RealTalk: We Recreated Joe Rogan's Voice Using Artificial Intelligence
21 replies, posted
https://youtu.be/DWK_iYBl8cA
Alright AI, now recreate this more convincingly:
https://files.facepunch.com/forum/upload/527/6a5c3882-bbf5-4897-a7d5-9ce7a5650cec/chimps joe rogan.webm
Well that's terrifying
You'd have a hell of a time convincing drunk me this is fake just saying
Holy shit that’s too real. Also the fucking chimp hockey team, god damn.
alright we need an AI model of alex jones's voice now
you wouldn't need much to feed it, just use this:
https://youtu.be/GD9rU9u9szw
id honestly do it myself if i knew how to do it
id do it for robert culp as breen as well
To be honest nothing new here. Google/amazon/apple has reached this level somewhere around 2016. People just haven't noticed it because they have not used famous person as a base for their training set.
Now I want to know how Joe Rogan's voice sounds like when trying to speak a foreign language.
I would assume it would sound off, while the AI knows how to make it sound like Rogan, it probably doesn't know how to properly pronounce the words, so it would probably end up sounding like those text-to-speech programs made for other languages and speaking a language its not made for
I love how it immediately goes to the chimps
Considering that's what Joe does, pretty much the best part of the bit.
Tinfoil hat time: One day the film industry will use things like this and CGI so they never have to pay actors again significantly decreasing the cost of making movies and increasing a film's revenue.
Purchasing likenesses is actually quite expensive.
What about creating whole new ones? And a voice?
Likenesses are the actors, not the part, and MIT perfected 1:1 voice modulation including singing, over 15 years ago. The question is not if, but when.
But what if we never have to pay for the film either?
One day perhaps, but I mean even today there is still value and status in having an orchestral soundtrack performed by an actual orchestra, or going to see a live band instead of listening to the recording, or a realist drawing rather than a photograph. Arts like voice acting won't go away anytime soon and I imagine it'll take a long time until it even becomes viable to do most voice acting with AI generated voices.
The only reason they've managed to get this close to Rogan's voice is because there's hundreds of hours of high quality audio of his voice due to his podcast.
The thing that speech learning algorithms are still missing things like inflection and emotion, so there isn't much control apart from just letting them say the words.
This website is created by the guys who made this voice synthesis: Faux Rogan
I can still tell the voice clips apart(I was right for every single one), but that's mostly because I listen to the podcast a lot, but also because Faux Rogan only sounds like Joe Rogan. It doesn't speak like Joe Rogan does.
I don't think we're far off from that point though.
Actors will say "that's based off of me," like with GTAV.
Sorry, you need to Log In to post a reply to this thread.