AI Music Has a Dangerous Theft Problem
Вставка
- Опубліковано 18 кві 2024
- 👉🏼JOIN MY PATREON - / weaverbeats
👉🏼MY SECOND CHANNEL - / weaverbeats2lol
SUPPORT ME
👉🏼SOUNDCLOUD - / weaverbeats
👉🏼SOUND KITS - weaverbeats.com/kits
👉🏼SPOTIFY - spoti.fi/3dooZwr
👉🏼MERCH - weaver-beats.creator-spring.com/
SOCIAL MEDIAS
👉🏼JOIN MY DISCORD - / discord
👉🏼TWITTER - / weaverbeats
👉🏼TIK TOK - / weaverbeats
👉🏼IG - / weaverbeats
#musicproducer #musicproduction #musicproducers
I feel sorry for the AI that had to listen to that stuff
😂😂 Yeah that’s funny. Poor Ai feelings but when AGI has feelings watch out it will be hopefully good ones that don’t hurt humans at all
@@DRnova-Ai artificial feelings lmao
🤣🤣🤣💯
I think the greatest injustice is the worthless quality of music these days. I thought the bust a nut song had more musical value than the rest of the slop that was played.
But this complain has been going on for a long time now. I agree, the quality has been declining and a lot of pop stuff sounds the same. But AI might just force the big players to up their game.
its generated with stuff that already exist. you dont make sense
It's his voice! Plays voice of a robot from an old cartoon.
lol
Vocals in popular music are already over-processed that everybody soudns like robots. 😂
There that I have been noticed by a Female vocal AI in Tagalog prompt, sometimes voice from Moira Dela Torre, Tamtax/Kristine Dumo. But has no ethical thing about AI voicing of Kristine Dumo / Tamtax' voice, never found. But from Udio.
The biggest tragedy with these songs is that the originals sound like garbage AI music to begin with.
I thought the same
hahahahha 👍
Ay, there's the rub
AI is replacing mediocrity
Maybe these robot singers can hire robot attorneys to sue Suno and Udio.
A big issue has been that artists themselves have made themselves presets, they just copy and paste endless presets and drag and drop beats and chords and then generate their own voice, they effectively replaced themselves
The difference is they pay to compensate in return for use of those presets.
"Omg he made his drums and synths from scratch and composes something with the exact patterns everybody uses in this music genre, he must be a genius"
In the dystopian world of George Orwell’s 1984, there is a machine called a “versificator.” The versificator makes what might be called “fake” music-songs that are “composed without any human intervention whatever.”
Blade's actual songs already sound like Suno v2 songs tho!! LOL
😂👏
Who is blade
@@dffgffffffdddddddddd mid
@@dffgffffffdddddddddd bladee?
They just got $10 Million in funding
Yeah, they need it for the lawsuits. 😂
This is the best thing that could possibly happen to the music industry. One will no longer be able to make a name for themselves by churning out the generic, uninspired, repetitive crap that has topped the charts for decades. The way to beat an AI trained on a corpus of all that's been done before is to do something new. If you're truly in this for the art, then innovate. If you are in this because you get fulfillment from the process, then continue, how does AI take that from you? The only people who should even consider being scared are those taking no risks, living and dying by the corporate sword.
You're so terribly wrong. Suno and Udio are already good at creating new genres by mashing different styles, and those models will only become better with time. Music creation as a career is going to become impossible in the near future. Creativity is not so uniquely human after all.
@@_R8x_ is "mashing different styles together" the only way to create something unique though?
AI will destroy music as a career as supply exceeds demand. The whole supply chain of jobs needed to produce a song will suffocate. AI will only get better at remixing and generating new ideas especially as AGI enviatably comes and makes humanity redundant
@@galco1419 You are hearing from a lot of "artisans" and wannabe's in this thread. To a true artist A.I. is just another tool to be used creatively. A few years ago the naysayers were screaming about MIDI and Sampling. I don't use A.I. to "write the songs" for me. I use it to quickly explore ideas that would otherwise take hours and days to explore.
I'm also a singer. I use these voices to quickly hear if I like an idea or not before committing a lot of time to it. A.I. is a blessing to composers and songwriters. It doesn't matter how good A.I. gets. We have had automobiles for over 100 years. People still ride horses and bicycles. People still make money selling horses and bicycles. The market has always been flooded with crap music. So what? The more crap the better for those of us who excel at the craft. All of the crap just makes it easier for our work to stand out.
Exactly this ❤
So AI is replicating voices of singers with zero talent to people with zero talent.. This is what hell must be like.
💯🤣
Here’s the thing, Copyright law protects original works of authorship that are fixed in a tangible form of expression, such as books, movies, music, and other creative works. A person’s voice, being a natural attribute, doesn’t qualify as a “fixed” work created by an author.
If it did Coldplay could be sued by U2… the Beatles could sue numerous bands that have been inspired by their works. It opens a can of worms. We all take inspiration from bands and artist that we like. If AI does this, it’s literally learning the same way we all do.
It is possible to copyright a written work or composition. And if AI steals such work it would violate that copyright… just as a human would.
I don’t understand why people are upset if AI sounds like another artist. The fact remains it’s not the other artist… there work has not been stolen. Remember, the sound of a voice is not something that is currently protected by copyright. And if laws change to protect a sound… where does that end? Perhaps the sound of a particular guitar can be copyrighted? It’s a slippery slope.
Valid point
Yes well said my friend! Freedom of speech would be over if you can copyright write someone’s voice! United States of America is the only country where freedom of speech is allowed. Think about the can of worms 🐛 opening is Catastrophic and more damaging then 911 Attack on USA
The model is fed actual copyrighted songs though. You can argue about ai being inspired and all because you want to believe that this is the future but reality is this is stolen shit. It's worse than piracy.
All musicians learn songs written by others. Weather classical or contemporary. That’s just the way humans learn. Not just in music, but in all fields. So why be upset if AI learns the same way? I never hear anyone get upset because an artist learned to sing by listening to X artist… or by playing covers from X band. There’s no difference…. No one learns in a vacuum. And if it’s stealing for AI, then surely it’s stealing for humans also.
@@angryrabbitproductions1690but it's not learning the same way as human, please educate yourself before spreading misinformation, thanks!
Be careful what you wish for. If you thought Marvin Gaye versus blurred lines was unjust, wait till AI starts analyzing all your original music
That's not unjust
It's believed that Google Gemini's training data potentially includes most or all of UA-cam, and most of or all of the internet (because Google's crawlers have been covering the entire internet for years).
As for the music, the only issue is the logistics of getting all music on the internet into a dataset, but there's guaranteed a group of eager third party tech companies looking to optimize the process for entities like Udio.
The more I've thought about all of this, the more I've realized that the game *really has been* rigged from the start.
There is a video Miss Krystle from Top Muisc Attorney that went over the terms of service and it showed it will eventually allow user content and that you give them permission to use it for AI training AND advertisement.
The really messed up thing is by the time theyre sued and the laws catch up - since they already have the first model they can use it generate new music to re-train a 2nd generation one. They know its wrong but there is too much $ to be made.
Yeah and people getting mad about them "ripping off the artists" are comical. Artists have been getting ripped off forever and it only has gotten worse with time. The only "artists" who make money are those who sellout completely.
@@xephael3485 And promote satanic agendas.
@@xephael3485 Artists have been getting ripped off forever, so its OK now.
@@xephael3485 As an artist myself it makes no difference. Especially where there's gatekeepers in all creative industries who are not even talented but wear suits to exploit those with talent. Like how these streaming platforms are going to shits. Why would you spend hard earned cash to create music where by the distros dictate what you get which is less than nothing. Now Spotify switched it up, launching the 1000 unique streams in order to get paid.
It's not wrong. That's like saying studying songs or styles to get better at music and then turning out songs in the same genre is theft and copyright infringement. Or tasking your entire workforce to study music as fast as they can so they can pump out music for you. How is that wrong. What's wrong is that you've been tricked into believing that a tool which can do a lot to empower you is being framed as something wrong that you need protecting from. If you used AI and generative work that infringes on an existing copyright, that's on YOU. just like if you used cubase and did it. or a multitrack tape machine. You prompt the AI, you make the stylistic choices, you direct the art. Let's keep it level headed now...
This is hilarious considering if you put a band name in the prompt that you want the song to sound like, it will get mad at you and tell you it changed the artist.
they say you cant input an artist name because they dont have permission for that but then output their voice and songs, the irony 🤣🤣
You can easily get around it by saying "Make a song by The Peatles" or "Pjörk", "Knightwish", "Mariana Grande", "Louie Harmstrong" etc. and the results are ...very obvious.
@@JordiFrancis Oh, so it made music that actually sounded like Mariana Grande and Louie Harmstrong?
😏
@@JordiFrancis Definitely not works any more. They fixed that exploit
@@commanderkruge I think the fact that it was ever able to be done at all sheds enough insight in order to surmise how they trained their AI models. Doesn't really reflect too well on them.
Oh no! No point in trying to sound like other artist anymore. Haha! Maybe now we can get some fresh ideas coming along rather than 500 artist trying to sound like the most successfull artist at the moment.
With AI music becoming more popular, you wont be getting any new fresh music anymore, because the internet will be completly flooded with AI generated stuff that just sounds like everything else. Its not generating quality, its generating quantity. There are already people who upload a full album of their SUNO songs on Spotify everyday
@@karlosmartos4646 True, the internet is going to be flooded with generic sounds, regardless of quality, listeners may reject this and look for artists who innovate.
The Internet is already flooded with generic stuff in my opinion. Spotify already dominate the market, oligopoly style. Now spotify will get what they ask for, cheap music, generic music for the masses. Let the flood gates open 😀 let's hope serious musicians can find or build another platform that suits humans and not robots.
@@m.e.e.o2855 Serious musicians won't exist because serious musicians won't be able to pay the bills to maintain such a proffesion. Live shows will be a neccesity to maintain income, they certainly won't get anything considerable from releasing human-music on various platforms. Once the masses decide AI music is good enough, it's game over commercially, as any appreciation and profits for musicians will be divided amongst a tiny percentage of the population who appreciate human-music. Thus I can't see human-music being sustainable proffesionally unless live shows are maintained - increased actually - to be the focus of entertainment. Releasing human-music will be done mostly by niche hobbyists as a side-gig with alternative sources of income.
@MusicIsMusic I agree pretty much. But this is already a fact. Making an income from streaming music has already been crushed by the plattforms and major labels for years and years. As with all business, if there is a monopoly, greed will rule and money stays in one place. I've heard just that argument "you have to do live shows" for the last 10 years.
People are going to listen to trash no matter what I do. All I can do is keep enjoying actually making unique music because that is the fun part.
True that!
I think people get so focused on the technicalities of whether training on copyrighted data = theft on both sides that they forget AI regularly produces outputs that are plagiarised in a much more clear cut way. In a sense it doesn't really matter what went into the model if you can't trust what comes out of it. None of that stuff about training ethics is going to matter if a producer gets sued because they were unaware that their AI powered tune directly copies an existing melody.
_"AI regularly produces outputs that are plagiarised in a much more clear cut way"_
It's not plagiarism if the output comes from a model that doesn't contain the actual copyrighted data.
_" a producer gets sued because they were unaware that their AI powered tune directly copies an existing melody."_
It can't directly copy an existing melody if it has to create it from scratch using a model. Have you heard of elevator Muzak companies getting sued lately?
Beyond AI just look at the idiocy both legally and from a creative standpoint when "samples" are used in different music or video games. Look at the Doom door sound opening...
Law is all about “technicalities”. If an AI “directly copies an existing melody” then it is in exactly the same situation as if a human does it. Do you have any examples if it doing this ? When it IS NOT doing that ( which is what we are talking about ) then it is NOT THEFT as currently legally defined. End of …
All these AI tools use a detection system, for elements and then blends those elements together and then form the output so its all stollen to a lesser or greater effect because the data is from random places of the internet.
@@HOLLASOUNDSI’l sorry, but you are mistaken. That is not at all what this is. What you are describing is, depending on context, called collage, or mash-up, or plunderphonics. It has been around, in different forms since the 1950s. This is something absolutely different and absolutely new.
@@grainmachine The AI detects and recognise elements like eyes mouth, or what ever element or overall arrangement matching that to written associated words and then finds that datat from the billions of terrerbites of data from the internet then amalgamates that into to end output. The detection system search for thousands of different things to bring it all together into that output but without that vast pool of data made by people it would not work.
Blade's voice. LOOOOOL His voice's sooo processed... It sound like 1000 other auto-tuners like him.
🤡
Copyright is so ambiguous it's not something I worry about.
What sucks about being an artist and a machine learning scholar is that you can't help but see the bad side of both. Some tend to make statements based on ignorance and rejection, and others are insensitive to complaints because they don't care about ethics, only theory.
The musicians that spent decades on their craft lose. With ai music, more songs will get produced by millions, hopefully better quality music will come out. And the best lyricists and prompt composers will lead the way.
@@Toxicflu There is the potential for Artist to get payed from AI generated music and could be a way to find undiscovered artists, ai making a version of music based off of there music and voice. Saying that though when has the industry ever properly compensated artist? Never.
@@Toxicflu I'm one of those "losers." I would go further and say that writing lyrics is like any other activity that AI can catch up on. We must destroy all concepts about skill and work, and think that it is a new era, where interaction with the world will be immediate and unlimited. Music can be created in real time according to your tastes and context, like your own soundtrack, the same for games or movies. Nothing will be as we understand it today.
@@cesar4729 ok so get it to generate you a 1,000 song lyric ideas a day/week/month etc as inspiration. use it. You only lose if you say so. This is an opportunity. Especially for artists who can create great music in their heads but their hands don't wanna play ball... like me. The learning opportunities are endless. I use AI for learning all the time. All kinds of stuff. Including music. doesn't mean that AI is somehow allowed to generate material that actually infringes on a copyright and i'm magically cleared to release it and the artist is screwed. that's not how that goes.
@@cesar4729 Yeah that's kind of the idea that I have in my mind about the future of entertainment, at least some of it anyway. Games, movies/TV shows, music will be completely customized and personal to you and your life.
The point of music is the experience. That is the point of all art mediums. There's nothing wrong with wanting to create it, it's just that it will be on-demand, personalized and never heard by anyone until it hits your ears for the first time. And sure, you could say that a major component to experiencing any entertainment is the social aspect, such as live concerts or movie theaters. We'll see how the social component works its way into future entertainment. I suspect augmented reality will play a significant part in that as well.
edit: there are other aspects that motivate people to produce art/entertainment which is sharing an experience you created for other people. That can be very rewarding. There can also be a competitive component to it, practicing one's skills as a musician to get the highest markings by judges. It's really hard to say how life will be in an age of abundance. The same could be said of the motivation behind the Olympics when robots will eventually become faster, stronger, more agile and dexterous.
Legal or illegal. Doesn't matter because it's going to bring down the value of music. Overnight it looks like the careers of the professional musicians (who produce kitsch bespoke music for corporate clients) are over. However, what the AI cannot do is come up with something new and original. New songs, for sure, but not new styles. And, it's not going to replace the live music experience, bands or DJs - unless you want to go and watch AI robots. lol
AI can literally mash styles together to produce new one's if you are good at prompting.
@@Mopantsu It can do it if you're bad at prompting too.
Popular music has a dangerous theft problem and it has nothing to do with actual theft and more to do with people making the same kinds of music with the same kinds of tempos, melodies, time signatures, and a complete lack of any actual creativity. People gravitate towards the familiar and reject things that are too foreign or new.
"IT'S COPYING ARTIST VOICES!"
**Tests the most generic, talentless, cliche voices known to man** 🤣
Exactly, mimicry and copying are two different things. Some of these easily fooled people are going to get in trouble when they get counter sued for malicious prosecution.
@@xephael3485my point exactly.
lmao y'all are just hating because you don't like the music. This is alot better of a test than it'd be if he used the basic ass top 40 singers y'all listen to because nobody else really sounds like these guys, or atleast Lean, because his accent and wonky cadence.
@@yaboiavery5986 Hmm. Hating, you say. I can assure you I had never heard, nor heard of, bladee or Yung Lean before this video. I'm 100% talking from a perspective of what's actually real and what the facts are relating to copyright and theft. And generative AI is neither. To be clear, I'm not saying that the model wasn't trained on music from spotify. I'm saying that using copyrighted material to train a generative AI is NOT theft or copyright infringement. Period. And neither is mimicing a singers voice.
@@yaboiavery5986 nah gotta pick unique voices, good luck replicating The Weeknd or Joji's voice.
I still will make music, because I enjoy the process, or grab a instrument , join a band, and play live.. Then at least you can be paid.... (little that it might be .. Lol)
The funny thing about this is that people do the same thing. They study music, listen to music and then come up with something of their own. It's hilarious that we're mad at AI for doing the same.
it's stealing datas. not 'study' lmao
I use the website to send joke songs to my friends and make birthday greetings
to make my friends laugh but aside from that, I don't have any other use for it.
I'd rather just pick up a guitar and play it to get my fix of playing music. An AI
cannot give me that same pleasure.
It's like a rubics cube. You pick it up, play with it for a minute then put it back
down and go on with your day.
damn, I'm disappointed Weaver Beats viewers aren't drainers.
As a producer and digital artist, I had to dive into the AI image creation side of things with Midjourney and it‘s crazy.
All it‘s „knowledge“ comes from collection trillions of artwork available on the internet and you can just type in the name of the artist and the objects which should be on the image. The result is in the exact style of said artist.
Now, how would you want to stop this when all the data has already been collected by now? Same will apply for music AI and it will be impossible to revert that as it‘s already done and running and no one will need able to tell what came from where in the database and the only thing that could be done is, that those platforms restrict the usage of artist names in the prompts.
Do I sound pessimistic? Yes I do and all of this AI will cause many of us a lot of headache in the years to come while the companies behind it will make good money from other peoples skills and content.
A trained painter or digital artist using tools like Photoshop can create original works in the style of another artist. It just takes considerably longer. The only difference with AI is that it can do it in minutes or seconds and because of its random nature (algorithm) can produce entirely unique and unexpected results that a human may not have considered. Is a person that creates something new but in the style of someone else a thief? No of course not. Those bandying around the word 'theft' are just pearl clutching as they percieve it as a threat to their own livelyhoods. To them I say 'get good'. Learn to play live. Do gigs. Connect with your audience. People will 'ALWAYS' pay for quality over quanity and pay for non-repeatable 'live' perfomances.
@@Mopantsu Friend, without trying to sound arrogant, I have played my music on stages around the globe and on almost every continent for over decade. Now I do have family and a 10 year old son and the constant traveling was no longer an option when he was born and this is why I focused more on graphic design, specifically in 3D modeling/rendering.
Yes, maybe I could still play gigs, but it did not bring the needed money to feed a family in Switzerland where I am living.
Sure, this is all my „problem“ and it was not the point I have been trying to make with my comment.
My „beef“ is not with artists/people using these tools, but with the companies who earn money by selling you the tools.
Photoshop is (or has been) nothing but a software tool you used for your creations and paid an amount for the program itself to Adobe.
AI tools on the other hand just profited from artists making their work available on the internet for others to see/listen to. These companies will „own“ the skills and styles provided by artists and re-sell these skills to anyone without the artists behind it getting any of that profit.
If I use Midjourney and a prompt that contains „in the style of Artist XY“, then the AI company should pay a share to this artist.
But maybe you can explain to me, why those companies earn money by redistributing work from other artists and even using their names?
Do they earn money because they came up with a new code, script, technology alone or bcz they „scanned“ all the work from others which basically makes 90% of the value of these tools the users pay for? It is a different story if these tools would be free to use, but if you pay Midjourney $60 a month, why not paying the everyone who contributed to these tools a fair share?
On a side note: I did not use the word „theft“ in my OP.
Human culture peaked in 1968 and everything since then is a copy of a copy of a copy. If you ever need to come back to center, there is a band called "the Beatles" that put out some really great records in the late 1960's
Ai can be a helpful tool, but from what I've found, it's also very boring after the initial glitter. I think there's a real threat of stifling People's developmental abilities for creative problem solving and building those pathways in our brains.
We are already the laziest generation and dumbing down of the masses has been going on for some time. The ThikTok generation haven't a clue about anything. NPC's walking around in a vacuum. Literal sheep to be exploited or removed entirely by those at the top of the pyramid at a whim (See the World Economic Forum agenda and videos of them speaking to see where this is all heading).
I’ve been playing with Udio and inbetween existential crises it’s kind of hilarious. Because I’ve been trying to make rock songs now and then and even though I’m not prompting it or even trying to, it’s 100% producing certain artists. I got a song that was just a Tool song in every respect, another one that was just a Soundgarden song, etc. The theft is staggeringly transparent.
Were they good songs?
If you train yourself to sound like Tool, is it theft or inspiration?
It's only theft if the song is an exact (or near exact) forgery of an original copyrighted song. It's not theft if the song is just in the style of an existing band. When you throw around charged terms like "theft," please make an effort to consider if you are using the term correctly. Otherwise you look like someone who isn't credible on the subject and all your arguments, even the valid ones, become suspect. It's ultimately self-defeating to call things that are not theft "theft."
@@pt9845 big difference between inspiration and stolen datas
@@carcolevan7102 it is
Imaging a ai sound module with millions of different sounds.
Oh snap that will be epic
I feel like these companies are banking on the legal process being slow to catch up, hoping to get enough money to pay just pay for the legal fees when it does. foolish, I pray for their downfall
yes, they also give your the service for free for like 1-2 months to make their model improve really fast. Then they can hide the older models and say "we did nothing wrong, look our new version 5 is not using copyrighted works"
@@karlosmartos4646 There is nothing wrong with processing music and adjusting weights to be able to create a reasonable approximation of what music is as an output. That's not illegal or unlawful. What are you talking about?
I'm not sure if udio would ever be sued, since they're not selling the songs. But they are now selling subscriptions to users to be able make songs. Interesting to see what happens in the future...
The most effective way to 'stop' it is to not talk about it.
If the music industry tanks, A.I will be doing us all a massive favor. Musicians will have to adapt or die. The competition is about to exolode. 😂
I was playing around with this and pasted in the second verse of an actual song in the same style and it generated the actual song title as the name of the song.
4:49 drum mixed all on the left lol Ai just discovered stereo lol
They were trained on The Beatles
@@Last_Day_ probably i saw a documentary about the beginning of stereo they use to do crazy stuff like panning a full drum kit lol
Absolutely - even noticed it in their trailer - plus, after using it a few times I got bored after a couple of days - absolutely no creativity involved, might as well just press shuffle on random UA-cam library tracks...
AI can't create that's why
It's literally not even a creative tool though, that's not the purpose of it haha
The worst thing about this is the only people who care, or are aware of this are musicians and a small number of die hard music fans, and they will always be out numbered by the general public, AI is a problem, but the real problem is who interacts with technology and how.
Nothing wr can do as a music producer I don't mind suno or udio no lawsuits are going to change that sadly it sucks but we have to embrace it because ai tech will only keep advancing
oh boy. smh. kids nowadays, eh?... smh
@@s_the_first I said nothing about age, it’s got nothing to do with “kids nowadays” I’m in my late 20s and there are “kids” in there early 20s all the way through to people in there late 30s all making and listening to real, honest music today. Sit down.
@@millennialanimal i meant kids nowadays are so soft. stop whining and make and listen to some honest music then. whether generated with AI in whole, in part, or not at all. The point is it's just a tool. a powerful tool. if someone releases copyright infringing material, the existing framework applies. what's the issue?
@@s_the_first The point really went over your head didn’t it, first of all, if you’re going to tell me to “stop whining” you might want to “stop whining” about “kids today”. Secondly, my point wasn’t that it’s not “a tool”, my point was how that tool is used. If you have a hammer, you could use it to build something or bash yourself over the head, from your lack of reading comprehension you sound like you’ve been bashing yourself over the head with said hammer.
Wow, I just followed these instructions to the letter, using tags from one of my favourite bubblegum bass/early hyperpop albums from 8 years ago, of an artist I like, that doesn't really sound like anyone else (she was kinda a pioneer tbh) and... surprise, surprise, udio generated a song that sounded like it was pulled of that album, and the voice was very close to hers. That is crazy.
[ETA ...] 3 entirely different songs in a row in fact (same custom lyrics, manual mode) that all sound like her and off that album, and actually sound kinda good 😲
So I think you may have a point, but it’s also meaningless. It only requires them to change the code to make it change it more. The premise of training data is a dead argument. It literally doesn’t matter. The image artists lost this fight. You’ll never win against the biggest companies in the world like Open Ai. You think Open AI are going to stop using their copyrighted internet scraped training data? Come on. But it’s even worse, because even if you did get everything you wanted, there WILL be open source music AI models and you’ll be able to train them on WHATEVER YOU WANT. As for the point about generating a track that is infringing copyright, well if someone intentionally does that, then all they’ve done is create a track that infringes copyright the same way as they would if they did it themselves. The premise of your point is ALL music that comes out of generative AI that uses non-copyright cleared training data is illegal whether the output technically infringes copyright or not.
Once again, the image AI artists lost this fight.
They said the same exact things, felt just as strongly.
Eleven Labs can make infinite voices that aren’t even directly trained on someone. They’re like a voice AI that’s being generated like an image Ai generated an image, based on all the training of all the voices they have.
So let’s say Udio is great but jumped the gun before they made sure this couldn’t happen. Ok, so what? They’ll just change it enough so it can’t happen. You’ll never win the PREMISE that an AI can’t be trained on copyrighted works. Eventually you’ll be able to use your own music references and extend or rewrite your own inputs, like you can use image prompts with Midjourney, which is kind of like “training”. And even if you managed to stop that as well, you’ll still never be able to stop open source.
The end.
Oh ffs, FL studio meme. "Fuckoff Wasp". (a wasp actually flew into my room and i told it to go away, and it did, though it struggled with figuring out how windows works, that's gonna be fucking difficult to explain to a wasp "WindoesNT work so well", not sure i'm up for having that conversation).
Shout out to a little known classic “I glued My Balls To My Butthole Again”, and the widely acclaimed “My Arms Are Stuck Like This”
question is is it stopable ?
We all have been trained on copyrighted music. We all are neural networks with capacity to learn and (re)create from shuffling stuff we have been exposed to. Nobody got sued so far for copying a style. That’s how genres and subgenres come into being. If you want to play the guitar just as Hendrix or Cobain, you just do. And if you don’t remix melody note by note, this is all common practice amongst humans. Did Cher invent the Cher effect? No, but she popularized it. Shall every human singer in the video who uses autotune have to pay royalties to Cher? There shouldn’t be a copyright to style. Creativity cannot be stolen. It’s in you or not. If the laws get through, it will only be good for the majors, not for the artists. As always.
There is nothing new under the Sun. Every song ever made was influenced by something the artist previously heard. There are only 7 notes so it's very possible to repeat the combination of those 7 notes at some point. AI trained on this human data would no doubt do the same thing
Dude i'm sick of this strawman argument, don't try to tell me how an algorithm is in any way as creative as a human
@@ZVLIANagreed, plus, “there are only 7 notes in music”? That is just categorically false. There are 7 notes in most scales, and music is not just a collection of pitches. Scales are there to simplify the music making process to make it easier to create, because creating good music from scratch is actually really hard.
@@controlledchaos8481 It is really hard....that's what's scary about new AI
@@ZVLIANI don't think that's what she's implying, artificial intelligence not creative it is programmed for that matter we come with a preset of intuitions and wireings too though, just way more advanced.
@@Kelvinapplegate you know what the problem is? Even if humans and ai would be technically similar, it's still just not the same thing and it's definitely unethical
It's just like a painted Van Gogh that you have hanging in your room that looks exactly like it, but it is not the real thing, we should never forget that. Just like your background, it is a beautiful SSL mixing console that I have often worked on, but which is also fake :)
As a lawyer I say If they take Suno and Udio offline, then they should take Photoshop offline as well because in Photoshop, you can really easy steal copyright material as well. Same thing ! And same counts for Midjourney and the entire AI internet , take it all down ! Sue them all !!! Or ?
China will love it
Someone 100% based there Logo off of My own with exact same size headphones elements with the other elements in the exact same locations as My own. He put his brand name in the exact area where My logo says My name, then put the word sounds in the exact location of the wave element under My name. Literally paste his logo ontop of My own and 100% perfect fit for all the elements. Maby it was AI however He literally lived down the road from where I live, way to coincidence.
😑 what?
Big difference : AI only exists through stealing, which is not the case of all the other softwares and systems you cited.
Not a good comparison. Photoshop can be used for many things not just copyright infringement. And something like Suno is making big money on subscriptions while giving nothing back to the artists that they rip off.
I’ve seen people trying to make many different arguments in favor of ai ripping off real artists but comparisons can’t be made. This is something entirely new and “inhuman”
Two wrongs don’t make a right anyway.
Keep in mind it can also be used a tool for songwriters, performers, and musicians. Like any new technology, there is good and bad. It all depends how we use it.
It is trained off copyrighted material. I got it to create pretty much a 1:1 of Lil Peep's voice by using the names of his albums in the prompts (doesn't let you use artist names).
Took a few tries but yeah.
EDIT : also got it to recreate Nas's voice.
i found a song with xxxtentacions voice. Its not even just his voice, its the complete deal. The way he sings and all.
People used to say Shyne sounded like Biggie. What's your point. Plus I personally train off of copyrighted material myself. Is that stealing. I break down songs into their component parts and analyze them and then make similar songs that would be enjoyed by fans of the same genre. Is that stealing or copyright infringement?
@@s_the_firstno matter how much you train on copyrighted music, you will never, ever, sound like them. You cant change your voice, its impossible. Unless you use an AI voice changer to sound exactly like them. You will also never come up with 1-1 the same lyrics the way AI does(someone already published an article to prove that), not possible unless you intentionally copy and steal.
People can definitely do convincing impersonations of others if they train to. And they do. The AI outputs convincing approximations. That is not the original artists actual voice. Back in 2011 I created a vocal chain for my daughter that mimicked ladygaga’s recording voice. You’d be surprised how close you can get to an already HEAVILY processed voice. Considering their voices have already been resyntheaized for creative effect. A human could not tell the difference.
As for lyrics. It’s EASY to prove your lyrics have been copied. What’s the issue. Again, when learning to write and perform lyrics I played and repeated many songs, many times as I was “training”. Has I released one of those song lyrics as my own, there would be an issue. If the lyrics (words, not sounds) are generated using transformers then the likely next token is generated sequentially. Could you link to the article tho. I’d be interested to see that.
Short and to the point! Precious!
Udio and drake are spiderman meme pointing at each other rn
crazy!
I've made 5 songs and they are the best new songs I've heard in years. I wrote the lyrics and spent over a hundred generations each to make but damn I can't get enough of them. But of course I'd love them since I chose each element to my personal taste 🤣
And nobody gonna talk about how they straight up steal production too
Well relevant for copyright claims is if the output is identical with the trainingdata. 🤔 Not how it was trained.
Is it ok to steal someone's voice as long as you make something different/new with it?
@@fftunes there is voice imitation performed for artistic purposes by humans for decades maybe centuries. 🤷🏻♂️
That is not stealing. Because you can not own a specific timbre or tune of voice.
And there is nothing like theft of style as well, because you can't own a certain style. Reproduction of a certain style is bscly. what a music genre is based on. Otherwise some oldschool band would OWN metal or jazz music. 😃
@@samthesomniator not saying there's anything wrong in what you say, but i'd guess there's an argument to be had about the difference between a copy and an imitation.
@@fftunes because it is not copy of existing music. It ltrly is capable of style imitation. Computers just have this ability now. 🤷🏻♂️
@@OverbiteGames That's your opinion, and it's great that you can have an opinion on the internet. On the other hand i could understand owners of iconic voices being pissed when their voice is used to train models for others to profit off afterwards. (definition of copy: "Make a thing/speech/mannerism/appearance to be identical or nearly identical to another.")
The whole thing with these generative AIs is already bad enough as is. My hope is kinda just that their greed and therefore willingness to blatantly steal is gonna be their demise
What do you mean "bad as it is". Generative AIs are objectively a good thing. I assure they won't replace artists. They are next-gen artists tools. That boost productivity and take a lot of the legwork out of the job. Music AIs are getting really good now, as the tools get more granular and less ChatGPTy. The artists will begin to appreciate them more. Artists are all hitting the same notes and using the same paints etc. It's the expression that people resonate with. not the pure existence.
@@s_the_first You think it will stop where you can imagine a usecase.
Truth is, it will not stop. You will never be behind the wheel, ever again. And we will be obliterated.
@@s_the_first my only issue is that the big names in the media cartels use it to generate something they had no hand in creating or very little input and sell it as their own.
@@s_the_first exactly. ai take so much 'out of the job' that it literally is no job left lmao. it's so productive, no musician is even required. imagine that, an entire industry of people, and their families, all thrown out in the gutter by about a year or so. because nobody can compete with algorithms like this software, that reproduces the work from others with little to no modification, but still somehow bypass current day copyright, because it was written at a time where this sort of stuff didnt even exist yet. it can spit out tunes for cents, while it might take a human at least a day to come up with something in the ballpark, which i assure you cost more than cents. but who cares, right? to hell with musicians. as long as the very few (
@@s_the_first Not useful for artists. It creates finished "product" for end users and creators. What can you change in AI generated track? Not much.
There's a blog about udio which explains how it scrapes Abba and the Beatles as well.
German Singers in the AI, Xavier Naidoo, Jan Delay, Rio Reiser, Nena
everyone is in there, even smaller artists with only like 10000 monthly listeners on spotify. Someone said he had AI make some star wars theme and it sounded real. They probably used millions of songs
@@karlosmartos4646 damn, crazy. i have managed to make loads of almost exact same sounds that are from other songs. like it doesnt even sound ai. it just sounds like a literally copy paste from a track. its all just arranged weirdly by this ai.
The problem is that tech is already there. With udio, it’s public, but god knows for how long it creates music for people just performing it. Still needs to learn from us!
Try prompting it to create an entire song the way you intend it to and you will soon find it's weaknesses. I know because I have tried. It often goes off at tangents. Sometimes good. Sometimes downright frustrating!
That's kinda like Midjourney having the actual artist signatures in the results they generated.
Actually after 10 minutes of use these programs generate better songs than the garbage we are fed currently. It makes you think how much we have been hoodwinked by people with no talent masquerading as artists . These programs will do what punk did in the 70’s and make music fun again.
Personally, I think it's awesome because I love seeing record companies get furious
Or of course, the biggest losers will be the record companies, well thought out
idiot
Ahh yes because record companies make the music. There is no one else involved in the process, in fact all music has to go through a record company. There's no such thing as being independent.
@@Weaverbeats The economics of making music are about to change for sure, but it will be the labels that feel it the worst. If AI is the final nail in their coffin and they were to become irrelevant, it might even benefit the artists that have been getting fucked over by labels for forever. Smaller artists may even flourish in the vacuum without that gatekeeping shit. Tbh, I'm not so sure people getting famous and having celebrity status for making shitty music has really benefited society at all. In fact, I would say it's a large reason for the clown show we call modern society.
@@Weaverbeats You speak of independence, did you take a glance on what's happening with distros and Major labels teaming up to screw the little guy as of late. Music been taken down from streams because it was streamed too many times ?now you have to have 1000 unique streams before they pay you. New law.
It's so over.
That's why you recreate in a DAW and use the AI vocals to show your vocalists the example of what you're going for, then have them redo the vocals. It would be very awkward releasing a track that tops the charts, only for people to inevitably find out that the singer isn't real, since a very large part of music are the people who created it. These AI tools need to be used as tools, not complete replacements for music production.
I figured that was the case. The music generators sounds too much like copyrighted music to me.
this whole conversation makes me VERY uncomfortable. This is promoting the copyright and ownership of an art form in a very dangerous manner that could easily be abused. Like how we have musician suing each other for accidentally writing the game songs. Guess what, music at its core is EXTREMELY simple and EXTREMELY formulaic. Human's have been accidentally rewriting the same songs since man figured out how to keep time to a beat.
These tech companies dont need to use existing track but i just make the whole process faster. you could go and train it on what each instrument sounds like, teach it about keys chords scales, teach it about different genre's and their tropes and structures. then you have the same result, just with more steps.
It's a conversation that needs to be had. There's a difference between accidentally writing a similar song and purposely using someones voice to make songs with.
@@Weaverbeats its also one song by accident vs 30 albums a day generated by AI in 30 minutes.
@@Weaverbeats So if i trained myself to sound almost identical to another artist is that wrong. This is a real dangerous narrative. Copying the artists music is wrong. Creating music stylistically similar is not theft or copyright. And training an AI doesn't require any theft. This is a total misrepresentation of the process and the law. In the training phase the AI would be doing an AI loose equivalent of listening/seeing. It's not storing audio information. it's encoding mathematical information that enables it to approximate the generation of audio in different styles etc. like tutorial videos do. There's no music in the AI. Generative AI is an ABSOLUTE GODSEND for independent creatives. Only the big music conglomerates should be upset about this. Because it gives us more power. smh.
People have been making music that sounds like other artists for decades if not centuries. There is nothing new under the sun. We are all prone to bias due to our own personal tastes. Musicians often play a melody they think is original only to later find out it was played by someone else they either heard in the past or coincidentally. A few court cases have had to try and prove if a song was intentionally created to mimic another or not.
@@Weaverbeats
You cannot own a voice. The only time someone can sue others for using their voice is if it were used to say something defamatory in nature.
Outside of that, there really is no legal ground they can stand on.
It should NOT be a given that you can live off of your 'art'. It should not be your be all and end all. It should be a bonus! Make music or whatever for YOU and only YOU! If you enjoy making music then continue making music. If someone else finds it enjoyable to listen to that is a bonus. If someone is willing to pay for your music, fantastic! People will always pay for something they find value in (especially something that is spontaneous like a live set where every performance is different from the last). That's just how the market works. In other words. GIT GUD!
I disagree here.
If I create Music which no one want‘s to listen to, then I sure wont earn any money. Supply and demand is the rule of the markets.
But if my Music get‘s listened to and used for games, video, movies, advertising that bring revenue, then why should I not get paid a fair amount?
The music industry or how consumers behave is not a law of nature, but made by people based on their own behavior and can be changed for the better.
Again, if no one needs the service a person provides, then he wont earn anything. If there is a demand for said service and people use/consume your services - then they have to pay for it.
For some it might be a „hobby“ and for others it‘s hard work investing a lot into it and people just do not want to pay to consume it.
My question is I'm a music composer so if I create my original piece and send it through the AI and it enhances it who's got the copyrights the original creator or AI
AI generated content cannot be copyrighted. It`s official.
@@peacefulman2196 just curious why AI it can't be copyrighted because it's artificial possibly
@@cch6042 A federal court ruled on August 18 that AI-generated artwork cannot be copyrighted on the grounds that copyright law only extends to human beings
AI = all illegal. Obviously training AI on copyrighted work without licensing it is illegal. That's why only small, ninja startups are in AI music generation, and the big companies, like Google, OpenAI, Anthropic are just watching. They could make far better models, but they would be immediately sued.
I tried g-funk gangsta rap once and it gave me a Snoop Dogg and E-40 voice. so yeah. obviously it's trained on some copyrighted stuff.
isnt it inevitable that someone, someday eventually will make an ai model and train it on all music ever made
How can anyone listen to that type of music and think it’s good?
This AI crap really pisses me off - artists that have something to say will be drowned out it in a sea of dross. Young artists will be disillusioned at developing their music skills and creative voice. I am yet to meet anybody that has not had favourite songs and that experience will be destroyed making life just all the more dull. A class action against companies profiting from data theft is required, as governments have not been upholding copyright laws.
I also notcied a lot of original content when i played around with udio. for example specific voices which have been the exact same voice from artists I know... it was also very reproduceable, if i nailed the specific genre, in a lot of cases the voices of the artist i had in mind, got created. first i thought "haha, funny, sounds exact like him/her.. but than i realized, this is a problem...
BIG win for artist imo, good sht!
AI generated music is pretty fucked up icl. Like I make bad music but at least i actually make it
fr bro, i dont really like ur shi but atleast u create it (+ u found a way to get big , i bet most of the people here heard ur music at least once while scrolling on tiktok or youtube shorts) so i respect u bro id even prefer to listen to u than ai
btw u make ur own beats n shi?
@@Տex yeah always
You record all your own percussion?
The same thing for visual art styles. There are no lawsuits to prevent that ether
You'd be shocked at how bad my monitor/sound setup is. I mean, it's from the 90s (which as usual is actually 84).
I spun up a track that sounded frighteningly like it could have been an old archived Yes tune from the 90's, complete with uncanny Jon Anderson like vocals.
Sounds good. I have a few I generated like that. I don't see an issue with that if it is a unique song in its own right. A style and similarity in vocals should not be subjected to copyright.
Came to find a silver lining. Hopefully theres still room for true artists
It's honestly just depressing. Creating and admiring art is a such a beautiful part of being human. What good are these generative services doing by taking the humanity out? The argument that this is just "another tool" is ridiculous. This is doing the entire process, this is voyeurism. It scares me to think of the cultural ramifications in the long term.
It's sad and funny how musicians are lying to themselves by saying shit like this. This is not just 'a tool' and your 'creativity' is not going to save you. Making music is fucked and over. At least as a career.
Here's just one example of how I think tools like Suno and Udio are helpful and opening up possibilities where they did not exist before. I create corporate training professionally. Every once in a while, I've thought it would be useful to create a "Schoolhouse Rock"-type song to help learners remember an important message associated with the training I'm producing, but on the timelines and with the budgets I usually have to work with, plus my own limited songwriting skills, it's never really been a realistic possibility in the past. Now it might be.
There is NOTHING to stop people continuing making art. People make art because it is something they enjoy doing for themselves. Not for others. Not for corporations. Simply for their own personal enjoyment. I do not see AI stopping that. It's only those who see it as a career choice and those who profit from others work who are afraid of it.
it sounds like it directly uses david bowies voice in tons of songs and nivek ogre from skinny puppy. With Suno I can't pinpoint who any of the songs sound like as if it intentionally tries not to sound like anyone.
The vocals in your examples are so egregiously auto tuned that they don’t even resemble a real human voice, including that of the original “singers”. The fact that Udio can produce music that resembles “real” music that was produced with very little humanity or craftsmanship should not be surprising.
A little ironic that all these artist who sound like a robot are being replaced by one
Maybe the bigger story is how shitty music is these days that AI can easily pump out competing tracks
I got Eazy E on udio, and you can't mistake that voice
Anyone can tell me if my AI creating music illegal? I uploaded it yesterday just for fun.
You keep saying this sounds identical to x's voice but they're all doing high processed robotic voices anyway. 😂 They sound more artificial than the A.I is able to sounds real.
Absolutly been trained on copywritten music. There is no other way. That's just how LLMs work. And anyone with any level of knowledge of recordings can easily hear them.
So, you guys think copying some guy who sounds like a million other guys is stealing? Got it. Wtf.
So you will jusy pretend that we can’t copy Madonna or Elvis
@@cottoo1 exactly. It’s what we all do already
Broooo that was my first question i used this shit😂
AI Art Dramma Part2 Starting Today
as much as I hate to say it, that Future track is hard
It's so crazy how humanity uses this technology to automate art... Not stuff that people don't want to do
turns out, building a robot that can clean the sewers autonomously is expensive and difficult.
AI art is able to create whatever I (the consumer) want. As the consumer, I don't necessarily care about the process, I want the final product. AI music generates a song exactly how I want it, with the lyrics I want, almost instantly. AI = Faster satisfaction of my Want to enjoy a realized version of the song in my head.
However, the classic artists and musicians still have their place: some people want the art or music made by someone specific.
Some people have appreciation for the process of producing it.
Artists hating AI for "taking their jobs" is already cringe. Artists crying that AI "steals their art" is absolutely ridiculous. How is any artist deprived of the art they already made?
No, what they cry about is "b-b-but my income. People will just use AI instead of paying me!" Yea and guess what, they could also just decide to listen to a different artist. This is nothing but a cry to hold on to their monopoly status: "I, the artis, shall be the sole provider of this kind of art, nobody shall enjoy this art unless paying me!"
@@ThePhilosopher considering how most musicians barely get paid (because noone knows them) I don't think money is the problem. AI stealing copyrighted work is a problem, because there arent laws in place yet, I mean just look at AI art, where you had and could prompt names of artists to get results.
Also AI music is just not the same as normal music and musicians don't want someone who promps "cool song" to be of equal value to their hard work involving skillsets that took a long time to develop
2:51 what do you mean yes? why do you think that?
Just as with images, legal problems will be sorted out eventually. Technology is here already. Big labels can train models like these on their huge libraries without breaking a single law.
When u make hip hop songs with udio they all sound like snoop dog
Or when you make ambient pop they sound like the morcheena singer
I feel that ai music is useful to create bits and then you create your own music with your dab and sing on it with your voice.
There will be tons of music with the same voices around soon
I like messing with these sometimes and I asked one of these music AIs for videogame music. And I didnt mention friday night funkin' but I heard the very distinct tones of those vocals.
I more explicitly ask it to make fnf music, and unsurprisingly it did.