    Quiet, bot.

  • GeneralDisorder

    I get the feeling that using the stuff learned with DeepFake and RealTalk we can use that to analyze audio, video, and photo for indicators of digital manipulation. Now if only we can figure out how to make sure that gets used with some kind of honesty and integrity. Welp, we're fucked alright.

  • Wilf Smith

    You're making a huge assumption that it's a reversible process, a two way thing. The art of the ultimate fake is that it's not - the end result is indistinguishable from what it's copying. Very achievable, particularly if you have that goal in mind to cover your tracks. Adding digital noise is easy. Ultimately you will not be able to analyze for digital manipulation - in the meantime whilst it goes through imperfect iterations - and no one yet cares about being analyzed - you might have some success.

  • PUNX

    is see this being used for more bad than good because people are dicks

  • Ivanna Pewpalott

    creates plausible deniability when frazzledrip finally leaks

  • Irina Abramovich

    1st!!!!

    I want GW’s voice to read books to me........

blog comments powered by Disqus
Previous Post
Next Post