Microsoft is 100% Chinese now?
Why do the "speaker prompt" recordings have such bad audio quality compared to the "ground truth" recordings?
Otherwise interesting I guess but I thought this technology already existed.
>we didn't do anything new but we're rich so we used more hardware >golly gee it works exactly the same as ever >we are so innovative unlike the gwailo
It's actually got a lot of application in video games. No need to pay through the nose for Union Voice Actors when you can just lease John from accounting's voice and have the AI do it better. Likewise for anime, film, and television dubs and all. Hell, you could feasibly sample the original voice for a character, then have the AI speak the dialogue in their voice in a different language.
at this point AI already can into >illustrations >storytelling >voice acting
If allah is willing the west will finally catch up in terms of visual novels
seethe about it shekelstein
people realized they can no longer get ahead by slaving away and that hard work doesn't *actually* pay if you're an employee, so might as well enjoy yourself rather than slave away for the man
5 months ago
Anonymous
>I don't care about white society
This is what consoooomerism does to civilisation.
5 months ago
Anonymous
i don't live in a white or asian country
5 months ago
Anonymous
>seethe about it shekelstein
You know "shekelstein" is the first that'll profit from this, right? >can fire all now useless voice actors or blackmail them by telling them they're easily replaceable >can fabricate recording for deep fake purposes >can make virtual pajeets to scam grandmas
Actually it improves society quite a bit or do you prefer having most things being voiceless because "muh voice actors should be paid $100k for voicing a video game"
>fun is bad, I don't do anything fun, all I do is work 16 hours a day for shekelberg to afford another yacht, except when I stop to defend my cuckoldry on BOT
Unironically, have a nice day.
>fun is good, shekelberg provides me with new toys to play with, therefore shekelberg is good >the more toys I have the less I have to think about improving my life or the current state of society >I'm deadly afraid of shekelberg removing my toys, therefore I will defend him to the death and do anything he says
Grow up.
My favorite potential application is speech to text to speech. Voice transfer is a thing but imagine being able to play a game, speak into the mic, have the voice turned to text for the persistent chatbox, AND have the text (or voice) modified/transformed to match your character. Also if your character can understand the language being spoken or not in RPGs. No more accent memeing, no more thirsty fucks lusting over ethots (well, they'll always be a thing but it will greatly mitigate the issue anyway). Much better roleplay capabilities across the board, etc.
This is going to make for kino eroge voicepack mods
>ai generated video and voice
Yeah I’m thinking the TRUE era of cooming is almost here
>We are sorry, but our API does not allow pornographic or otherwise disturbing content. Your API access will be revoked on repeated infringements. >As per European Union's AI Safety Act (AISA), we are not allowed distribute the model freely and our API is the only way to interact with our models. Thank you for understanding.
Or, you know, realise the technology is coming and is here regardless as to what you want and learn that you can't trust people's voices. Boomers are gonna have trouble but so long as you aren't retarded and realise that this technology can be used against you, you will be on guard against audio-only information
>you will be on guard against audio-only information
and video information
and written information
and your lying eyes irl in ~15 years
welcome to the dark ages where you can only trust your Microsoft (tm) AI assistant
It exists so video, Audio and photographic evidence become as reliable as eye witness testimony. We already live in a post truth world but we are about to see what peak post truth looks like.
I think the reason for all Skyrim npcs sounding the same is not problem of no voice actors, but problem of not enough space. Sound files take lot of space and making every npc with unique set of 40 dialogue options is great waste of resources that could have been filled with music, 3D models or animations.
>audiobook narrators >voices for your animated movies >realistic screen readers >funny pranks on your friends using hidden speakers in your living room where you clone their voices and make them say funny things
Competitive shitposting is your only chance. If you really wanted to try and get competitive at this age with no former abilities then almost all doors are closed. Some fatties have had some luck in getting good at long distance running in their 30s though. Also much slimmer once they do.
No worries i work for the government and those fucking dinosaurs will be the last to adapt to anything high tech. They'd rather employ retards like me and only monitor the number of hours we work instead of monitoring efficiency. Makes them feel powerful or something. Joke's on them i eat and drink for free all day every day and i'm having good fun with my coworkers.
I'm going for the cyber rebel shitposter job, what are some unethical activities or ideas to do with ai in order to speedup the fall of modern civilization
I’m going to start training to be a pilot. It’s the one job that is guaranteed to never be automated because people will never get on a plane without a human flying it.
It was his research paper at MIT.
And honnestly, "we" asked him countless time. Or even to just tell how things work, we could have figured.
But he wanted to wait for his paper publication.
Oh well.
>But he wanted to wait for his paper publication.
What a moron, that is not how research works in deep learning.
If you have something, you rush to put it on arxiv, then you publish at nips or iclr or something of the sort. Those accept prepublished papers. Nobody publishes in journals in this field except fake bullshit for clout. The big guys all signed a pact/open letter saying they would never touch journals as well.
Since he didn't do that, it means he had nothing whatsoever.
Well, his website works, and very well.
But I know for a fact that some companies wanted to buy/rent his techs.
So, yeah, that's the burden of closed sources I presume.
I didn't lurk /pppv lastly, but last time I did 15 posted some examples and ask for feedback.
The quality wasn't better or worst than a year ago.
But...
He claimed that he now could keep quality high while adding any new voices in a very small training time (like a day or less, I don't remember exactly).
Which would be beaking too.
His fucking emotion are damn good too.
>his website works
I'm pretty sure it's been down for more than 75% of the time (it's currently down btw). His twitter is just him bitching constantly. He seems incredibly unpleasant so it was never a surprise to me he refused to open source anything since it's obvious he's just that kind of moron
The samples sound way worse than that tortoise one posted here the other day. However they are more consistent. Nobody sucking helium at the end of a longer clip.
>Uses a language model to emulate a person's voice with only three seconds of sample audio
Is that even novel? Not my expertise but I'm pretty sure there're already plenty of commercial software that does that, and does it faster. I remember seeing Yamaha doing something similar a couple years ago.
It's not novel, the novel claim is just a new performance level for the task. To be clear: just like image generation and text generation, those things have been done since the 70's and with deep learning since the 80's-90's depending on task, sometimes earlier (but it never became workable until ~2010, and it was pretty bad until recently for things like images and text generation).
>almost all artsy shit replaced by ai in 2-5 years >almost all computation-related stuff replaced by ai in 5-10 years >a fucking janitor or a truck driver will have a more marketable skillset than your average gayman hotshot
ahahahahaha what a fucking clown world, i can't wait until normies realize half the jobs worldwide are gone and that the rich fucks have zero intention of spreading their wealth around
>we will live to see the FBI create CP through stable diffusion and work to plant it on people of interest in order to justify the possession of literal digital art aka 0s and 1s as grounds for arresting people just because muh feelings and use fabricated voice lines thanks to AI as well as proof of admission of intent to distribute
Anyone can hire an expert witness to prove that the recording isn't legitimate. It's actually excellent because with the easy access of these technologies the burden of proof that the source is legitimate is greatly increased. This is no different the FBI hiring a voice actor to impersonate someone and admitting guilt.
How about you give a modern example instead of talking a political hot potato like the Holocaust? Oh yeah, because you don't have anything. Paper and verbal testimony isn't enough to convict anyone today. And rest assured 4chan gay, when they arrest you there will be plenty of real evidence.
5 months ago
Anonymous
The holocaust is a well known example it maximizes the potential that you understand what the argument tries to tell you. You are just a moron who feels scared of these words to the point that you start whining lmao
5 months ago
Anonymous
>Paper and verbal testimony isn't enough to convict anyone today
Btw please explain to me how that is? Why and when did we decide yes this year paper and testimony is truth and the next is not. I just want to know which cases we apply it to and why
Actually criminal courts (And all courts) have always worked that way and still do. You might also be surprised to learn that courts do not work off of written law, but rather on the basis of precedent. Only when no precedent exists is the written law actually consulted. When someone wants to contest precedent, it's a massive uphill battle that requires creating a case off of an edge result from the previous precedent that's against the written law in a part that hasn't had a precedent yet.
It's clear you're still learning, so this might be a clearer way to put what I think you're trying to say:
MoCap makes animating easier yet the hand animation industry has not collapsed.
My point was literally opposite. The argument was, if they don't have to, they would get rid of real actors. MoCap uses real actors, and they don't have to use it, but they still use it.
>Nothing happened with that >He doesn't realize that all Google services use tacotron nowadays >He doesn't realize that tacotron research paved the way for diffusion based neural network generation
I might be uninformed, but you are less informed.
https://i.imgur.com/MlUU6y2.jpg
Uses a language model to emulate a person's voice with only three seconds of sample audio.
https://valle-demo.github.io/
I personally only really want this so I can have AI read to me as I follow along because I have to read like 400 pages a week and hunting down audiobooks to bulldoze read shit I don't want to touch is a bit of a pain in the ass.
fuck me, i still haven't heard a single ai audio thing where the artefacting wasn't awful
why hasn't anyone trained a network that will upsample generated audio.
What's it like being a wonder-free head-in-sand zoomer, you cunting meatbag? Must be dreadful, honestly. Maybe it's bliss. Maybe in your abiding fucking ignorance of how insanely impressive this is, you can somehow postpone the reality of your deprecation. Fucking retard.
Have you used Google translate, maps, or assistant in the last year? It sounds great. When the AI misreads a street, it sounds like the voice actor doesn't know about your local podunk culture, not like an AI is broken.
Your brain has been turned to mush by an electronic training regimen that has made you flatly incapable of human thought patterns that the species has had for hundreds of thousands of years, and this programmatically-induced ignorance of the inhuman speed at which these things are occurring will ultimately be your undoing. You're already dead, kiddo.
hey cunt, if the speed at which things are progressing is human then why would having human thought patterns help at all.
keep your brainlet takes for your undergrad essays.
Humans can see this technology and say >that is fucking inane
Phone-ruined zoomers look at it and the only thing they can think is >yeah well it hasn't perfectly replicated a voice from THREE FUCKING SECONDS of training data so it's trash
You literally cannot understand the difference. You are physiologically incapable. This incapacity will be your demise. Take your booster, slave.
>all authors are chinese
What's going on here that I can't see? What guarantees that those guys won't flee back to china at some point and just give top trade secrets to the government?
>just give top trade secrets to the government
The chinese government is a corrupt joke. Give them any technology they will fuck it up somehow. Nobody tells each other the truth such is communism
Microsoft is 100% Chinese now?
Why do the "speaker prompt" recordings have such bad audio quality compared to the "ground truth" recordings?
Otherwise interesting I guess but I thought this technology already existed.
Stop samegaying street shitter. AI will soon replace your jobs and bring your country back 50 years.
Microsoft is 100% Chinese now?
Why do the "speaker prompt" recordings have such bad audio quality compared to the "ground truth" recordings?
Otherwise interesting I guess but I thought this technology already existed.
m$ china org is the only one that has been allowed to do ML at MSR for years. They used to have a team in montreal as well but it got killed off due to chronic mismanagement. Now they still have a token US team (MS turing) but it's really just do-nothing fuckery.
Mismanagement is fucking rife at Microsoft tbqhfam. They ruined muh gaems.
The whole contractor business is such a fucking clown show too, people make rube goldberg legal entity machines to remain working at ms for over a year
Last year I had an AI voice scam call. The voice was scary real. It was clear the pajeet on the other end was speaking into a conversion device. Social security boomers don't stand a fucking chance.
>Last year I had an AI voice scam call. The voice was scary real. It was clear the pajeet on the other end was speaking into a conversion device. Social security boomers don't stand a fucking chance.
some telecommunications companies are outright banning voip providers that use their fixed line/mobile networks. some already ban sms with any bit dot ly links too (used for device hijacking - banned on this site as well). future will be fine. telcos need to actually do their fucking jobs and protect their networks.
>"During the pre-training stage, we scale up the TTS training data to 60K hours of English speech which is hundreds of times larger than existing systems" >with only three seconds of sample audio.
what a dishonest and delusional fucking chink.
I can barely perceive any improvement over baseline and it's not clear to me the improvement I do notice is from actual better generation or saturation issues.
I will build a model with audio from my waifu seiyuu and hook it up to a chatbot emulating her personality. The future is bright from my moms basement.
I mean copying an existing vtuber's voice and making "her" tell the viewer to cum on her toes or whatever. Possibly while also ripping off that vtuber's avatar and streaming setup so it seems like an authentic piece of content. I feel like that's inevitable but I don't know how far we are from it happening.
I think if they continue replacing creative jobs, it will be Marvel and Pixar (Disney) soon making them their ropes. Imagine being the dev who made AI based on a movie character and then the creator of that very movie character goes up to you and spits in your face while trying to sue you for that very AI.
>be me >minding my own business >be accused of being a terrorist >wtf.jpeg >demand proof >proof is a deepfake of me saying "I love isis" >mfw sent to jail for no reason
big tech bids for a.i. tech companies all the time, only costs a few million to cause lots of misdirection and deny others tech. I wager there are many hidden gems waiting for the right time to go public.
Microsoft is 100% Chinese now?
Why do the "speaker prompt" recordings have such bad audio quality compared to the "ground truth" recordings?
Otherwise interesting I guess but I thought this technology already existed.
>Microsoft is 100% Chinese now?
Poojeet
This is what happens when you say it's racist for white men to do well in america and try and keep them out of corporations and universities.
>Microsoft is 100% Chinese now?
Chinks are the only ones who can make improvements in this space, and you know it.
guffaws
literally 100% of advancement in the space are made by garden gnomes and some whites.
If you look at the paper, clearly not the case.
>we didn't do anything new but we're rich so we used more hardware
>golly gee it works exactly the same as ever
>we are so innovative unlike the gwailo
Where?
There's no legitimate reason to make this technology. It can only be used to trick people into doing something they wouldn't otherwise do. It's evil.
It's actually got a lot of application in video games. No need to pay through the nose for Union Voice Actors when you can just lease John from accounting's voice and have the AI do it better. Likewise for anime, film, and television dubs and all. Hell, you could feasibly sample the original voice for a character, then have the AI speak the dialogue in their voice in a different language.
This is going to make for kino eroge voicepack mods
at this point AI already can into
>illustrations
>storytelling
>voice acting
If allah is willing the west will finally catch up in terms of visual novels
you ready for thousands of shitty social justice blm vns on steam for 69.95?
>It's actually got a lot of application for my electronic toys.
Ahah, wow! Fuck improving society, I get better toys guys!
>Fuck improving society
yes
quite literally not my problem
>not my problem
Actual Black behavior.
seethe about it shekelstein
people realized they can no longer get ahead by slaving away and that hard work doesn't *actually* pay if you're an employee, so might as well enjoy yourself rather than slave away for the man
>I don't care about white society
This is what consoooomerism does to civilisation.
i don't live in a white or asian country
>seethe about it shekelstein
You know "shekelstein" is the first that'll profit from this, right?
>can fire all now useless voice actors or blackmail them by telling them they're easily replaceable
>can fabricate recording for deep fake purposes
>can make virtual pajeets to scam grandmas
>chud reddit cuck frog
>HUR DUR NOT MY PROBLEM LOL IM SUCH A CHAD! XD
Actually it improves society quite a bit or do you prefer having most things being voiceless because "muh voice actors should be paid $100k for voicing a video game"
>fun is bad, I don't do anything fun, all I do is work 16 hours a day for shekelberg to afford another yacht, except when I stop to defend my cuckoldry on BOT
Unironically, have a nice day.
>fun is good, shekelberg provides me with new toys to play with, therefore shekelberg is good
>the more toys I have the less I have to think about improving my life or the current state of society
>I'm deadly afraid of shekelberg removing my toys, therefore I will defend him to the death and do anything he says
Grow up.
Games aren't toys any more than film and tv is. They are vehicles for capturing attention and programming people.
>Muh vidya
>Muh Chinese cartoons
Go consooooooom elsewhere gay
I have never bought a video game
My favorite potential application is speech to text to speech. Voice transfer is a thing but imagine being able to play a game, speak into the mic, have the voice turned to text for the persistent chatbox, AND have the text (or voice) modified/transformed to match your character. Also if your character can understand the language being spoken or not in RPGs. No more accent memeing, no more thirsty fucks lusting over ethots (well, they'll always be a thing but it will greatly mitigate the issue anyway). Much better roleplay capabilities across the board, etc.
Its great for my coom purposes
>ai generated video and voice
Yeah I’m thinking the TRUE era of cooming is almost here
>We are sorry, but our API does not allow pornographic or otherwise disturbing content. Your API access will be revoked on repeated infringements.
>As per European Union's AI Safety Act (AISA), we are not allowed distribute the model freely and our API is the only way to interact with our models. Thank you for understanding.
They just released the papers explaining how to make the AI yourself assuming you have the talent and resources.
> Just build your own $250 million AI bro
good morning kind grandma, i am rajesh your grandson, i am having police issue, please send $500 in giftcard
*in masculine holly wood actor voice*
good morning sir this is john smith I'd like to congratulate on winning a free trip to the Bahamas
kek but if they are going to do this surely they will also use ai to general the text
Such an easy trick to pull off that there's no way it won't become common.
Why not? Translation followed by speech synthesis sounds useful for communication.
Or, you know, realise the technology is coming and is here regardless as to what you want and learn that you can't trust people's voices. Boomers are gonna have trouble but so long as you aren't retarded and realise that this technology can be used against you, you will be on guard against audio-only information
>you will be on guard against audio-only information
and video information
and written information
and your lying eyes irl in ~15 years
welcome to the dark ages where you can only trust your Microsoft (tm) AI assistant
It exists so video, Audio and photographic evidence become as reliable as eye witness testimony. We already live in a post truth world but we are about to see what peak post truth looks like.
>There's no legitimate reason to make this technology.
unlimited voice acting in your skyrim mods.
I think the reason for all Skyrim npcs sounding the same is not problem of no voice actors, but problem of not enough space. Sound files take lot of space and making every npc with unique set of 40 dialogue options is great waste of resources that could have been filled with music, 3D models or animations.
it exists to finally put voice actors out of employment, fuck those overpaid seiyuus
I will custom mod my anime shows with whichever voices I like
>audiobook narrators
>voices for your animated movies
>realistic screen readers
>funny pranks on your friends using hidden speakers in your living room where you clone their voices and make them say funny things
I don’t get it. Haven’t we been warning everyone about automation for about 20 years now? This is exactly how it begins.
>It can only be used to trick people into doing something they wouldn't otherwise do.
Crypto exists for this reason and half the zoomers shill it.
>tfw you chose anything as a profession besides being a plumber
Gonna pivot to being an athlete. What sport should an out of shape 34yo take up?
>What sport should an out of shape 34yo take up?
competitive shitposting
Competitive shitposting is your only chance. If you really wanted to try and get competitive at this age with no former abilities then almost all doors are closed. Some fatties have had some luck in getting good at long distance running in their 30s though. Also much slimmer once they do.
Run at high speed until you collapse. Do it again tomorrow. All your ills and pains will magically fade away.
At that age only bb gayry on social media trying to sell your shitty supplements while pinning on the low
unironically curling or bowling
don't mind me just 3d printing a pipe
There are only so many shitters to unclog.
No worries i work for the government and those fucking dinosaurs will be the last to adapt to anything high tech. They'd rather employ retards like me and only monitor the number of hours we work instead of monitoring efficiency. Makes them feel powerful or something. Joke's on them i eat and drink for free all day every day and i'm having good fun with my coworkers.
I'm going for the cyber rebel shitposter job, what are some unethical activities or ideas to do with ai in order to speedup the fall of modern civilization
I’m going to start training to be a pilot. It’s the one job that is guaranteed to never be automated because people will never get on a plane without a human flying it.
15.ai guy is going to have a meltdown lol
>only three seconds of audio needed
>it's potentially open source
Yeah. But he got too complacent and is not active, so...
don't care about the childrens show addict
good, gay should have open sourced his shit while he had the chance
Yeah, /pppv already had a look at that.
It was his research paper at MIT.
And honnestly, "we" asked him countless time. Or even to just tell how things work, we could have figured.
But he wanted to wait for his paper publication.
Oh well.
>But he wanted to wait for his paper publication.
What a moron, that is not how research works in deep learning.
If you have something, you rush to put it on arxiv, then you publish at nips or iclr or something of the sort. Those accept prepublished papers. Nobody publishes in journals in this field except fake bullshit for clout. The big guys all signed a pact/open letter saying they would never touch journals as well.
Since he didn't do that, it means he had nothing whatsoever.
Well, his website works, and very well.
But I know for a fact that some companies wanted to buy/rent his techs.
So, yeah, that's the burden of closed sources I presume.
I didn't lurk /pppv lastly, but last time I did 15 posted some examples and ask for feedback.
The quality wasn't better or worst than a year ago.
But...
He claimed that he now could keep quality high while adding any new voices in a very small training time (like a day or less, I don't remember exactly).
Which would be beaking too.
His fucking emotion are damn good too.
>his website works
I'm pretty sure it's been down for more than 75% of the time (it's currently down btw). His twitter is just him bitching constantly. He seems incredibly unpleasant so it was never a surprise to me he refused to open source anything since it's obvious he's just that kind of moron
Gonna be interesting when AI replaces all of hollywood. All actors, voice actors, writers, editors, and directors. All filmmakers. All crew.
Interesting though, flops will be virtually impossible, you can make a movie for free.
>ai starts churning out so many movies we will need an ai to review them and tell us which are worth watching
>thinks garden gnomes will replace garden gnomes
First it better replace some good lawyers because the entire fucking hollywood isn't exactly the same as some starving brazilian coom artists.
The samples sound way worse than that tortoise one posted here the other day. However they are more consistent. Nobody sucking helium at the end of a longer clip.
>Uses a language model to emulate a person's voice with only three seconds of sample audio
Is that even novel? Not my expertise but I'm pretty sure there're already plenty of commercial software that does that, and does it faster. I remember seeing Yamaha doing something similar a couple years ago.
It's not novel, the novel claim is just a new performance level for the task. To be clear: just like image generation and text generation, those things have been done since the 70's and with deep learning since the 80's-90's depending on task, sometimes earlier (but it never became workable until ~2010, and it was pretty bad until recently for things like images and text generation).
>almost all artsy shit replaced by ai in 2-5 years
>almost all computation-related stuff replaced by ai in 5-10 years
>a fucking janitor or a truck driver will have a more marketable skillset than your average gayman hotshot
ahahahahaha what a fucking clown world, i can't wait until normies realize half the jobs worldwide are gone and that the rich fucks have zero intention of spreading their wealth around
you r missing the big picture
the more ephemeral the economy becomes the less their money is worth
the difference in effort between being rich and poor keeps shrinking
>Microsoft has made their own
>Nvidia made their own
Not looking good Loonix and AMD sisters...
>we will live to see the FBI create CP through stable diffusion and work to plant it on people of interest in order to justify the possession of literal digital art aka 0s and 1s as grounds for arresting people just because muh feelings and use fabricated voice lines thanks to AI as well as proof of admission of intent to distribute
Anyone can hire an expert witness to prove that the recording isn't legitimate. It's actually excellent because with the easy access of these technologies the burden of proof that the source is legitimate is greatly increased. This is no different the FBI hiring a voice actor to impersonate someone and admitting guilt.
They accepted paper and word as truth for the holocaust. I think it doesnt matter if they say so
Completely irrelevant for modern criminal cases so you can fuck off your 4chan shit.
Relevant as the truth is what they want it to be. They dont give you bootlicker any reward either if you need to disappear you will
How about you give a modern example instead of talking a political hot potato like the Holocaust? Oh yeah, because you don't have anything. Paper and verbal testimony isn't enough to convict anyone today. And rest assured 4chan gay, when they arrest you there will be plenty of real evidence.
The holocaust is a well known example it maximizes the potential that you understand what the argument tries to tell you. You are just a moron who feels scared of these words to the point that you start whining lmao
>Paper and verbal testimony isn't enough to convict anyone today
Btw please explain to me how that is? Why and when did we decide yes this year paper and testimony is truth and the next is not. I just want to know which cases we apply it to and why
>4chan bad
>antifa good
Your brain is rotten
Actually criminal courts (And all courts) have always worked that way and still do. You might also be surprised to learn that courts do not work off of written law, but rather on the basis of precedent. Only when no precedent exists is the written law actually consulted. When someone wants to contest precedent, it's a massive uphill battle that requires creating a case off of an edge result from the previous precedent that's against the written law in a part that hasn't had a precedent yet.
The first world uses civil law, not Common law, mutt.
Both are used in different parts of the first world, although it's debatable if the anglosphere is actually first world or 4th world in 2023.
>the highly tight knit voice actor industry is about to collapse
Thank god
Animators could animate without MoCap since forever. Than why MoCap is still used, huh?
>than
ESL, Go
It's clear you're still learning, so this might be a clearer way to put what I think you're trying to say:
MoCap makes animating easier yet the hand animation industry has not collapsed.
It basically has though. The only thing done by hand now is cleaning mocap data.
I don't care what the conversation was about, I was trying to help him with his English grammar.
My point was literally opposite. The argument was, if they don't have to, they would get rid of real actors. MoCap uses real actors, and they don't have to use it, but they still use it.
I have seen the same thing before a few years ago and nothing happened with that.
>Nothing happened with that
>He doesn't realize that all Google services use tacotron nowadays
>He doesn't realize that tacotron research paved the way for diffusion based neural network generation
I might be uninformed, but you are less informed.
I personally only really want this so I can have AI read to me as I follow along because I have to read like 400 pages a week and hunting down audiobooks to bulldoze read shit I don't want to touch is a bit of a pain in the ass.
the world is so close to ai generated anime that is regional dubbed and subbed produced like in a cheese cake factory
its a wonderful thing to imagine but in reality it certainly create a complete new generation of degens
>in reality it certainly create a complete new generation of degens
A new de-generation, if you will.
fuck me, i still haven't heard a single ai audio thing where the artefacting wasn't awful
why hasn't anyone trained a network that will upsample generated audio.
What's it like being a wonder-free head-in-sand zoomer, you cunting meatbag? Must be dreadful, honestly. Maybe it's bliss. Maybe in your abiding fucking ignorance of how insanely impressive this is, you can somehow postpone the reality of your deprecation. Fucking retard.
what's like being a low-standards, boomer with hearing loss and so few interactions with human beings that you think this is passable.
The only thing ITT that isn't passing is you, chud. have a nice day.
deaf people like yourself are a drain on society, cocksucker
Your tombstone will carry your birth name.
Have you used Google translate, maps, or assistant in the last year? It sounds great. When the AI misreads a street, it sounds like the voice actor doesn't know about your local podunk culture, not like an AI is broken.
>It sounds great.
it really doesn't.
no one will come to your funeral
Dilate
seethe
(you wanna make out?)
Your brain has been turned to mush by an electronic training regimen that has made you flatly incapable of human thought patterns that the species has had for hundreds of thousands of years, and this programmatically-induced ignorance of the inhuman speed at which these things are occurring will ultimately be your undoing. You're already dead, kiddo.
hey cunt, if the speed at which things are progressing is human then why would having human thought patterns help at all.
keep your brainlet takes for your undergrad essays.
>is inhuman*
Humans can see this technology and say
>that is fucking inane
Phone-ruined zoomers look at it and the only thing they can think is
>yeah well it hasn't perfectly replicated a voice from THREE FUCKING SECONDS of training data so it's trash
You literally cannot understand the difference. You are physiologically incapable. This incapacity will be your demise. Take your booster, slave.
take some mood stabilisers, malcom gladwell.
https://voca.ro/12Q7WCzdeuaj
death of artistry, humanity
we are more than organized information producing more organized information
>all authors are chinese
What's going on here that I can't see? What guarantees that those guys won't flee back to china at some point and just give top trade secrets to the government?
>just give top trade secrets to the government
The chinese government is a corrupt joke. Give them any technology they will fuck it up somehow. Nobody tells each other the truth such is communism
Stop samegaying street shitter. AI will soon replace your jobs and bring your country back 50 years.
m$ china org is the only one that has been allowed to do ML at MSR for years. They used to have a team in montreal as well but it got killed off due to chronic mismanagement. Now they still have a token US team (MS turing) but it's really just do-nothing fuckery.
Mismanagement is fucking rife at Microsoft tbqhfam. They ruined muh gaems.
The whole contractor business is such a fucking clown show too, people make rube goldberg legal entity machines to remain working at ms for over a year
ITT: bot fight
wheres the female voice? I want to write my own custom naughty stuff
Last year I had an AI voice scam call. The voice was scary real. It was clear the pajeet on the other end was speaking into a conversion device. Social security boomers don't stand a fucking chance.
>Last year I had an AI voice scam call. The voice was scary real. It was clear the pajeet on the other end was speaking into a conversion device. Social security boomers don't stand a fucking chance.
some telecommunications companies are outright banning voip providers that use their fixed line/mobile networks. some already ban sms with any bit dot ly links too (used for device hijacking - banned on this site as well). future will be fine. telcos need to actually do their fucking jobs and protect their networks.
its over
>"During the pre-training stage, we scale up the TTS training data to 60K hours of English speech which is hundreds of times larger than existing systems"
>with only three seconds of sample audio.
what a dishonest and delusional fucking chink.
read this...
Stop spamming /v/, schizo.
Any demo? Sauce code?
https://github.com/microsoft/unilm/tree/master/valle
Not yet.
I can barely perceive any improvement over baseline and it's not clear to me the improvement I do notice is from actual better generation or saturation issues.
Wallah!
Fucking CCP spies
Looks like voice actors will be out of job. Sad.
Great for indie game development. RPGs can actually have a lot diverging dialogue options again too.
I will build a model with audio from my waifu seiyuu and hook it up to a chatbot emulating her personality. The future is bright from my moms basement.
>woman claims that you raped he
>she shows up in court with a recording of '''you''' saying that you did in fact rape her
>itsover.png
How long until someone trains this on vtubers and we get anime girl jerk off instruction videos?
>on vtubers
You mean for vtubers?
I mean copying an existing vtuber's voice and making "her" tell the viewer to cum on her toes or whatever. Possibly while also ripping off that vtuber's avatar and streaming setup so it seems like an authentic piece of content. I feel like that's inevitable but I don't know how far we are from it happening.
Y'know that VTubers already use AI for their voice, right?..
What difference does it make if the voice is authentic or synthetic for the person trying to copy it?
>vall-e
>one letter off of dall-e as to avoid getting sued
fucking kek
You're aware that "DALL-E" name is a parody of WALL-E, a Pixar movie?
>
Also Salvador Dali. It's pretty clever
>muh pixar, muh marvel
hope devs get roped by AI insurrection first
I think if they continue replacing creative jobs, it will be Marvel and Pixar (Disney) soon making them their ropes. Imagine being the dev who made AI based on a movie character and then the creator of that very movie character goes up to you and spits in your face while trying to sue you for that very AI.
>muh lawsuit
Not enforceable.
>be me
>minding my own business
>be accused of being a terrorist
>wtf.jpeg
>demand proof
>proof is a deepfake of me saying "I love isis"
>mfw sent to jail for no reason
Y'know that video footage can't be used as proof? It's now new, it couldn't be used as proof since inception, because of tempering.
didn't know that
Still can be used in court, just not as a proof.
> no open source implementation
straight to the trash it goes
why the fuck people in this field came to agree this is acceptable
>artists defeated
>voice actors destroyed
Just need to get AI wrtiting to a decent quality and then I can generate my own VNs and anime
So how many gay companies have a model like this and who's is the best? In the past few months Nvidia and now Microsoft has made one
big tech bids for a.i. tech companies all the time, only costs a few million to cause lots of misdirection and deny others tech. I wager there are many hidden gems waiting for the right time to go public.