New AUDIO DEEPFAKES Give You any Celebrity Voice
Vložit
- čas přidán 2. 04. 2022
- Squarespace ► Head to squarespace.com/corridorcrew to save 10% off your first purchase!
Join OUR Exclusive Streaming Service ► www.corridordigital.com/
THIS EPISODE ►
Sam and Niko use an experimental voice-changing software called Altered.ai to create fake celebrity cameos for their friends!
Voiced with Altered.ai ► www.altered.ai/
SUPPORT ►
Join Our Website: bit.ly/Crew_Membership
Instagram: bit.ly/_Corridor_Instagram
Twitter: bit.ly/_Corridor_Twitter
Buy Merch: bit.ly/Corridor_Store
OUR GEAR, SOFTWARE & PARTNERS ►
Our Go-To Gear: bhpho.to/3r0wEnt
Puget Systems Computers: bit.ly/PC_Puget_Workstations
ActionVFX: bit.ly/TheBest_ActionVFX
Lighting by Aputure: bit.ly/CORRIDOR_LIGHTS
Cinema4D: bit.ly/Try_Cinema4D
Insydium: bit.ly/Insydium_Plugins
Octane Render by OTOY: bit.ly/Octane_Wrender
Boris FX - Mocha, Sapphire & Continuum: bit.ly/2Y0XLUX
Motion Captured with Xsens Suit: bit.ly/Xsens_MoCap_Suit
Reallusion: corridor.video/Reallusion_3Ds...
Unreal MegaGrant: bit.ly/Unreal_MegaGrant
MUSIC ►
Epidemic: bit.ly/Corridor_Music click this link for a free month! - Zábava
This would be an excellent tool for lone indy game developers to create a whole cast of characters for their games.
wow
good idea!
@BenBenson wow, really expensive stuff
That was also my first idea, and then mic trolls in games. Or, scambaiters.
I’m not sure how that would work with copyright and actually selling a product for profit, technically using someone else’s voice just with your inflections.
@@VeniVidiMortui That’s cool. I wasn’t aware of what the legality was so that’s cool that you have looked into it and are able to do that
These guys just nail it, They always do new things they find online while the channel is still going fresh even at almost 1.000 videos and no drama has ever happened with them, Everyone on the Crew deserves the support they'll always get with every video they make back then and now.
Hey, don't jinx it!
The app monthly price is ridiculous for the average users, speaking in my oscar-winning Will Smith voice. 😬
They also have all the toys that make these things simple
Hey thanks for the kind words. It’s the support of everyone here that let’s us create and experiment, and our appreciation for that runs real deep. -Niko
"no drama has ever happened with them"
Define "drama" because this video literally mentioned the AI Jake voice drama.
Also LTT RED cameras drama.
If this tech was more affordable I imagine it would be a valuable asset for Indie games, and rather than rely on a cast you’d likely only need 2-3 voice actors. Depending on whether or not the Voice deepfakes could pull off convincing emotion is another story, since from what we see here the Ai voices sound mostly neutral.
Gotta admit though The Morgan freeman segment broke me with starting as a guide to America and then requesting VFX help 😂
give it 5 more years and we are there.
@@Heinz76Harald potentially closer to 2 or 3 years at this rate
Yeah, I was already imagining how great this would be for modders and video game designers.
Why is it that these creepy technologies always get romanticized by telling people it will help small creators?
Wasn’t that what they told us when they removed the dislike button? To help small creators?
Don’t small creators adapt creatively to limitations and become more creative in the process instead of having all their whims on a silver plate?
and solo writers and Animators
This is perfect technology for Sam. Him making scripts up on the spot is always incredible
At some point, Corridor is going to introduce a 'new hire', have them participate in multiple videos, then reveal that the face and voice were deepfaked and it was someone else from the Crew the whole time.
They definitely should do this
That will officially be the day I'm terrified of deepfakes
Plot twist, it's Sam.
IT'S FENNER
Nah, not a deepfaked new hire... the person was entirely CG the whole time.
Honestly, there could be a huge audio book market for this. You could choose a narrator's voice based on the vibe of the book. You could also make every character literally have a different voice
actually Im trying to do this. Im writing books and thought "My voice sucks, no one will listen, why not choose a voice that actually works?"
audiobooks, indie animators, game makers, etc. As much of an ethical blackhole AI is, the plus side is it's going to give amazing capacity for small teams and individual creators who don't necessarily have the resources to hire talent for voice, music, etc to have "not bad" production values. Or even just for prototyping.
@@SolWake Every technology has good and bad aspects to it's use. Discovering how to split the atom creates relatively clean energy or radioisotopes for medical uses but it also lead to the development of some of the most destructive weapons ever made by man.
@@SolWake am pretty sure that breaks copyright law not sure tho maybe am wrong
@@marveyzing1495 ah, I think I wasn't clear. I mean licensing AI altered voices from a legitimate service as they did here (where the actors licensed their voices to the service), not pirating other people's voices without their consent, it would be like using stock assets.
These guys clearly have so much fun together that it's like a weekly shot of serotonin! Keep up the great work!
Here's an idea for a future video: Test when the voice breaks down. Weird non-speech sounds? Beatboxing? Slurred speech?
Does it still sound like the original or does it not sound like a human voice at all or does it produce weird artifacts?
I mostly want to see what happens if you try to speak another language through it
As a voice actor who is trying to make it in the VA industry, this is stunning and terrifying at the same time, but at least this still has to be driven by a performance. Someone still has to be hired to deliver the reading. Once we can fully dial in a performance through key strokes and sliders, then we'll be in trouble.
so in like 5 years.
@@vectoralphaAI a little generous when the tech is still in its infancy.
Every time someone says it'll take "X number of years" it's always twice that or more. I can't imagine voice acting unions would stand for it either.
@@TheRealAlpha2 Exactly
Two more papers? Down the road?
I want an entire animation voiced by ONLY ONE person with this AI.
I mean, isn't Star Wars the Clone Wars basically that, but without the AI voice?
Jokes aside, that would be interesting though
Maybe when it gets more voices and the voices gets less robot like
@@instinctisfiercenotcruel.958 Not a diss at all, I find it impressive the dude can make so many characters with the same voice, but not the same at the same time, each character having their own unique sound and such.
@@Dert26 I know you weren't sorry, i was just kidding man.
@@instinctisfiercenotcruel.958 Dee Bradley Baker*
It would be nice to see Rodrigo speaking in portuguese and see how the A.I. would react.
Muito feliz em ver um brasileiro no Corridor, parabéns Rodrigo!
BRASIL NO CORRIDOR EU TO MUITO FELIZ
sopa do macaco,delicia
@@Batnano for fuck's sake
I was hoping that they would have done that instead of making him "murican".
That’s pretty amazing! Some of the voices missed but the technology is their. This was another fun one!
If they tweaked their accents a bit more, they’d nail it 100%
all this is showing is that these actors cant do the celeb voices, not the ai.
That and the cadence. You can change someones voice all you like, but if you can't match their cadence it'll sound off.
@@dccrulez the cadence around the end was actually pretty good
@@DankMemer42013 true, but in the cases where they were lacking, cadence was the big breaker.
Cadence, word choice, pitch variation, and other affectation of delivery, as well as some of the more esoteric aspects of the target regional accent must be supplied by the operator/speaker.
"Let me dry your tears with your...money" had me dying. Sam is so funny and clever at coming up with this stuff.
Whomever you have ever been able to have in the world and the world of things that poor dog's life and your own heart and soul and the world of godess is not a good thing.
I died laughing when he finally cracked as Morgan Freeman, shit was just too hilarious
the snickers bar and the gun wheel chair bit killed me
I’d love to have Sam Elliott’s, James Earl Jones, or even Peter Cullen’s voices.
Yes. To all.
George Clooney
Duuuuuuude yeeeees, sam elliots voice is so damn good
You guys are amazing and always do everything I have always wanted to do. You make the things I have wanted for the last 10 years a reality every single episode. There is not a better channel I watch on youtube. Thank you for all the work you do.
Tell me there's Nick Cage.
We need Nick Cage, and his voice on his deepfakes.
I love you guys but the sponsored segments are awful now...used to be funny
Make it Niko Cage
Not the Will man, not the Will
@@dorkknight86 they're still funny. Everyone has different tastes
@@xg3990 they used to be creative and reflected their humor, now it's just skip material
Patrick Stewart saying "Hit me up" was a dead giveaway. Palpatine sounded more like Andy Serkis, which works. Much like deepfakes, it relies heavily on the performance to make it authentic.
yeah, it's perfectly adequate to emulate a podcast/audiobook
but i don't think you can have that much mileage trying to get an emotion out of it that's not softened downed too much
"Hit me up, Scotty." - Jean Luc Picard
"Live long and prosper"
*Chotto Matte*
Patrick Stewart saying EVERYTHING was a dead giveaway. Some of these voices were great, but some of them were not even close.
@@justaguy12345 Yeah, Patrick Stewart's sounded absolutely nothing like him. The Michael Cane wasn't bad, especially the reading at the end, but his voice is so stereotyped by now that practically any older guy with a cockney accent is going to read as him.
the voices are perfect, they just need an actor that can speak like them but might not have the same voice.
The ‘ta-ta’ sign-offs were the best!! Super-fun video!
Sam is actually so so so funny. I wish he had more opportunities to let his comedy shine in videos and more corridor content.
Watch old Node videos with him in it and the old node D&D series, he can be grumpy sometimes but when he’s in a good mood he’s hilarious.
@@rodom303 probably just doesn’t sleep enough
8-[==3
The audio quality is perfect, but none sound like the actors.
It is UNCANNY how close to perfect this technology is especially when compared to deepfakes. It just shows how much more visual information we take in compared to all of our other senses.
I think uncanny is the correct word, it might become this new phenomenon in using it for things like Animation and the uncanny valley for human faces.
Here's the thing about the gory newspaper and other things: a little more than half the world is being used as a special kind of way for people to be geese or moose.
This is super entertaining and I’m so glad I stumbled onto this … into this?
Whatever. It’s great!! 😂 Adore this entire crew.
I want to see an impressionist use this technology to turn their performance into absolute perfection.
I just want this for D&D, 10/10 would make the game feel way more immersive and make things way easier for the DM
I was thinking the same!
ME TOOO!!!
Imagine giving this tool to a GM like Rob Hartley...
It'd be hilarious to use this on young children over the phone. Imagine Elsa from Frozen scolding them for not eating their peas.
Come on boy, eat your frozen peas
This will ABSOLUTELY happen in the future
So much trauma
Oh shit that's a badass idea, and will dead-sure be used that way lol
😂😂😂 best idea I've ever heard
You guys always give me a great laugh but knowledgeable info at the same time
thanks for creating this episode. it's funny that the voice mail marks as 'SPAM and Dangerous'!!!!!
I'd be really freaked out if I got a Cameo call from Alan Rickman because he died six years ago.
The phone scammers will have a field day on this.
and the porn industry too
And shadow government....
@@tekknorat get your tinfoil hat off because no one asked
Well scammers want to make money, not spend $120/month for 30 minutes of morphing that's not even live.
@@M50A1 you stupid? Deep fake has *already* been used to push propaganda by one government or another. Don't be so naive.
This is sooo trippy... omg please do more of this
I love these boys.
They are lucky to have each other.
I love how upon reading “Revenge of the Sith”, I immediately knew what was going to be read out.
Morgan Freeman cracking up during his Cameo recording is so hilarious 😂😂🤣🤣
This episode was so wholesome, them dudes almost cried because of feeling valued and honored by other greats in the industry! I would have flipped it and let them hold on to that moment for a while. But then now they know that you guys care even more just to go and do that for them! Either way, great stuff!
Audio most important in any animation or movie ! Thanks guys for this great application 👍 !
As soon as that "Snape" impression started, I literally shouted at the screen "That's a horrible Alan Rickman but THAT SOUNDS LIKE JEREMY IRONS!". I was about to scroll down to the comments but they actually heard it too. Well met!
Now what about someone doing an impression of a celebrity's voice going through this deepfake? Would it sound any better than a person speaking in his or her regular voice? I mean, would it get any closer to the celebrity's voice just because the source is already a close match?
If you've seen Erik Singer react to accents, dialects, even actors playing presidents, you'll see just how much goes into matching a celebrity voice.
If John Malkovich uses the synthetic John Malkovich voice, he will enter an alternate universe. It's very dangerous
That's what i was wondering also
same thing happens with regular face deepfakes: The closer you already are to your target, the better the result
I've used a more basic software that can alter your voice. The better you match all the little nuances of a certain voice the closer you will get. So impression experts would absolutely fool us through this app for sure.
You've got to put this to the test with a karaoke contest.
This brings both the feelings of "best thing in the world" and "it ruins everything".
Niko's "Judy Dench" sounded so much like Robin Williams as Mrs. Doubtfire!
Majel Barrett-Roddenberry has recorded a library of phonemes before her death so that her voice can be used for every Computer in future Star Trek Productions
Sadly, I don't think they still use her voice (I imagine it's just cheaper to hire someone else) although I wish they still did for consistancy.
@@TheRealAlpha2 she didn't provide the voice for Star Trek Discovery. But I read about the phoneme library in several occasions. Otherwise I wouldn't have posted it.
11:02 "Hello Jordan, it's your friend, Winnie the PWGHU!"
I can see all sorts of shenanigans with this tech… like prank calling your boss or a coworker - props of you manage to replicate their boss’s voices.
You should find someone who actually does an impression of one of the celebrities and see how much closer they sound using the AI than a regular person.
Bring the impersonator from the Tom Cruise deepfake video back in order to complete the effect with the proper voice!
you should do a DND session where the GM is using this for their npc's!
That's genius
I run DnD sessions and thought the Same thing! SADLY two things 1, it isn't instant from what I can tell from the website. and 2 You have to pay $140 USD for 30 min.... With DnD sessions going 3.5 hours, I think I will stay away for now till the tech becomes little more widely affordable.....
@@puffinmaster8 I know VoiceMod works pretty well for instant conversions, but you would need a powerful-ish pc for it to run at a stable rate
that be awesome
@@puffinmaster8 all valid points. I was thinking more for son of a dungeon where there’s a lot is post processing and editing.
This is awesome and hilarious XD I completely lost it at Sam saying "Stor Wors" XDD
goddamn, that credits segment was cursed AF. Amazing technology though, both exciting and also worrisome how well these are already performing. Just as with the image Deep fakes, this is going to bring both amazing new creations and also great new issues of personality impersonation, rights to your "likeness", and possibly security issues. This is truly an exciting time to be alive!
I only think a couple of those voices actually sound like who you thought they sound like, but also there is a distinct digital quality to every piece of audio. Just hearing it, that digital aspect can't be hidden... at least not yet.
I agree though with more practice you could potentially minimise its weaknesses and concentrate on what it does well.
It seemed to sometimes struggle tracking if the person talking suddenly speaks at a higher octave.
agreed, I would say it sounds around 70% of the way there. I'm sure if you took your time with speech and mimicked the voice of the voice actor, you might get less artifact-y results.
If you take your time to actually mimic the person, and do some audio magic it could work, also, they are hearing on a phone speaker, that mostly makes it more 'natural'
All I’m taking out of this video is that Sam is an amazing voice actor
He's really more of an amazing improviser
Can't wait for something like this that works in real time for our DND sessions, as a Dungeon Master this would be world changing XD
Corridor really know how to do it they even make the sponsored segments interesting to watch 😅
Imagine using this software to record an audiobook.
really wanted to check it out but that price 🤕
Just checked out their pricing plans. It's fucking ridiculous. Their smallest plan is 120$ per month and that only gets you 30 minutes of audio max. Fuck these greedy mfs
How much?
@@mohammedismail3027 £120/month for the cheapest package which only gives you 30 minutes of morphed audio per month.
This really need an open source deepfake voice AI... I bet you could use actors voices from movies to build a data set.
@@Soul-Burn that seems expensive for such an insignificant amount of time lol
I was having a little trouble sleeping because of a fever but after that last bit i laughed so hard that i feel better thanks
DYING of laughter over here. Corridor for life.
The voices don't really sound much like their celebrity counterparts BUT as far as slipping into an AI voice, it's incredibly convincing!
They weren't actually the voices from the actors they tried to sound like, they found voices that sounded the closest to them in the AI voice library of other voice actors that contributed, still really cool.
scary
It won't be long before they perfect it.
Stewart and Caine have unusual inflections so they were quite well off with those
It comes down to speech mannerisms, Morgan freeman just sounded like a deep white southern voice but that's just because of how the guy spoke. A lot of them didn't sound close except for Jeremy Irons cause has similar speech mannerisms to the person who voiced him as well as movie reference. While using some voice AI that represents another English speaker.
20:43 IM DYING! its trying SO HARD! With all those voices.
This would be great for acting out a movie script made by an AI. Specifically a fan fiction movie script made by an AI. It would be halarious as hell
This stuff can bring old voices back to life.
The Corridor Crew is getting too smart now, you guys should do a video where you ACTUALLY give everyone a cameo bit so when they think it's a generated voice it's a real turn around.
This is how they voiced Luke in Book of Boba Fett and Mando.
Using this actual software?
@@JonahNelson7
Don't know about the software, but its the same technology.
i knew this was there but didnt know it was this good ,wow imagine having any voice u can use in a game as audio now that would be cool
This is incredible.
I like how the Brazilian guy still sounds Brazilian with the American accent 🤣
Seeing a fellow Brazilian being part of the crew fills my heart with joy.
It's not just the voice that you use it's understanding their manurisms, the way they articulate their voices, and how they express it. Then it could be hardly distinguishable.
This would be great for podcasts
"Give me Jason Statham!"
"Can't, but here's John Oliver taking about fish and chips in the loo."
Gott love how MOgon Freemun mispronounces his own name on his cameo. 😂
at this rate every single AI person regardless of the nature of their service should probably bar any IPs that can be traced back to a "Sam Gorski" lmao
Wait, it's all Sam Gorski?
@@catfish552 Always has been.
Todays episode is sponsored by ExpressVPN!
Mind-blowing 🤯
Also this would be cool for radio, many people get rejected by stations for not having a “radio” voice but if it can be simulated, any person with the talent and energy but lacking that one thing has a shot.
Absolutely true. Future will be about who has the most acting talent and best ideas, not about who has the voice or looks.
You literally have the ability to take a scene from a movie and completely reshoot it with different stories and make it R Rated
The way I could eventually see this used for is when the voice actors who read Audiobooks for a living dies or their voice goes out and retires.
The author can just read the book in their voice.
17:07 aw no he's back again and actually says something :(
I really want to bring this to my RPG tables
At this point, I would love to see them recreate scenes from movies. They could use virtual production for sets etc. Maybe have some foreground elements and props.
They could do the costuming a deepfake the faces. They would just need to voicefake their audio. It'd be interesting to see how close they could get to the real deal.
Bruh the crew really is living on the edge right now. How long till the whole gang ended up recruited by some shadowy alphabetic government agencies?
How do we know they haven't and these videos is just a cover?
they hire ppl with 1000000 times more talent
@@forge23 this video was actually all deep fake and AI voice lol
Oh man, could you imagine using this for D&D as a DM? What a game changer.
Love the content guys!!
I know you guys did a video about AI changes for video, but I have a request. May we see a video about how to spot deep fake video and audio from news media? Also, could you guys and Hak5 collaborate on this or a future video?
This has to be the funniest video yet...my humor and theirs doesn't always line up but HOLY SHIT...when they were just messing around at the beginning and riffing with different voices ...the random shit was so funny. "And this is where I keep my gun". Loved it!
I wonder how this technology will affect voice actors if unique voices can just be synthesized. All you need is someone with good annunciation and believability.
Just because you can copy a person's voice doesn't mean you can copy their performance and talent
I think that this tech will be used as it is now, for facial deepfakes. To bring back older characters by actors who are no longer able to vocie them anymore. WIth both a talented voice impersonator and a deep fake on top to enhance it.
As facial deepfakes won't replace actors, voice deep fakes won't replace voice actors.
In many ways, Siri already does this. If you notice, she can say the same thing and they often can sound different based on context.
@@aliciafraser1835 Yet
yep voice actors will use this and be even better
Wow. Totally didn’t double think about that truck video. I even thought “is is a Ford commercial?” And was even into it
WE WANT MORE OF THIS. MORE NOW, WE WANT MORE NOW
Hmmm... just so you guys know. This was the first time I was made aware of the fact that you guys have a podcast. Seriously, if not for those clips I would never know it.
Yeah they barely push it. It's not even in these descriptions
Yeah we could probably get the word out more 😬
@@CorridorCrew same with me.
Glad there's more content!
For the truetons only 😎
Maybe mentioned on their website? Not that I've ever been to it
perfect choice of voice on that outro
didn't expect id find u here lol
IT SOUNDS SO REAL HELP the emmes that could be made-
I don’t know if it’s because I do a lot of voice stuff, but there was a lot of tells in the audio itself which could be discounted as horrible bitrate or microphone distortion, but the biggest tell is the tone and cadence. Absolutely fun video though 🤣🤣
At first I was hyped by this because it would be so cool to voice my own characters for animations and stuff until I could hire some voice actors
however it costs at least $130 per month to use, so it's not worth it unfortunately
That is such a good price when you compare it to what a typical narrator charges. The narration of 2500 word script costs around 350 USD.
@@tonit4233 ah, makes sense. However as a nobody freelancer, the monthly fee is a bit pricey for me or other artists who are starting off
Except it’s only $130 for 30 minutes of audio which is a fucking scam
@@chairio6212 A good narrator would charge you over a 1000 bucks for 30 minutes of Audio.
@@tonit4233 If you're able to give the pitch and you're really indie with low budget, you could have someone narrate for lesser price
These techs will benefit more video games. Imagine just writing text and AI voicing the npcs, maybe even in real time. Or deep fakes to make easier TONS characters faces with realistic facial animations.
FromSotware wont have to be cheap on voices and facial animations.
We might one day realise a common dream: having an NPC pronouncing our names when talking to us 😅
@@NicolasLeperlier ikr! Shit would be nuts!
To be honest, the thought of that worries me because it could take away jobs for the voice and mo cap actors
YEEESSSS!!!! YEEEEESSSSSSS!!!
@@t.rexking441 Not every game has the budget/interest in doing that sort of thing. Definitely seems like good tools for indie developers.
This would be perfect for virtual reality characters it would really help you feel the part
This would be fun to play with and send messages to my kid to get to to do her chores from her fav people.. wish there was a free version
This is by far my favourite channel its just so interesting with a great group of people. The guests are amazing always providing an insight into the behind the scenes magic that happens with movies. Thanks for bringing great content.
You guys are doing fantastic work. Keep it up :)
The best part of this video is Sam's ad-libs - "This is where I keep my gun" lmao
To people who are interested to know more about what's happening, this is made with a technique known as voice cloning while the previous video was on speech synthesis. Both are based on recent advancement in deep neural networks
Nem acredito que temos um representante das terras tupiniquins no Corridor agora! Parabéns Rodrigo 🙌
Alguém tem que achar o insta desse cara
Acho q o nome dele é Rodrigo Gorski
É sempre lgl ver o gui nos reacts tb kk
Porra, foda demais! Me empolguei de verdade hahah
Que dahora!!!! Brazil dominando hahahaha
I feel like they really need to give the crew some legit gifts (while filming them) just to throw them off the trail from the next few things they'll do to them.
Tbh I'd love it as a thing just so see how suspicious they are xD
I really appreciate the sponsored segment timer…it makes it very convenient to know exactly when I can stop skipping…
Whoa the hot-mic recordings are gonna shoot to the moon, I tell ya!