And where youtubers can be cancelled because someone made an AI speak as them. And later people can be framed for crimes, or video and audio evidence will no longer be reliable in court, because an AI could have made it. All of the AI dangers that were predicted decades ago are coming true...
You can basically take any voice, with no background noise and turn it into an AI voice. DougDoug uses it regulary for his content, it's basically just a website where you can do it.
@@samszczepanski2281if you havent seen the full beginning of the pajama sam stream when hes setting up the voice there are some incredibly funny moments
You said you’re fine with it. I think the thing that people don’t like most about the ai voice issue is that even if they are okay with it or not, you don’t have a choice. They can get a fairly decent voiceprint of you from a 30 second voicemail. Good enough to make social engineering calls to fill your elderly parents into wire transferring money because they think you’re in jail. The tech is out there and it’s fairly easy to use and no one can do anything to stop it.
Yeah, some guy who I had a falling out with on a competitive server even tried to make a fake clip of me saying the N word with a similar AI, luckily I managed to fill in the admins (he was not too bright and literally gloated to me and I screenshotted the convo).
Honestly, stuff like this straight up terrifies me. Just imagine how many poor innocent people will be framed by AI generated voices in the future... Edit: glad to see I'm not alone. We'll pillage the wastelands together brothers!
Maybe in a few years, voice recordings of any sorts won't be trustworthy enough to be evidence for anything anymore, with how much this kind of technology is improving.
That future is now. That guy probably fed some of Phoenix voice to some free ai voice cloning tool that's on the internet. This decade and the next few are going to be FUN. Everyone will use AI, ai voice mimicry, self driving cars, quantum computers that can Crack the strongest passwords in the world under seconds, Cyber wars between countries, yadayada. FUN 🎉
I’m pretty sure the way the AI works is that it takes audio of just your voice and basically nothing else, then makes a synced transcript (the same way things like Siri do) to figure out what words correspond to what you’re saying, then from there it breaks up the words to find what different phonetic sounds sound like when you say them, and then it can do the whole thing in reverse to make words then sentences then read scripts that are fed to it. I have no idea honestly, I’ve never worked on anything like that but it’s my best guess on how to do it.
Very close, however the real process is a bit more tedious (from what I've seen). You have to split each word he says into a different audio clip and write what word is being said in each separate clip. Then you leave the AI overnight and in 8 hours you may have an usable voice synthesis yay.
Let's not forget there are old technologies like talkloid, vocaloid etc. That makes words out of existing voice samples. They're not flawless (either is one phoenix is in), but gets job done
AI has exploded in capability in the last 6 months. It is now possible to feed a minute of your voice into a voice analysis tool (powered by AI, naturally), give it a prompt, and in just a few minutes, you have an end result. None of the longer sampling process, and little wait time. For better or for worse, it is that simple. Future internet security rules may involve keeping recordings of your voice private, with how ridiculously easy it is to emulate it now.
Honestly, the AI voice kind of freaks me out a bit. The idea that AI can replicate someone’s voice so WELL is so much to imagine. Like- someone could make FAKE EVIDENCE of a crime with this or ANYTHING like that also buff phoenix is something I’m surprised I’ve never seen before 😭
Yep, AI can do this these days. You just need your voice, I think you don't even need subtitles, just the sound. It is pretty scary once you realise what a bad actor could do with this :(
Neuro-Sama is an AI streamer that actually sounds even more convincing than this AI. It’s also already said that it wants to take over the world and it hates humans. 💀💀💀
The worst thing is that there are already ai that can change the voice IN REAL TIME, preserving the intonation of the speaker. It essentially replaces your voice with someone else's.
The wikipedia thing is probably text to voice using some ai which is trained to mimic voices from(probably) 10 seconds of your voice. And the other 2 are probably voice to voice using more samples... they sound more like you so there is a chance that it's a custom model trained only on you... but i don't believe it is
Yo need way more than 10 seconds. For a decent audio quality you need from 5 to 10 minutes plus transcripts and you still get a lot of flicker and noise. It's easy to hear when it's a real person and when it's AI. For realistic voices it's around 20 minutes of clean audio, and the AI still cannot put any emotion without glitching out.
2:05 it’s a voice deepfake. From what I understand it works with two ai’s going back and forth with each other one making new content and the other fact checking that content based on what the user wants it to make. It then tells the first ai to fix those errors and the process repeats over and over again until the second one can’t find any errors or the person making it is happy with it. Because of the way it works it only really works well on people who there’s a lot of vocal data. Same tech is used for photo and video deepfakes.
It does kinda sound like him with the trash audio, but the wierd pause when it first mentioned Jeb, REALLY gave it away. How do ppl not be getting these things, man?
The scary part is, that with a decent enough computer (and/or enough patience) it's not actually that hard to make these covers and voice imitations anymore.
Ok, while I'm all up for technological progress this sort of thing makes me want to go back on my view of that. It's just uncanny. *Identity theft is not a joke, Jim! Millions of families suffer every year!*
1:07 - Everybody knows of Mojang's famous Israeli software engineer, "Jensjeb... Bergenstein." 🤣 Nice try, AI, but you're not skilled enough at human speech patterns (yet).
Im pretty sure there is an AI that needs only 1 minute of talking footage and it can recreate the voice perfectly. Im pretty sure that happened here Edit: im starting to think that isnt actually a thing but im not sure lol
This video got me so interested, I looked into real-time voice transformation, and it took me 5 minutes to sound like GawrGura. This tech fucking scary bro
people dismissing this tech are missing the point. right now yes its fairly easy for us to tell its not phoenix. but A. not everyone will be as familiar with his voice and could be more easily confused especially if they have hearing problems. b. This technology has progressed leaps and bounds in just a few short years, it wasnt long ago that to produce something of this quality you needed a professional studio and hundreds of hours of voice recordings to get a voice this good. now we have programs that do it from 30 second audio clips. how this tech will develop in a few more years is terrifying to think about...
Hello Phoenix, the softwere they are using is called "Project RVC" if you want you can look into it! it works kinda like a voice changer, so the pronunciation is based on the audio it is given, and yes it can be used in real time if someones computer is strong enough xdd
Through my phone speaker it was actually pretty hard to hear that its ai generated but as soon as i plug my headphones in it was pretty clear. It is really convincing already but you can still hear the difference clearly if you know what to look for. It could be confused with a low Bitrate recording though which is really crazy to me how good those ais have gotten
NO UR NOT USELESS U MADE ME HAPPY IN MY LIFE I DONT WANT U TO LEAVE (Apparantly while I was listening to ur classes u said that u would leave but PLS DONT)
Oh. My. God.
th-cam.com/video/4LKkL_xkhBs/w-d-xo.html
oh
58s ago?
HOLY SHIT
How????????????
How do people make this
Finally, Phoenix has the power to make himself say anything.
@padifythe world would be better if you stoped existing
@padifyoh my god another cringe slander video maker?!?!?
He can be controversial and blame it on ai!!!!
@@Writivite🗣🗣🗣💯💯💯🔥🔥🔥💪💪💪
@@CorruptedNoob4220000000000 IQ
3:27 “I'm just gonna get rid of the x”
Phoenis.
so true
Pheonics
OK so I'm gonna play the AI songs on stream instead of here cos I don't wanna be banned - sorry.
www.twitch.tv/phoenixsclive
Yes
Yes
Yes
Why my message no load 😒 😑
ai songs suck
"following several early testing versions" 0:58 sounds so spot on it freaked me out
We are entering an era where youtubers in apology videos can blame AI voices to defend themselves
How?
@@addison_v_ertisement1678they might make excuses saying I was framed and that the someone used ai to frame him or something
Phoenix: Man making that apology video sounds really exhausting. If only someone could just do it for me. Wait a minute….
Cool!
And where youtubers can be cancelled because someone made an AI speak as them.
And later people can be framed for crimes, or video and audio evidence will no longer be reliable in court, because an AI could have made it.
All of the AI dangers that were predicted decades ago are coming true...
Petition for Phoenix to rename his Channel to PeenixSC
Voted
Nope
real
Agreed
Yes
Therapist: buff Phoenix cant hurt you, he doesn’t exist
Buff Phoenix: 👁️👄👁️
Tick Tock, Heavy as a brinks truck.
Peenix*
@padifyidk man, your vid looks like kids bop trying to pull up some slangs but failed because the slang expires in August of 2020
@@totally_ahumanbeingIts a bot, dont reply to it
Wait, is OPs pfp actually is what i think it is!?
You can basically take any voice, with no background noise and turn it into an AI voice. DougDoug uses it regulary for his content, it's basically just a website where you can do it.
"I'm pajama Sam! AHAJHAGSJAKAJAHAKAJAJJAKAJAKAJAJAJJAKAJAJAHA"
High demon Elgrim is getting to me
@@samszczepanski2281if you havent seen the full beginning of the pajama sam stream when hes setting up the voice there are some incredibly funny moments
"The darkness is really getting to me"
uberduck
1:16 it sounds like phoenixSC if he had a bad mic
You said you’re fine with it. I think the thing that people don’t like most about the ai voice issue is that even if they are okay with it or not, you don’t have a choice. They can get a fairly decent voiceprint of you from a 30 second voicemail. Good enough to make social engineering calls to fill your elderly parents into wire transferring money because they think you’re in jail. The tech is out there and it’s fairly easy to use and no one can do anything to stop it.
Yeah, some guy who I had a falling out with on a competitive server even tried to make a fake clip of me saying the N word with a similar AI, luckily I managed to fill in the admins (he was not too bright and literally gloated to me and I screenshotted the convo).
Honestly, stuff like this straight up terrifies me. Just imagine how many poor innocent people will be framed by AI generated voices in the future...
Edit: glad to see I'm not alone. We'll pillage the wastelands together brothers!
you need therapy
@@arosbastion7052 what
Maybe in a few years, voice recordings of any sorts won't be trustworthy enough to be evidence for anything anymore, with how much this kind of technology is improving.
@@arosbastion7052 that's legitimately starting to happen. Scammers are using ai voices to trick people into paying them money
That future is now.
That guy probably fed some of Phoenix voice to some free ai voice cloning tool that's on the internet.
This decade and the next few are going to be FUN. Everyone will use AI, ai voice mimicry, self driving cars, quantum computers that can Crack the strongest passwords in the world under seconds,
Cyber wars between countries, yadayada. FUN 🎉
Hearing Phoenix SC's voice sing in Japanese really got me with that cover.
I remember when people were making TH-camrs sing things by using voice clips, but now we have full on ai the can almost perfectly mimic their voice!
2:24 the song is Renai Circulation, it's one of the openings for the Bakemonogatari anime, and had some memes made about it for a quite while.
The x is clearly designed for me
Alright we have PeenixSC's voice, now we can have him sing Ballin'.
i gotta new fawgees onna gee
Let's make him sing anime songs
Make him read the entire bee movie script.
Even better @@bowxfire5275, turn him into Barry Benson.
@@Twiddle_thingscringe
Renaming your channel to your real name (PeenixSC) then?
💀
☠️
💀
lmfao💀
TH-cam won't let me.
I’m pretty sure the way the AI works is that it takes audio of just your voice and basically nothing else, then makes a synced transcript (the same way things like Siri do) to figure out what words correspond to what you’re saying, then from there it breaks up the words to find what different phonetic sounds sound like when you say them, and then it can do the whole thing in reverse to make words then sentences then read scripts that are fed to it. I have no idea honestly, I’ve never worked on anything like that but it’s my best guess on how to do it.
Very close, however the real process is a bit more tedious (from what I've seen).
You have to split each word he says into a different audio clip and write what word is being said in each separate clip. Then you leave the AI overnight and in 8 hours you may have an usable voice synthesis yay.
Let's not forget there are old technologies like talkloid, vocaloid etc. That makes words out of existing voice samples. They're not flawless (either is one phoenix is in), but gets job done
AI has exploded in capability in the last 6 months. It is now possible to feed a minute of your voice into a voice analysis tool (powered by AI, naturally), give it a prompt, and in just a few minutes, you have an end result. None of the longer sampling process, and little wait time.
For better or for worse, it is that simple. Future internet security rules may involve keeping recordings of your voice private, with how ridiculously easy it is to emulate it now.
Alternatively, vocaloid.
@@kuba4ful I assumed it would be more work than what I’d described lol, thanks for letting me know
Kinda want a mod for Phoenix's voice on text to speech apps for reading stuff online or audiobooks or something
We all know he has a calming voice
Everyone, let’s make sure AI doesn’t develop in such a way
Too late
humanity made a huge mistake when we invented the transistor
@@solinus7131 Without the transistor, artificial anime girls won't be possible tho, and that's the only bright thing that awaits us in the future
@@suspicioussandwhat about the sun exploding? Its that not bright enough for ya?
Nah.
Renai Circulation sung by Phoenix brings a tear to my eye, it's beautiful.
if heaven was a song i think this is the closest we can get to hearing it
Padify thinks he makes BETTER VIDEOS than PhoenixSC!
(he doesn't)
The AI sounds like it has a cold
Handsome phoenix doesn't exist handsome phoenix can't hurt you
Handsome phoenix: 0:49
No AI voice can replace Phoenix SC!❤❤
Honestly, the AI voice kind of freaks me out a bit. The idea that AI can replicate someone’s voice so WELL is so much to imagine. Like- someone could make FAKE EVIDENCE of a crime with this or ANYTHING like that
also buff phoenix is something I’m surprised I’ve never seen before 😭
it is still not perfect it has weird artifacts
Petition for Peeenix to become an anime girl again
It’s OBVIOUSLY PeenixSC, how are you watching his video and doesn’t know his name.🙄
@@ICanSmellYourFear69 I'm so sorry, I editied it to fix my horrible mistake.
@padify no you don't
Padify thinks he makes BETTER VIDEOS than PhoenixSC!
(he doesn't)
Yep, AI can do this these days. You just need your voice, I think you don't even need subtitles, just the sound. It is pretty scary once you realise what a bad actor could do with this :(
Bro had us in the first half
As long as that X doesn't keep me up all night, LET'S GO
Ok, that's way too good. I've never heard such a realistic sounding AI
Most of them have this robotic-y hint to them.
Peenix already sounds robotic-y
you don't need to add -y to robotic, it's already an adjective🤓
nah u can hear some hints
Neuro-Sama is an AI streamer that actually sounds even more convincing than this AI. It’s also already said that it wants to take over the world and it hates humans. 💀💀💀
That's crazy 🐢
Ah yes buffed peenix
le buff
Thats hot
2:36 TheMisterEpic, is that you?
0:51 deveLŒp
1:08 THE "jeb" SOUNDS LIKE THAT AI VOICE THAT PEOPLE USE
Woah, there are two of Phoenix Arizona on one channel.
3:38 Correct! :D
The peenix is no more😔
2:41 Oh hey, it's me!
I need Phoenix SC Ballin' AI Cover
ALL THESE VIDEOS HAVE ME WHEEZING I FEEL LIKE MY HEARTS GONNA STOP, I need to get my hands on one too HAHAHAH
The worst thing is that there are already ai that can change the voice IN REAL TIME, preserving the intonation of the speaker. It essentially replaces your voice with someone else's.
I would absolutely have an issue with it. AI replication of anything is scary.
Phoenix isn't pregnant, but he always delivers 😃
@TheTrueTroller
...dad?
cursed
@TheTrueTroller… j?
@@TheWagonroast Rick?
He is pregnant on monster school videos
And I only subbed now. Which is so stupid since I've been watching your EVERY NEW VID SINCE FOREVER.
Phonics XP really dropping another good vid like that 🔥
I'm very impressed by how well peenix sounds as a vocaloid. What a bop.
The wikipedia thing is probably text to voice using some ai which is trained to mimic voices from(probably) 10 seconds of your voice. And the other 2 are probably voice to voice using more samples... they sound more like you so there is a chance that it's a custom model trained only on you... but i don't believe it is
Yo need way more than 10 seconds. For a decent audio quality you need from 5 to 10 minutes plus transcripts and you still get a lot of flicker and noise. It's easy to hear when it's a real person and when it's AI. For realistic voices it's around 20 minutes of clean audio, and the AI still cannot put any emotion without glitching out.
@@Ivi-Tora i know a lot about ai, believe me...
2:10 yes, yes they were
LMAO PHOIENIX RENAI CIRCULATION COVER??? THATS AMAZING
You had the opportunity to make this video with the AI voice and reveal it at the end
The ai voice thing is both awesome and terrifying. It could be used to make some really cool stuff but also be used for terrible things...
We need to put a stop to it before it's too late
2:05 it’s a voice deepfake. From what I understand it works with two ai’s going back and forth with each other one making new content and the other fact checking that content based on what the user wants it to make. It then tells the first ai to fix those errors and the process repeats over and over again until the second one can’t find any errors or the person making it is happy with it. Because of the way it works it only really works well on people who there’s a lot of vocal data. Same tech is used for photo and video deepfakes.
It is very surreal hearing Robot PhoenixSC, not because of AI but because of him not speaking in memes or getting sidetracked.
No crossword break. 💯 AI generated content.
Congratulations Peenix, great video.
Now I'm scared
Congratulations on the rebranding my guy. You've really out done yourself. 🤣😎👏
penix sc is real
Nice vid like always
Nuts
Behold poenix pc
It does kinda sound like him with the trash audio, but the wierd pause when it first mentioned Jeb, REALLY gave it away.
How do ppl not be getting these things, man?
0:36
*It's not my family, in your head, in your head, they are fighting*
Why is peenix skin so weird here 0:02
The scary part is, that with a decent enough computer (and/or enough patience) it's not actually that hard to make these covers and voice imitations anymore.
Why did they make an AI voice when Phoenix already sounds like an AI?
i know, right?
@padify no you dont
That's why it works so well.
Buff hot Phoenix
Terminator is slowly becoming more real
Ok, while I'm all up for technological progress this sort of thing makes me want to go back on my view of that. It's just uncanny. *Identity theft is not a joke, Jim! Millions of families suffer every year!*
Phoenix SC’s voice can never be replicated bc his fans will always find the flaws in AI generated voices.
unless if the ai generates a prompt in such a way that it formulates his entire personality that we see online.
1:07 - Everybody knows of Mojang's famous Israeli software engineer, "Jensjeb... Bergenstein." 🤣 Nice try, AI, but you're not skilled enough at human speech patterns (yet).
Petition for phoenix SC to sing renai circulation so they we can compare the ai version to the real one
2:38 bro I swear I just looked away for a second and I legitimately thought you were actually talking into the mic for the first 2-3 words 😢
Im pretty sure there is an AI that needs only 1 minute of talking footage and it can recreate the voice perfectly. Im pretty sure that happened here
Edit: im starting to think that isnt actually a thing but im not sure lol
what the fuck
If only it was possible, wah-wah
Phoenix has more than a thousand videos saying all kind of things, so no lol
Damn i wish you best and hope all things go well, PenisSc 💙💙
💀💀💀
I gotta put those songs in my playlist, they're straight up 🔥
Wonder if we can make as good use of this as good as the NileRed AI voice, gotta be funny.
Also: the AI generated PhoenixSC "As the world caves in" Cover is pretty funny, cant believe you didnt mention it
This video got me so interested, I looked into real-time voice transformation, and it took me 5 minutes to sound like GawrGura. This tech fucking scary bro
Remember, only the *real* PhoenixSC comes with the color changing stickers.
They talk like him, be like him, doesn’t give a f*ck like and might just be the next best him but not quite him!
That AI voice is so accurate
dont let him find out about the welcome to the internet cover guys (its really good)
Peni…. (OMG I want to stop my dirty minds)
Yooo thats my model!
glad everyones enjoying it lmao
Petition for the AI Phoenix and the AI Jschlatt to duo a performance
1:38 @PhoenixSC you can’t that requires a new AI chat bot. And that’s on a whole different beast
Well now... In 2024..
people dismissing this tech are missing the point. right now yes its fairly easy for us to tell its not phoenix. but A. not everyone will be as familiar with his voice and could be more easily confused especially if they have hearing problems.
b. This technology has progressed leaps and bounds in just a few short years, it wasnt long ago that to produce something of this quality you needed a professional studio and hundreds of hours of voice recordings to get a voice this good. now we have programs that do it from 30 second audio clips. how this tech will develop in a few more years is terrifying to think about...
For a second, I thought he started talking again, until he said "Mojang" with a j sound. That definitely gave it away.
I commented this before the video itself said that, so now everyone's going to think I'm copying it.
0:14 He got the Cash Banooca build my guy
I'll always remember you as PheonixSC >:)
please tell our bro peenixsc the nilegreen story, same thing here, your voice is always so calm so it's easy to make a deepfake
Hello Phoenix, the softwere they are using is called "Project RVC" if you want you can look into it!
it works kinda like a voice changer, so the pronunciation is based on the audio it is given, and yes it can be used in real time if someones computer is strong enough xdd
Im not surprised that the AI voice for him is so good. He talks in mostly the same tone so its probably easier to sample
HOL UP!!! People have used ai to make it sound like Phoenix did a cover of Renai Circulation!?!
That's impressive to hear, it sounds exactly like you with a lower quality mic.
Petition for PeenixSC to rename to PhoenixSC
Goal: 100 signatures
It's close, but I feel like I'm still able to tell you and the AI apart.
This is good, this is progress, now we need to pressure him into changing the second syllable to "nis"
that's honestly the most realistic ai voice i've seen it was genuinely kinda scary how close it was to his actual voice
Through my phone speaker it was actually pretty hard to hear that its ai generated but as soon as i plug my headphones in it was pretty clear. It is really convincing already but you can still hear the difference clearly if you know what to look for. It could be confused with a low Bitrate recording though which is really crazy to me how good those ais have gotten
NO UR NOT USELESS U MADE ME HAPPY IN MY LIFE I DONT WANT U TO LEAVE
(Apparantly while I was listening to ur classes u said that u would leave but PLS DONT)
ngl your voice in music brand will be pretty welcome