Follow Jesse @SyncMyMusic : www.youtube.com/@SyncMyMusic 🟡Ready to become your own record label? Get started now!: www.topmusicattorney.com/becomeyourownrecordlabel 📺 Get Your Music Into Tv/Film/Games: www.topmusicattorney.com/sync 💻 #1 Resource For Artists And Producers: www.topmusicattorney.com ⚖ Hire An Attorney: www.delgadoentertainmentlaw.com 📰 Get My Newsletter: www.topmusicattorney.com/newsletter/ 📄 8 Essential Music Contracts Bundle: www.topmusicattorney.com/contractsbundle 🎧 Listen To The TMA Podcast On All Streaming: www.topmusicattorney.com/podcast 🟢 Free Stuff: www.topmusicattorney.com/free-stuff
In the 1970s, welders taught machines how to weld ships. Then there were welders' strikes in shipyards. They were afraid for their existence. It was the same with horse cart owners when the car was created. It's the same with AI. In the end, it is the recipient who decides which product he will use, not his grandchildren. If 1,000 people steal something, they go to prison. if 10 million do it. This becomes law.
Nothing is original. Everything is stolen to some extent. , "If I have seen further than others, it is by standing upon the shoulders of giants," Newton
@@SekundSun1986 So are you. Music basically is copying. The genres are for a reason. There are also very few options that makes sense musically or mathematically in order to make modern music. If there is a certain harmonies or chord-progressions you only have so many options to get a feeling across. There is a reason why the perfect fifth sound so good. (Although in our equal temperament we don't have perfect fifths since all "errors" are smeared out). People have the misconception that they own the songs. They do not. They can have a copyright on their *expression* of ideas. But not the underlying ideas themselves. Taking a body of work and analyzing it for common methods of technical nature isn't taking the expression away. Even if it offsets musicians it does not offset one single work in any appreciable manner. To quote an AI: "Summary AI Training on Music: Using music to train AI by analyzing aggregate features (not the specific expressions) is generally considered less invasive and more abstract than human listening, though it involves complex legal considerations. Ownership Misconception: People often misunderstand what they own in a song. Copyright protects specific expressions, not the general musical ideas or common elements. Legal Distinction: The distinction between owning a song and owning the rights to its specific expression is crucial and informs how copyright applies to both human and AI-generated music. In conclusion, while copyright law can be complex and nuanced, the core principle remains that it protects the specific, original expression of creative works, not the underlying ideas or common elements. This distinction is vital for understanding the legal landscape surrounding AI training on music and the broader implications for music ownership and creativity."
1:21 just like to point out that while Udio is fast, it literally doesn't "spit out full tracks in 30 seconds". It generates 2 30 seconds of a possible track, and then you have to extend it in 30 second chunks at a time to make it a full track,
The paid version of Udo was just released and depending on what tier you get you can generate more songs than that and the update has way more control of how you can edit your songs. It's going to just keep improving.
Yep. I'd say it actually takes someone with a good ear for music and a talent for writing decent lyrics, to get an output from udio that would give anyone in the music industry good cause to feel "uneasy." And as you point out, that sure as hell ain't happening in 30 seconds..
I think the problem here, and what actually makes people afraid of AI (or at least should make them afraid), is not the technology itself but rather that no government is willing to regulate the market in such way that musicians and their jobs are protected. The paradeigm as we expirience it is that governments -obey corporate interests- lobby with corporates, and not human beings.
I know some members of Congress are trying to get that fair wage for musicians act past that has to deal with streaming we'll see if somebody can come together for some AI regulation. The film industry was able to make some moves in regards to AI. The problem is musicians don't unionize in the same way. We'll see if that changes
I'm more worried about the 0 regulation in "defence" technology, tanks jets you name it its integrating. the pollution that would come from a world battle bot arena is a worry
In my opinion, the best AI music will be generated by actual musicians. Knowing the lingo is essential for prompting, having a grasp on genres, styles, and moods is equally essential. With Udio, you are regenerating every 30 seconds, and determining which extensions sound better will be optimally achieved by a trained ear, not the average joe. There will always be a job for those already in the industry, you just need to adapt. Who is going to make the AI music for the companies? Not the company themselves, and while anyone can make AI music, musicians will still be the best at it. Musicians will need to create and CURATE the best results. They already have the industry connections and network, and can work to a brief. An average joe with Udio does not have those same necessary qualifications to just take over your job. Adapt, embrace, evolve.
Bro just tarin chatGPT or Lama or Claude 3 to prompt for Udio.... You got a point but unfortunately other AI can be trained to be a musiccian and know how to prompt, lot of people already do this to prompt on AI music generators... So maybe there will always be a human to guide, but no need for him to be musician.. All is going the wrong way for humans unfortunately.
Same with software engineering. Experienced seniors become true 10x with AI assistance over juniors. Clients can prompt all day long but still need someone with skills to put the requirements together. It's a process working with AI to build higher quality products.
🙏 Thank you so much for being on the show. You're insight it's incredible. I love how candidly and honest you speak about the situation. I look forward to our channels collaborating again in the future 🖖
The only people that care at all about most music used as underscore on television are the composers and music libraries that provide the music for the shows. I talk to non-composers all the time about my participation in this industry and almost every single person I've talked to about it said something like "I never really noticed" or "I had no idea there was so much music used in a single episode." We are a niche population, 99% of the world gives zero s***s about what we do, how it sounds, or how it was created in the first place. ;)
@@SineEyed yes, that is exactly how streaming works. However, a license is granted by the copyright holder to, spotify for example, so that they can stream. Here the copyright holder is not granting the ML model a license. Hence likely infringement. (See, e.g., American Broadcasting Companies, Inc. v. Aereo, Inc.,) You can make an argument for fair use, like Open AI did in their suit with NY Times. But its a poor argument for music, because in the case mentioned above, it is the facts of the articles that are being reproduced.
8:00 I disagree. It’s not just about money. It’s about human culture. Humans learning from other humans is much different than AIs learning from humans. AIs reduce an artist’s style and unique qualities, from composition techniques to their actual voice, to an algorithm. This creates a “push button” world for complete novices who don’t even need to be inspired by the artist, to simply generate endless variations on this artist and cheapen them completely. You could effectively troll an artist with perversions of their own cultural contribution. Those Udio Delta Blues examples do that to the legacy of Muddy Waters. It’s disgusting. No one needs to have any skill or sustained motivation or true artistic inspiration to generate these perversions or just knock-off replicas. It’s much different than a human inspired to cultivate music skills, who then is inspired by a specific artist, and puts the time and energy into developing an influenced style and pays homage to the original artist (like the Beatles did to Buddy Holly etc). AIs learning from humans will make true artists stop creating.
So Landr (the distributor) now has an opt-in when you register and upload a title with them. You can opt-in to be used in their AI learning and be paid a royalty every time your music is learned from. You can also opt-out and not be included in their AI database but still use their services. I can definitely see this as a solution for the AI learning aspect.
I have a funny feeling of déjà vu, and we went through the 70s and 80s, with the fear of drum machines and synthesise music. also remember the "bedroom musician"
I really think we need to revamp IP legislation and regulation to ensure original HUMAN artist get paid for their work. How about encoding a unique block chain ID into the metadata of every copyrighted song that MUST be recorded by every AI training algorithm. Once the algorithm is trained, every UDIO-like corporation pays a license fee to the all of the artists the algorithm used for a piece of AI work, and every individual end user pays a one-time license fee for individual use and every corporation using that AI product pays a license fee per play/broadcast.
Dear LORD, that man Jesse Josefsson should be a lawyer! No offense to you of course, Krystle. His answers, opinions, and commenting is just so quick, precise, right to the point, and genius! I was so blown away, that I had to make a comment on this video. And I RARELY comment on videos! LOL
@ThatOrko I don’t believe you can copyright a song that contains AI generated content. Which also means for this new medium there is zero Fair Use to consider for brands using say an Udio track. ESPECIALLY when they generate AI music exclusively for their brand and bring it in-house. The turnaround cycle for film/tv music supervisor just improved 10x and they no longer need to worry about clearing samples or verifying splits etc.
Debate about the ethics of the training data always seem to leave out how the model works. They're statistical models. They're not sampling tracks, they're counting how many times music with specific tags do specific things with the melody, chord progression, structure, everything that makes a song qualify for that tag. Newspapers can print box scores because the events of the a baseball game are facts, a thing that happened. Sports leagues fought them over this multiple times because the newspapers are profiting off printing a sports section. That a thousand surf rock songs do the same thing a hundred times and a different thing fifty times and so on is a fact, a statistic. I have no idea how Udio and Suno are defending their training data, but unless the model works differently than they're presenting it, they're covered by plenty of precedent.
Ive been a fan of @syncmymusic for many years. Was recently wondering what his thoughts about all this is....and here I find this video! Thanks Miss Krystle for all the great information/content!
I’m not the lawyer here, but my understanding of the fourth factor of fair use is that it isn’t generally applied the way Jesse is describing. Typically it’s used to determine if a specific work is harmed commercially by a specific other work in a similar medium, not because the resulting service can produce a result that serves a similar functional purpose. Admittedly, AI is a fairly novel scenario, but I think the “replacement” argument is substantially weaker than what Jesse represents it as. Think about it this way-if something is roughly as good of a replacement for a work that has been created as one that has yet to be created, what does that say about the nature of it?
In the US legislative law stated fair use applies to humans only (look into the photo taken by a monkey case). This should mean AI content can not be copyrighted for starters, but there’s also AI law getting written up right now so who knows… certainly not me.
That's not correct. The fourth factor of fair use has to take into account the impact of AI on the current market of the stolen copyrighted work. If I have a song, then AI will be competing directly with my song on the licensing market. It's a fact. Moreover, the fourth factor has to consider also the *potential* market which is damaged. Labels can argue that tech companies take away their exclusive right to train an AI with their catalogue. These simple arguments will be *deadly* for AI companies.
@@federicoaschieri I assume the plaintiffs will make arguments to that effect (or something similar), but I don’t find them very compelling for the reason I outlined in my original post, as well as the de minimus factor of an individual work’s impact on the resulting model.
@@GoranBackmanMusic you’re confusing whether something can be copyrighted with the fair use defense, which are separate concepts. The selfie case refers to the first, not the second.
@@pokepress Fortunately, it's not you the one that has to find those arguments compelling. The reality is that the impact of AI on the market is simply indisputable, otherwise none of us would complain about AI. And as soon as there is the damage to the market of the stolen works, it is no more fair use. AI companies have just to license data if they want to compete with us in the market. It doesn't matter at all what the impact of a single song is on the model, it is the fact the song is used that violates law, because the AI model uses it to compete with you. Your argument is like saying that the companies can steal everybody $0.50 because it is a negligible impact on their financial situation 😂
I assume you’re referring to the fact that shutting that down didn’t really accomplish anything because there was still a consumer need that wasn’t met, resulting in continued piracy and 15+ years of declining revenue.
Exactly. They disrupted the music industry and created a new way to sell and market records. Like I said, adaptation is key. What that looks like right now, not sure. But AI ain’t going away.
@@MASSDECEPTIONTV Is there even a reason for people to want to be artists anymore? Image streaming service dedicated to AI music. You have to pay one person
Is there a reason to be an artist anymore? Image streaming service dedicated to AI music. A record label dedicated to AI music. You can keep all the money
🙏 welcome to the community. We go live every Wednesday at 5:00 p.m. PST Make sure that you join in on the conversations with your questions and comments 💪
Generative AI is a simple business model- just harvest the creative work of hundreds of thousands of Artists, Musicians and Writers (and even Coders!) then use that harvest to train an AI that competes directly with those creators. The AI Developers would have nothing to sell if they had not first taken the creative efforts of others without payment or permission to build their systems. No new value is being created here- value is simply being transferred from Artists to those who took their work and used it to build a machine that replaces those Artists. And no- this theft of other people's work cannot be justified by pretending that AI's are just silicon people who are entitled to learn from anything they want-they are machines, not people- and if I build a robot out of parts that you own without your permission then I have stolen value from you- even if that robot looks a bit like a person.
As a video creator, I’ve struggled for years finding the right music for certain scenes. The music I’ve had to use for years is from subscription based services. It’s always hit and miss. I may have to spend hours going through libraries to find something that may or may not fit. It’s might work but be the wrong length. I don’t have the budget for John Williams and the London symphony orchestra. AI could create exactly what I want, faster and better than any composer I could ever afford. I’m embracing the future. And I’m happy about it.
Do you currently use AI for the music in your videos, out of interest? Or are you not at that stage yet and talking only about the future potential. If you are, Doesn’t it take a while to get what you want? Cheers.
You shouldn't be allowed to copyright music that has ai generating it....At some point tech is going to increase to where AI can and will produce every possible music option out there. * You will then have one company owning that AI, thus owning all music, and keeping people from being allowed to make any money off their music. ** Essentially: Only one company will be allowed to make money off of music...whoever owns that AI which is creating 99% of the songs. Also depends on people actually listening to it and other factors. AI is only going to aid those who are the most deceptive first.....and that is terrible for our future everywhere. They can't create, so they want to prevent us from creating as well.
Ai can only mimic and consolidate pre-existing styles. It's a copy cat. It can't create or invent anything new. Both on the visual and music sides. Play with ai long enough and you'll realize humans are still safe as the driver of the creative bus
At first I thought maybe as it can't do Wrestling it can't do Production Music as it's practically the same style. But then today I completely manually crafted a lot of Indian Instrument solos together into a completely custom instrumental track and the voice on parts of it doing ah ah ah was 100% Lakha Music's voice and all the solo instruments sounded 100% like his playing. PRS and Cocatalog needs to bring in a TV licence thing for AI training they don't pay it they are opting out or breaking the law just as you don't pay BBC you won't get any legal TV in the UK you are breaking UK media law by watching any non BBC TV as well with no licence paid to the Beeb. Production Music also has to open a second licence through the PMA as the other one would be performing rights for every track in existence but they need extra framework as they also have the stems generation rights to settle and the fact that it operates on having to pay for the track again after the licence expires after the annual. When they have to opt in by paying millions every year or opt out by not but then not having a service anymore then it would be a fully legal practice with framework paid off on the annual. In general the AI services themselves are bringing an extra layer to generate your own stems for making a manual track which is a positive for real musicians using half AI DAWS but they are training to do this off of web crawls it needs legal framework the owners of these services must pay off as for the services I love playing in them though and never want this progression to leave us altogether as stems creating is useful for musicians but I also feel strongly there needs to be custom music licences in place for training as you can't watch tv without paying BBC right. AI is awesome I even enjoy making my own music in it as well but all the lawful framework should have been put in place for the companies before the revolution began to make it 100% lawful on the training part of things it's Western Law to blame for not bringing in the licences when they knew there was upstarts coming from the first signs that was these AI Stem Generator sites that full blown AI Music was going to explode as a scene shortly after. I admit I enjoy Udio and have used it to make styles of music we shouldn't be playing with instruments from other places in them. But I'm one for the ai training license as well don't pay it you're out of business.
Sync (as in soundtracks) requires higher quality than is currently possible with AI. Eleven Labs sounds a bit better in terms of fidelity but given how sloppy AI stem extraction is, getting a true ATMOS mix is still years away.
AI generated music is just a subset of the questions around the large language models. As Miss Krystle said, the AI model “heard”” a lot of songs just like a “person” could and comes up with its own “ideas”, perhaps with the assumption that the tracks it “listened” to was legitimately paid for. At the end of the day what happens with the LLMs will decide the fate of the music and image AI industries. The 800lb gorilla will be what happens when musicians themselves start using AI to figure out what will be popular and sell the best. What kind of compensation argument will be made for the AI assist when that happens and the monetary reward even bigger?
my underlying concern is if the 'smaller' types use more and more AI, where do people get their skills to move up? if AI scoops up the low hanging fruit, how does someone 'grind' at it to get higher up?
Big business always seems to dictate what happens in the world. An AI is definitely big business in ways for businesses to not only make money but save money and cut costs.
@@ThatOrko I think that a healthy concern about A.I. is quite understandable, however, I'd argue that the music sector was stagnant before A.I. showed up this time around. That the music invented beyond the 1900s rarely has been worthy so far.
@@WillyJunior OK, lazy dogu. If that comment made you feel inferior, that's on you. If you can name just one (1!) worthy genre invented in the 2000s, I'd be impressed. For example both Techno and videogame music deff peaked in the 90s. And Rock related stuff even before that. Not by any means that Rock is that great of a genre to begin with. Fusion Jazz has very nice energy though, which is a genre that peaked somewhere 80s to early 90s. Whilst Classical music probably could replace all genres altogether, and that would likely have a positive impact on most peoples lives in all aspects. At least I am not aware of any new Classical music of iconic greatness composed in the 2000s. Latino related Jazz is awesome too, and that probably peaked in the 60s. Such genres was not driven by pesky political agendas. Anyway, Synthwave ain't it. If you enjoy a spectrum of Satanic emotional content, then your comment makes all the more sense, of course. Whether you're consciously aware of it or not, astronomical economic powers (which also taps into the political powers) do manipulate what is most presented/available on the market. Those powers (your/our invisable owners) are not only corrupting your thoughts and ideas with agenda-driven lies via mainstream media, but they're also corrupting your emotional state via *music* and woke movies, etc. Yes, it may not appear that obvious to most people, since there's simulataniously that forest of creativity out there that muddles the view. Btw, I held this position ever since I was a teenager. *Hah!*
okay so for producers who do NOT want to run into issues: generate melodies, download & import them into your DAW. open serato sampler & let it find key & bpm (so you don't have to manually do time stretching) loop the part you like the most. open chrome: find something that can extract midi from audio. then use your fire kontakt libraries (or lets be real omnisphere) to replace each instrument in the loop. ...that way there is no encoding built into the audio & no way for you to get in trouble. side note but if you feel up for the challenge, try to remake the loops just using your ears (it includes a LOT of listening to the same loop, filtering & gets old fast btw) but it will make you OP AF if you get good at this.
Awesome, productive suggestions! Going to try that. Most of these commenters just want to whine and play victim, though. I doubt they even read your instructions.
If you had the choice of seeing a live human band or going to an ai music show , whats more attractive, Dj included, people will always want the human connection so i think that in the wave of everything generated people will still support people.
This entirely depends on the ai show. Are we talking about a cool futuristic ai experience show, with robo holograms performing on a stage? The crowd would then be the human interaction factor, rather than the band. Programmed automatons have shows at Disney, people enjoy everyday. It could also be an ai VR experience... If one thinks creatively, a total bot music show being enjoyable is not out of the realm of possibility. It's simply a different experience. Neither good nor bad. Whatever your personal existential thoughts may be on the subject, people will and do, enjoy both kinds to varying degrees.
@@joanie-music Who the hell would want to go to a festival to put a VR headset on Lmfao might as well sit at home on ur PC. I think you are incorrect there. If anything it would be more like going to watch a fireworks festival. Plus the energy to run something like that for days on end... yeah i cant see it .
@@TheKanguru It's clear you would not. That's cool. FYI there are already VR concerts happening, and they are selling out. Maybe check out Web3 and what's happening in that space, if your curiosity leads you. "Different strokes for different folks", as they say. There is always a market for anything well done.
@@TheKanguru Actually, lots of people. Guess what? It's already happening. Clearly, it's not your jam. LOL I understand you 'can't see it'. I'm not here to proselytize for or against. I'm strictly stating facts. If you're curious, research VR music concerts in Web3. And ai concerts with holograms is an experience already being sold... and getting sold out. Success in Asian markets is encouraging such concerts to be replicated in the UK and US. Maybe read up on what's technologically currently happening globally, in the music concert space. "Different strokes for different folks," as is said. "There is always a market for anything well done" - John Braheny
Do you think that the car producers would accept, a ai company to scan all blueprints whithout approval ,of all the the car brands, and then start to design new models of cars ,using pieces of each car of the brands, and then call it theirs, The problem is that today a person dont own their own data, thats why big tec can harvest it. When the politicians make the law, that the data of a person ,is own by the person, and cant be harvested whitout your approval, then humans are safe and free. So why is the politicians silent and why has the law not been made. Its been 45 years of internet but the law were is it.
You‘re both producers. Both of you hear music and learn from it, both musically and technically. You get inspired by it. Then you do own tracks with that knowledge and inspiration, with the intention to moneytize it. Do you pay additional money for retrieving that knowledge and inspiration from existing music, or do you only pay for the opportunity to listen to it? Do you sign special deals to be able to use certain production techniques you learned by listing to existing music? You know the answer. Why should it be different for an AI? And furthermore, generally speaking about AIs learning from content, not only music: Do you really want the big companies to decide what an AI should learn and what not? I sure as heck don‘t want that. AIs should have as much information as possible to serve US instead of serving the interests of big companies.
Udio has been good... but ElevenLabs is GREAT! They put out some demos of different genre's yesterday. Holy moly... as an artist, i love this. New tools for all my mediums exploding at the same time. such an exciting time to be alive. naysayers and dinosaurs will complain, happens with every shift, then things get regulated and ppl pipe down. I for one, love the new tools intensely.
You're right. And Ai will quickly separate the music artists from the music craftists. ai music is craft based music. Art is the realm of expression, invention & experimentation - humanity. ai can't invent. Only mimic. A lot of music craftists unfortunately think they're artists. Though the two models can overlap to varying degrees, depending on the goal. Strict music craftists though, will have to learn how to be artists, if they want to stay ahead of, or stand out against ai music.
If you notice this, then that means your music most likely conforms to a standard sound. As in... sounds like everything else out there. For better or worse. That uniformity can serve many useful external functions. Like fit a genre, make people dance, get quick follows... etc. It's craft based music. It serves a particular type of goal. Craft is also the only thing machines can do. ai is only good at mimicry. It can't create new fresh ideas or invent. Invention is the realm of artists. Invention is humanity. Push yourself to create personally metaphoric music that truly expresses your ideas, feelings & stories without regard to genre or style, or fitting into some preconceived aesthetic or genre. I understand a lot of us work to write to a genre because we like it, or to land a job, even in order to be popular or sound familiar somehow to make it easier for others to accept us - 'our music'. I challenge you to express sound in a way that is Intrinsically you, because ai can't do that. And that is what the world needs right now. Your unique sound, you clearly have not created yet. Or at least, not put out into the public. It's scary to go down that unique road because many people may not dig it. It won't be industry standard. Guess what though? ai won't be able to make it... until it's fed your stuff... then if you're continuing to grow and push boundaries with raw childlike abandoned sonic expression, without clinging to music history formats... ai will be playing catch up to us. Not the other way around.
So Landr (the distributor) now has an opt-in when you register and upload a title with them. You can opt-in to be used in their AI learning and be paid a royalty every time your music is learned from. You can also opt-out and not be included in their AI database but still use their services. I can see that as being a solution for the learning aspect of this equation.
Here is what I’m having trouble with when I hear people say these models are stealing copyrighted music. If I, a human study someone’s catalog and then creates music in that style that is totally fine and legal. Why then is it wrong if I, a human trains my robot on someone’s catalog and then have it create music in that style?
Legally? The same reason the court ruled that no one owns the copyrights to the selfie taken by a monkey. Copyright is awarded to human creations only. Morally? Because it disrupts the process of iterating on music over time that musicians all over the world are engaged in and that as listeners we all value immensely. You have to ask, why do we listen to new music? Because we want to hear new perspectives on existing material. An AI can provide that in the short term but not in the long term since it has a fixed dataset of all the music that exists currently. If artists are impacted financially by this new technology and are thus unable to continue making music as they were before, then we loose that process of human interaction and the fountain of new and interesting perspectives eventually dries up.
that because the robot has to make copy of the that style. it means it has to break copy right to make copy for it's memory. copy allow copies in human brain but not copies by machine. in fact it has to have a lot unique songs in memory to produce good results. In more the better. with midjourney train on 16,000 artist. which means it could up 16,000 copy righted works in its memory.
because humans have ideas which are turned into expressions, then fixed to a medium. Humans extract an idea then reproduce a new expression. These models have no ideas - they measure averages in a body of expressions and produce new expressions. Furthermore, it's unquestionable that these companies have engaged in infringement - they admit to it freely. They hope to argue that their infringement is fair use, which doesn't hold water since their models create direct market substitutes for the material it trained on.
@@spark300c no, that's not how generative ai works. There are no copies of any copyright protected works kept within the ai models once training is complete. No generative ai has access to or memory of any of the data it was trained on. Your opinion is based on ignorance and misunderstanding. Midjourney is in the middle of a big class-action lawsuit against it right now. Early on in the case, it was quickly understood that the plaintiff's complaint of copyright violation - which included a bit about copies being made, as you describe here - had "no legal basis" to be taken seriously by the court. They're basically getting laughed out of court. That's what happens when you take action based on your feelings, instead of being based on the facts..
@@SineEyed the problem it produce copyright works. the music generation uido has output copy righted songs. so how image generator out put picture of batman if does not have picture of batman in memory. mean that generator AI can not produce certain images it not know what they look like. So far call it super hero test. we turely do not know want in these iA model because they have new release it to public. I also read some AI companies know they in wrong so have ia cover up it tracks because creating derivatives to be stored as training data instead of real works.
I'm happy if we can kick the music sharks at the top with AI. There are only a few who earn their way up there stupidly and stupidly. Music has been commercialized to death in the last 20 years. The music business is a hard and cold mafia-like construct. The smaller studios and production companies should not feel addressed here. The radio music that the AI generates is competition for the very top and comes into the homes of the big guys like a Swat team. Let The Games Begin....
To me the whole damn Thang is wicked. For some if not most, Music is their salvation from destitution and struggle. A.I. will turn it all into The Capital Wasteland.
I love AI music generation tools. Im here for it. SO fun. If training it on copyrighted music music is infringement when money becomes involved, then, by that logic... so is the influence human music creators/artists claim when they make money off of it as well. I love using AI to help in music production which I've been doing for over 2 decades now.
@@ajat3202 it accelerates the production time substantially. Its just not humanly possible to produce something of the quality offered with the aid of AI on my schedule and budget.
@@ajat3202 it accelerates production time. It's just not humanly possible to output at the pace AI has enabled. It's just a nifty tool added to my production suite. I think a lot of people are assuming it's gonna do all the work... Well, I like to add my personal touch including my own vocals/guitar work/keys/percussion, etc...
We're going to see how artist ends up using it in the future. I see the younger generations taking more advantage of it than those that have been in the industry longer
Gonna somewhat disagree with Jesse in terms of licensing and AI training. AI training is just a digital version of a human artist listening to a track and learning to play covers and turning that into experience. Does the human artist have to pay a licensing to learn a song or learn to play in someone's style? Not really. At most they might have a Spotify account or perhaps bought some tabs or sheet music. I mean you got talented artists like Anthony Vincent over at 10 second songs. Do you think he paid a license for all of the impressions he does? Doubt it. From a philosophical point of view here, why should it require a license to learn digitally versus a human learning simply because some perceive it as an existential threat? I mean Krystle does pose the same question to him. But I think the point is moot. For example - cover bands. They trained and learned a bunch of music. They get hired for a bar or some other gig. They're making money for that gig - a profit. And it's NOT at all transformative. They aren't paying licenses either. Or what about all the cover bands on UA-cam. They might not be making ad money from UA-cam but I can guarantee you a lot of them are making money from Patreon from fans. That's profit.
Sync has been broken since the 90s. Services like Taxi and Broadjam are absolute scams. There are 1000s of artists with dozens or 100s of tracks each for every singular need for a placement. The pay is abysmal unless you are the go to person for a studio or famous enough your name brings in viewers. That’s why so many famous chart tracks are show horned into shows, movies, and commercials. Blaming Ai sounds like a terrible case of survivor bias from someone mad their gravy train is heading out. I am singed non exclusively to three Nashville film sync libraries and have had several Amazon and Netflix placements. So no, I’m not bitter. I just understand the actual reality of what’s happening.
it also lets you straight up type the instruments you want to hear. lol this is so fun. also you can type shit like "reverb" and recordings to get more "record/sample" vibes. you can just type "sampling" too.
That red lobster ad was made by some tone deaf marketing executive. It literally sounds like hold music when you call the water company or something. Red lobster will go bankrupt if they're down to that.
I think the best thing about AI music is how it will help real artists produce albums far quicker. Imagine not having to wait another ten years for a Tool album? I am sure Maynard, Adam and Dan wouldn't want to use AI to help them make Tool albums; I am simply using this as an example.
i see the video that he made of udio and i think that he is a crying looser , he needs to go to other place as photography , art , he needs to get out of the box is like a crying baby
Follow Jesse @SyncMyMusic : www.youtube.com/@SyncMyMusic
🟡Ready to become your own record label? Get started now!:
www.topmusicattorney.com/becomeyourownrecordlabel
📺 Get Your Music Into Tv/Film/Games: www.topmusicattorney.com/sync
💻 #1 Resource For Artists And Producers: www.topmusicattorney.com
⚖ Hire An Attorney: www.delgadoentertainmentlaw.com
📰 Get My Newsletter: www.topmusicattorney.com/newsletter/
📄 8 Essential Music Contracts Bundle: www.topmusicattorney.com/contractsbundle
🎧 Listen To The TMA Podcast On All Streaming: www.topmusicattorney.com/podcast
🟢 Free Stuff: www.topmusicattorney.com/free-stuff
I havent watched this video yet but Udio has competition, and it’s more concerning at this point:“ElevenLabs”…
In the 1970s, welders taught machines how to weld ships. Then there were welders' strikes in shipyards. They were afraid for their existence. It was the same with horse cart owners when the car was created. It's the same with AI. In the end, it is the recipient who decides which product he will use, not his grandchildren. If 1,000 people steal something, they go to prison. if 10 million do it. This becomes law.
There's something to this for sure
Nothing is being stolen..
Nothing is original. Everything is stolen to some extent. , "If I have seen further than others, it is by standing upon the shoulders of giants," Newton
@@SineEyed the algorithm is trained on existing music.
@@SekundSun1986 So are you.
Music basically is copying.
The genres are for a reason.
There are also very few options that makes sense musically or mathematically in order to make modern music.
If there is a certain harmonies or chord-progressions you only have so many options to get a feeling across.
There is a reason why the perfect fifth sound so good. (Although in our equal temperament we don't have perfect fifths since all "errors" are smeared out).
People have the misconception that they own the songs.
They do not. They can have a copyright on their *expression* of ideas.
But not the underlying ideas themselves.
Taking a body of work and analyzing it for common methods of technical nature isn't taking the expression away.
Even if it offsets musicians it does not offset one single work in any appreciable manner.
To quote an AI:
"Summary
AI Training on Music: Using music to train AI by analyzing aggregate features (not the specific expressions) is generally considered less invasive and more abstract than human listening, though it involves complex legal considerations.
Ownership Misconception: People often misunderstand what they own in a song. Copyright protects specific expressions, not the general musical ideas or common elements.
Legal Distinction: The distinction between owning a song and owning the rights to its specific expression is crucial and informs how copyright applies to both human and AI-generated music.
In conclusion, while copyright law can be complex and nuanced, the core principle remains that it protects the specific, original expression of creative works, not the underlying ideas or common elements. This distinction is vital for understanding the legal landscape surrounding AI training on music and the broader implications for music ownership and creativity."
1:21 just like to point out that while Udio is fast, it literally doesn't "spit out full tracks in 30 seconds". It generates 2 30 seconds of a possible track, and then you have to extend it in 30 second chunks at a time to make it a full track,
The paid version of Udo was just released and depending on what tier you get you can generate more songs than that and the update has way more control of how you can edit your songs. It's going to just keep improving.
Yep. I'd say it actually takes someone with a good ear for music and a talent for writing decent lyrics, to get an output from udio that would give anyone in the music industry good cause to feel "uneasy." And as you point out, that sure as hell ain't happening in 30 seconds..
There's a new AI generator by ElevenLabs, sounds incredibly good with longer songs.
ua-cam.com/video/SdIi2c6hMDA/v-deo.html
Finally someone actually getting at the core of this issue! Thank you Jesse for bringing up the issue of the training data.
We are very fortunate to have him on the channel. Such Great insight.
I think the problem here, and what actually makes people afraid of AI (or at least should make them afraid), is not the technology itself but rather that no government is willing to regulate the market in such way that musicians and their jobs are protected. The paradeigm as we expirience it is that governments -obey corporate interests- lobby with corporates, and not human beings.
I know some members of Congress are trying to get that fair wage for musicians act past that has to deal with streaming we'll see if somebody can come together for some AI regulation. The film industry was able to make some moves in regards to AI. The problem is musicians don't unionize in the same way. We'll see if that changes
Do you really believe "the government" is around to help anybody, much less the little guy? Grow up.
I'm more worried about the 0 regulation in "defence" technology, tanks jets you name it its integrating. the pollution that would come from a world battle bot arena is a worry
@@NoDaysOff-oz2zl Sorry friend... YOU grow up... This is why you vote.... Vote the right guys.
@@TheKanguru also true!
In my opinion, the best AI music will be generated by actual musicians. Knowing the lingo is essential for prompting, having a grasp on genres, styles, and moods is equally essential. With Udio, you are regenerating every 30 seconds, and determining which extensions sound better will be optimally achieved by a trained ear, not the average joe. There will always be a job for those already in the industry, you just need to adapt. Who is going to make the AI music for the companies? Not the company themselves, and while anyone can make AI music, musicians will still be the best at it. Musicians will need to create and CURATE the best results. They already have the industry connections and network, and can work to a brief. An average joe with Udio does not have those same necessary qualifications to just take over your job. Adapt, embrace, evolve.
Yup. Adaptation is key.
Bro just tarin chatGPT or Lama or Claude 3 to prompt for Udio.... You got a point but unfortunately other AI can be trained to be a musiccian and know how to prompt, lot of people already do this to prompt on AI music generators... So maybe there will always be a human to guide, but no need for him to be musician.. All is going the wrong way for humans unfortunately.
Same with software engineering.
Experienced seniors become true 10x with AI assistance over juniors.
Clients can prompt all day long but still need someone with skills to put the requirements together.
It's a process working with AI to build higher quality products.
Finally the discussion on fair use makes sense. Thanks Jesse, you're perfectly right, this is not fair use. And thanks Krystle for inviting him.
Thanks for having me on! It was great to chat about these important topics 🙏
🙏 Thank you so much for being on the show. You're insight it's incredible. I love how candidly and honest you speak about the situation. I look forward to our channels collaborating again in the future 🖖
The only people that care at all about most music used as underscore on television are the composers and music libraries that provide the music for the shows. I talk to non-composers all the time about my participation in this industry and almost every single person I've talked to about it said something like "I never really noticed" or "I had no idea there was so much music used in a single episode." We are a niche population, 99% of the world gives zero s***s about what we do, how it sounds, or how it was created in the first place. ;)
Sync licensing sucks mam
I think the AI training on copyrighted music is likely infringement. The ML model makes a copy during the training.
We're going to be keeping a close eye on how the laws may change over the next few years
It's not infringement. Especially not for making a "copy". A copy is made any time you stream literally anything--that's just how the internet works..
@@SineEyed yes, that is exactly how streaming works. However, a license is granted by the copyright holder to, spotify for example, so that they can stream. Here the copyright holder is not granting the ML model a license. Hence likely infringement. (See, e.g., American Broadcasting Companies, Inc. v. Aereo, Inc.,) You can make an argument for fair use, like Open AI did in their suit with NY Times. But its a poor argument for music, because in the case mentioned above, it is the facts of the articles that are being reproduced.
8:00 I disagree. It’s not just about money. It’s about human culture. Humans learning from other humans is much different than AIs learning from humans. AIs reduce an artist’s style and unique qualities, from composition techniques to their actual voice, to an algorithm. This creates a “push button” world for complete novices who don’t even need to be inspired by the artist, to simply generate endless variations on this artist and cheapen them completely. You could effectively troll an artist with perversions of their own cultural contribution. Those Udio Delta Blues examples do that to the legacy of Muddy Waters. It’s disgusting. No one needs to have any skill or sustained motivation or true artistic inspiration to generate these perversions or just knock-off replicas. It’s much different than a human inspired to cultivate music skills, who then is inspired by a specific artist, and puts the time and energy into developing an influenced style and pays homage to the original artist (like the Beatles did to Buddy Holly etc). AIs learning from humans will make true artists stop creating.
Thank you for having him on the show 🙌🏻
if a niche library had specialty training data music, the artists used to train that model could be paid royalties on syncs w/o making new music
I just wish there was more of an opt-in process that was happening before these artists were being trained on.
So Landr (the distributor) now has an opt-in when you register and upload a title with them. You can opt-in to be used in their AI learning and be paid a royalty every time your music is learned from. You can also opt-out and not be included in their AI database but still use their services. I can definitely see this as a solution for the AI learning aspect.
Top Music Attorney is amazing for all things music related 🎶 the business and for the artists!
I’m one of those who jumped on the AI music bandwagon. Great video giving a fair comparison and warnings for others like me.
Jesse is always a thoughtful speaker and straight shooter
I have a funny feeling of déjà vu, and we went through the 70s and 80s, with the fear of drum machines and synthesise music. also remember the "bedroom musician"
Great Show appreciate Everything you are doing , Great intelligent interview thank you to everyone involved.
Yaasss 💪🏻 I absolutely love this interview ❤️
I really think we need to revamp IP legislation and regulation to ensure original HUMAN artist get paid for their work. How about encoding a unique block chain ID into the metadata of every copyrighted song that MUST be recorded by every AI training algorithm. Once the algorithm is trained, every UDIO-like corporation pays a license fee to the all of the artists the algorithm used for a piece of AI work, and every individual end user pays a one-time license fee for individual use and every corporation using that AI product pays a license fee per play/broadcast.
Dear LORD, that man Jesse Josefsson should be a lawyer! No offense to you of course, Krystle.
His answers, opinions, and commenting is just so quick, precise, right to the point, and genius!
I was so blown away, that I had to make a comment on this video. And I RARELY comment on videos! LOL
@ThatOrko I don’t believe you can copyright a song that contains AI generated content. Which also means for this new medium there is zero Fair Use to consider for brands using say an Udio track. ESPECIALLY when they generate AI music exclusively for their brand and bring it in-house. The turnaround cycle for film/tv music supervisor just improved 10x and they no longer need to worry about clearing samples or verifying splits etc.
Debate about the ethics of the training data always seem to leave out how the model works. They're statistical models. They're not sampling tracks, they're counting how many times music with specific tags do specific things with the melody, chord progression, structure, everything that makes a song qualify for that tag. Newspapers can print box scores because the events of the a baseball game are facts, a thing that happened. Sports leagues fought them over this multiple times because the newspapers are profiting off printing a sports section. That a thousand surf rock songs do the same thing a hundred times and a different thing fifty times and so on is a fact, a statistic. I have no idea how Udio and Suno are defending their training data, but unless the model works differently than they're presenting it, they're covered by plenty of precedent.
Ive been a fan of @syncmymusic for many years. Was recently wondering what his thoughts about all this is....and here I find this video! Thanks Miss Krystle for all the great information/content!
I’ve been waiting for this interview 👌🏾
This is something I can agree yo Jesse, my situation is very similar. And I don’t agree with AI doing a copy track without permission from the artist.
Great advice that is why I stick to playing and no sequencing even on my reasons software
I’m not the lawyer here, but my understanding of the fourth factor of fair use is that it isn’t generally applied the way Jesse is describing. Typically it’s used to determine if a specific work is harmed commercially by a specific other work in a similar medium, not because the resulting service can produce a result that serves a similar functional purpose. Admittedly, AI is a fairly novel scenario, but I think the “replacement” argument is substantially weaker than what Jesse represents it as.
Think about it this way-if something is roughly as good of a replacement for a work that has been created as one that has yet to be created, what does that say about the nature of it?
In the US legislative law stated fair use applies to humans only (look into the photo taken by a monkey case). This should mean AI content can not be copyrighted for starters, but there’s also AI law getting written up right now so who knows… certainly not me.
That's not correct. The fourth factor of fair use has to take into account the impact of AI on the current market of the stolen copyrighted work. If I have a song, then AI will be competing directly with my song on the licensing market. It's a fact. Moreover, the fourth factor has to consider also the *potential* market which is damaged. Labels can argue that tech companies take away their exclusive right to train an AI with their catalogue. These simple arguments will be *deadly* for AI companies.
@@federicoaschieri I assume the plaintiffs will make arguments to that effect (or something similar), but I don’t find them very compelling for the reason I outlined in my original post, as well as the de minimus factor of an individual work’s impact on the resulting model.
@@GoranBackmanMusic you’re confusing whether something can be copyrighted with the fair use defense, which are separate concepts. The selfie case refers to the first, not the second.
@@pokepress Fortunately, it's not you the one that has to find those arguments compelling. The reality is that the impact of AI on the market is simply indisputable, otherwise none of us would complain about AI. And as soon as there is the damage to the market of the stolen works, it is no more fair use. AI companies have just to license data if they want to compete with us in the market.
It doesn't matter at all what the impact of a single song is on the model, it is the fact the song is used that violates law, because the AI model uses it to compete with you. Your argument is like saying that the companies can steal everybody $0.50 because it is a negligible impact on their financial situation 😂
Imho, five years is a exceptionally optimistic forecast for background music. I'd say 2 years at best.
He mentioned Napster- just look at how that turned out. Adaptation is key.
I assume you’re referring to the fact that shutting that down didn’t really accomplish anything because there was still a consumer need that wasn’t met, resulting in continued piracy and 15+ years of declining revenue.
Exactly. They disrupted the music industry and created a new way to sell and market records. Like I said, adaptation is key. What that looks like right now, not sure. But AI ain’t going away.
@@MASSDECEPTIONTV Is there even a reason for people to want to be artists anymore? Image streaming service dedicated to AI music. You have to pay one person
Is there a reason to be an artist anymore? Image streaming service dedicated to AI music. A record label dedicated to AI music. You can keep all the money
Labels already keep majority of the monies 😂
Ok, I Subbed. Wow, what a channel.
🙏 welcome to the community. We go live every Wednesday at 5:00 p.m. PST Make sure that you join in on the conversations with your questions and comments 💪
Great points!
00:04:20. I believe that sample libraries companies like Spitfire Audio pay royalties to the musicians they sample
Generative AI is a simple business model- just harvest the creative work of hundreds of thousands of Artists, Musicians and Writers (and even Coders!) then use that harvest to train an AI that competes directly with those creators. The AI Developers would have nothing to sell if they had not first taken the creative efforts of others without payment or permission to build their systems.
No new value is being created here- value is simply being transferred from Artists to those who took their work and used it to build a machine that replaces those Artists.
And no- this theft of other people's work cannot be justified by pretending that AI's are just silicon people who are entitled to learn from anything they want-they are machines, not people- and if I build a robot out of parts that you own without your permission then I have stolen value from you- even if that robot looks a bit like a person.
As a video creator, I’ve struggled for years finding the right music for certain scenes. The music I’ve had to use for years is from subscription based services. It’s always hit and miss. I may have to spend hours going through libraries to find something that may or may not fit. It’s might work but be the wrong length. I don’t have the budget for John Williams and the London symphony orchestra. AI could create exactly what I want, faster and better than any composer I could ever afford. I’m embracing the future. And I’m happy about it.
I’m in a similar spot. I usually end up using credits on a couple of songs, then try and pick the best one.
Do you currently use AI for the music in your videos, out of interest? Or are you not at that stage yet and talking only about the future potential. If you are, Doesn’t it take a while to get what you want? Cheers.
You shouldn't be allowed to copyright music that has ai generating it....At some point tech is going to increase to where AI can and will produce every possible music option out there.
* You will then have one company owning that AI, thus owning all music, and keeping people from being allowed to make any money off their music.
** Essentially: Only one company will be allowed to make money off of music...whoever owns that AI which is creating 99% of the songs.
Also depends on people actually listening to it and other factors.
AI is only going to aid those who are the most deceptive first.....and that is terrible for our future everywhere.
They can't create, so they want to prevent us from creating as well.
Ai can only mimic and consolidate pre-existing styles. It's a copy cat. It can't create or invent anything new. Both on the visual and music sides. Play with ai long enough and you'll realize humans are still safe as the driver of the creative bus
At first I thought maybe as it can't do Wrestling it can't do Production Music as it's practically the same style. But then today I completely manually crafted a lot of Indian Instrument solos together into a completely custom instrumental track and the voice on parts of it doing ah ah ah was 100% Lakha Music's voice and all the solo instruments sounded 100% like his playing.
PRS and Cocatalog needs to bring in a TV licence thing for AI training they don't pay it they are opting out or breaking the law just as you don't pay BBC you won't get any legal TV in the UK you are breaking UK media law by watching any non BBC TV as well with no licence paid to the Beeb.
Production Music also has to open a second licence through the PMA as the other one would be performing rights for every track in existence but they need extra framework as they also have the stems generation rights to settle and the fact that it operates on having to pay for the track again after the licence expires after the annual.
When they have to opt in by paying millions every year or opt out by not but then not having a service anymore then it would be a fully legal practice with framework paid off on the annual.
In general the AI services themselves are bringing an extra layer to generate your own stems for making a manual track which is a positive for real musicians using half AI DAWS but they are training to do this off of web crawls it needs legal framework the owners of these services must pay off as for the services I love playing in them though and never want this progression to leave us altogether as stems creating is useful for musicians but I also feel strongly there needs to be custom music licences in place for training as you can't watch tv without paying BBC right.
AI is awesome I even enjoy making my own music in it as well but all the lawful framework should have been put in place for the companies before the revolution began to make it 100% lawful on the training part of things it's Western Law to blame for not bringing in the licences when they knew there was upstarts coming from the first signs that was these AI Stem Generator sites that full blown AI Music was going to explode as a scene shortly after.
I admit I enjoy Udio and have used it to make styles of music we shouldn't be playing with instruments from other places in them. But I'm one for the ai training license as well don't pay it you're out of business.
Sync (as in soundtracks) requires higher quality than is currently possible with AI. Eleven Labs sounds a bit better in terms of fidelity but given how sloppy AI stem extraction is, getting a true ATMOS mix is still years away.
AI generated music is just a subset of the questions around the large language models. As Miss Krystle said, the AI model “heard”” a lot of songs just like a “person” could and comes up with its own “ideas”, perhaps with the assumption that the tracks it “listened” to was legitimately paid for. At the end of the day what happens with the LLMs will decide the fate of the music and image AI industries. The 800lb gorilla will be what happens when musicians themselves start using AI to figure out what will be popular and sell the best. What kind of compensation argument will be made for the AI assist when that happens and the monetary reward even bigger?
my underlying concern is if the 'smaller' types use more and more AI, where do people get their skills to move up? if AI scoops up the low hanging fruit, how does someone 'grind' at it to get higher up?
Your brain doesn't have to be creative anymore
AI music will definitely thin the sync music creator herd.
They kill our passion, identity and hopes.
Big business always seems to dictate what happens in the world. An AI is definitely big business in ways for businesses to not only make money but save money and cut costs.
It's going to impact almost every industry eventually, just give it time.
Relax, people/consumers can still only listen to one track at a time at 1x speed
The world always has a knee jerk reaction to change I'm trying to have less of that. AI is here we'll see what happens
@@ThatOrko I think that a healthy concern about A.I. is quite understandable, however, I'd argue that the music sector was stagnant before A.I. showed up this time around. That the music invented beyond the 1900s rarely has been worthy so far.
@@wilfred309 You have got to be aware of how "old man yells at clouds" this comment is
@@WillyJunior OK, lazy dogu. If that comment made you feel inferior, that's on you. If you can name just one (1!) worthy genre invented in the 2000s, I'd be impressed. For example both Techno and videogame music deff peaked in the 90s. And Rock related stuff even before that. Not by any means that Rock is that great of a genre to begin with. Fusion Jazz has very nice energy though, which is a genre that peaked somewhere 80s to early 90s. Whilst Classical music probably could replace all genres altogether, and that would likely have a positive impact on most peoples lives in all aspects. At least I am not aware of any new Classical music of iconic greatness composed in the 2000s. Latino related Jazz is awesome too, and that probably peaked in the 60s. Such genres was not driven by pesky political agendas. Anyway, Synthwave ain't it. If you enjoy a spectrum of Satanic emotional content, then your comment makes all the more sense, of course. Whether you're consciously aware of it or not, astronomical economic powers (which also taps into the political powers) do manipulate what is most presented/available on the market. Those powers (your/our invisable owners) are not only corrupting your thoughts and ideas with agenda-driven lies via mainstream media, but they're also corrupting your emotional state via *music* and woke movies, etc. Yes, it may not appear that obvious to most people, since there's simulataniously that forest of creativity out there that muddles the view. Btw, I held this position ever since I was a teenager. *Hah!*
@@wilfred309 Your tastes do not define the legitimacy of art. There's too much to address in your comment. I don't have the time.
okay so for producers who do NOT want to run into issues:
generate melodies, download & import them into your DAW.
open serato sampler & let it find key & bpm (so you don't have to manually do time stretching)
loop the part you like the most.
open chrome: find something that can extract midi from audio.
then use your fire kontakt libraries (or lets be real omnisphere) to replace each instrument in the loop.
...that way there is no encoding built into the audio & no way for you to get in trouble.
side note but if you feel up for the challenge, try to remake the loops just using your ears (it includes a LOT of listening to the same loop, filtering & gets old fast btw) but it will make you OP AF if you get good at this.
Awesome, productive suggestions! Going to try that. Most of these commenters just want to whine and play victim, though. I doubt they even read your instructions.
The tractor put farmers & farm hands out of business. "Was it right?" I don't know. Did it require regulation? Youbetcha.
Do Record Companies own Musical Styles and Genres? No... This is the tight argument for AI companies.
If you had the choice of seeing a live human band or going to an ai music show , whats more attractive, Dj included, people will always want the human connection so i think that in the wave of everything generated people will still support people.
This entirely depends on the ai show. Are we talking about a cool futuristic ai experience show, with robo holograms performing on a stage? The crowd would then be the human interaction factor, rather than the band. Programmed automatons have shows at Disney, people enjoy everyday. It could also be an ai VR experience... If one thinks creatively, a total bot music show being enjoyable is not out of the realm of possibility. It's simply a different experience. Neither good nor bad. Whatever your personal existential thoughts may be on the subject, people will and do, enjoy both kinds to varying degrees.
@@joanie-music Who the hell would want to go to a festival to put a VR headset on Lmfao might as well sit at home on ur PC. I think you are incorrect there. If anything it would be more like going to watch a fireworks festival. Plus the energy to run something like that for days on end... yeah i cant see it .
@@TheKanguru It's clear you would not. That's cool. FYI there are already VR concerts happening, and they are selling out. Maybe check out Web3 and what's happening in that space, if your curiosity leads you. "Different strokes for different folks", as they say. There is always a market for anything well done.
@@TheKanguru Actually, lots of people. Guess what? It's already happening. Clearly, it's not your jam. LOL I understand you 'can't see it'. I'm not here to proselytize for or against. I'm strictly stating facts. If you're curious, research VR music concerts in Web3. And ai concerts with holograms is an experience already being sold... and getting sold out. Success in Asian markets is encouraging such concerts to be replicated in the UK and US. Maybe read up on what's technologically currently happening globally, in the music concert space. "Different strokes for different folks," as is said. "There is always a market for anything well done" - John Braheny
Exactly. Generic music will be disrupted for sure. But good music will be good music and it takes good musicians to create it, with or without ai
Do you think that the car producers would accept, a ai company to scan all blueprints whithout approval ,of all the the car brands, and then start to design new models of cars ,using pieces of each car of the brands, and then call it theirs, The problem is that today a person dont own their own data, thats why big tec can harvest it. When the politicians make the law, that the data of a person ,is own by the person, and cant be harvested whitout your approval, then humans are safe and free. So why is the politicians silent and why has the law not been made. Its been 45 years of internet but the law were is it.
You‘re both producers. Both of you hear music and learn from it, both musically and technically. You get inspired by it. Then you do own tracks with that knowledge and inspiration, with the intention to moneytize it. Do you pay additional money for retrieving that knowledge and inspiration from existing music, or do you only pay for the opportunity to listen to it? Do you sign special deals to be able to use certain production techniques you learned by listing to existing music? You know the answer. Why should it be different for an AI? And furthermore, generally speaking about AIs learning from content, not only music: Do you really want the big companies to decide what an AI should learn and what not? I sure as heck don‘t want that. AIs should have as much information as possible to serve US instead of serving the interests of big companies.
Udio has been good... but ElevenLabs is GREAT! They put out some demos of different genre's yesterday. Holy moly... as an artist, i love this. New tools for all my mediums exploding at the same time. such an exciting time to be alive. naysayers and dinosaurs will complain, happens with every shift, then things get regulated and ppl pipe down. I for one, love the new tools intensely.
will Elevenlabs destroy Udio?
AI is dangerous and will not only be the end of music and the arts but humanity itself.
Change is definitely coming at a rapid pace.
It will die off quickly. Kind of like the day many years ago when I saw a VR headset in Goodwill.
@@NoDaysOff-oz2zl You're either in denial or you simply don't understand what you're saying.
@@MagnesiumEnterprise Haha I heard this with NFTS, Crypto, Metaverse blah blah, AI is overblown by businesses to make profit from fools like you
Tyler Perry halts $800 Million studio expansion due to AI concerns. Just think about how many hundreds of jobs have been lost because of this.
6:02 it’s too late to fight. The data is now out there already, the horse has bolted.
If your trying to get music placed in Sync now I think you should go work at Walmart or Subway
This is a common argument.
You can't adapt to AI and you shouldn't regardless how good it sounds, there's no feeling in it it will just play straight notes and no add libs!
You're right. And Ai will quickly separate the music artists from the music craftists. ai music is craft based music. Art is the realm of expression, invention & experimentation - humanity. ai can't invent. Only mimic. A lot of music craftists unfortunately think they're artists. Though the two models can overlap to varying degrees, depending on the goal. Strict music craftists though, will have to learn how to be artists, if they want to stay ahead of, or stand out against ai music.
Gamed over
it spilt out in 30 seconds what take me a weak. the most unethical thing is that AI has make copies of copyright work in memory to generate music.
If you notice this, then that means your music most likely conforms to a standard sound. As in... sounds like everything else out there. For better or worse. That uniformity can serve many useful external functions. Like fit a genre, make people dance, get quick follows... etc. It's craft based music. It serves a particular type of goal. Craft is also the only thing machines can do. ai is only good at mimicry. It can't create new fresh ideas or invent. Invention is the realm of artists. Invention is humanity. Push yourself to create personally metaphoric music that truly expresses your ideas, feelings & stories without regard to genre or style, or fitting into some preconceived aesthetic or genre. I understand a lot of us work to write to a genre because we like it, or to land a job, even in order to be popular or sound familiar somehow to make it easier for others to accept us - 'our music'.
I challenge you to express sound in a way that is Intrinsically you, because ai can't do that. And that is what the world needs right now. Your unique sound, you clearly have not created yet. Or at least, not put out into the public.
It's scary to go down that unique road because many people may not dig it. It won't be industry standard. Guess what though? ai won't be able to make it... until it's fed your stuff... then if you're continuing to grow and push boundaries with raw childlike abandoned sonic expression, without clinging to music history formats... ai will be playing catch up to us. Not the other way around.
So Landr (the distributor) now has an opt-in when you register and upload a title with them. You can opt-in to be used in their AI learning and be paid a royalty every time your music is learned from. You can also opt-out and not be included in their AI database but still use their services. I can see that as being a solution for the learning aspect of this equation.
6:02 it’s too late to fight. The data is now out there
Forget SUNO, forget UDIO. 11labs is coming from Poland
In terms of quality ai aint quite there yet but nobody would notice because most people listen on lossy formats like mp3's
Here is what I’m having trouble with when I hear people say these models are stealing copyrighted music. If I, a human study someone’s catalog and then creates music in that style that is totally fine and legal. Why then is it wrong if I, a human trains my robot on someone’s catalog and then have it create music in that style?
Legally? The same reason the court ruled that no one owns the copyrights to the selfie taken by a monkey. Copyright is awarded to human creations only.
Morally? Because it disrupts the process of iterating on music over time that musicians all over the world are engaged in and that as listeners we all value immensely. You have to ask, why do we listen to new music? Because we want to hear new perspectives on existing material. An AI can provide that in the short term but not in the long term since it has a fixed dataset of all the music that exists currently. If artists are impacted financially by this new technology and are thus unable to continue making music as they were before, then we loose that process of human interaction and the fountain of new and interesting perspectives eventually dries up.
that because the robot has to make copy of the that style. it means it has to break copy right to make copy for it's memory. copy allow copies in human brain but not copies by machine. in fact it has to have a lot unique songs in memory to produce good results. In more the better. with midjourney train on 16,000 artist. which means it could up 16,000 copy righted works in its memory.
because humans have ideas which are turned into expressions, then fixed to a medium. Humans extract an idea then reproduce a new expression. These models have no ideas - they measure averages in a body of expressions and produce new expressions. Furthermore, it's unquestionable that these companies have engaged in infringement - they admit to it freely. They hope to argue that their infringement is fair use, which doesn't hold water since their models create direct market substitutes for the material it trained on.
@@spark300c no, that's not how generative ai works. There are no copies of any copyright protected works kept within the ai models once training is complete. No generative ai has access to or memory of any of the data it was trained on. Your opinion is based on ignorance and misunderstanding.
Midjourney is in the middle of a big class-action lawsuit against it right now. Early on in the case, it was quickly understood that the plaintiff's complaint of copyright violation - which included a bit about copies being made, as you describe here - had "no legal basis" to be taken seriously by the court. They're basically getting laughed out of court. That's what happens when you take action based on your feelings, instead of being based on the facts..
@@SineEyed the problem it produce copyright works. the music generation uido has output copy righted songs. so how image generator out put picture of batman if does not have picture of batman in memory. mean that generator AI can not produce certain images it not know what they look like. So far call it super hero test. we turely do not know want in these iA model because they have new release it to public. I also read some AI companies know they in wrong so have ia cover up it tracks because creating derivatives to be stored as training data instead of real works.
08:00 Wait, but you learn music to eventually make a profit, right?
I'm happy if we can kick the music sharks at the top with AI. There are only a few who earn their way up there stupidly and stupidly. Music has been commercialized to death in the last 20 years. The music business is a hard and cold mafia-like construct. The smaller studios and production companies should not feel addressed here. The radio music that the AI generates is competition for the very top and comes into the homes of the big guys like a Swat team. Let The Games Begin....
To me the whole damn Thang is wicked. For some if not most, Music is their salvation from destitution and struggle. A.I. will turn it all into The Capital Wasteland.
A lawyer that is trhis young and says dood :)
I love AI music generation tools. Im here for it. SO fun.
If training it on copyrighted music music is infringement when money becomes involved, then, by that logic... so is the influence human music creators/artists claim when they make money off of it as well. I love using AI to help in music production which I've been doing for over 2 decades now.
What does it help with that you couldn't learn?
@@ajat3202 it accelerates the production time substantially. Its just not humanly possible to produce something of the quality offered with the aid of AI on my schedule and budget.
@@ajat3202 it accelerates production time. It's just not humanly possible to output at the pace AI has enabled. It's just a nifty tool added to my production suite. I think a lot of people are assuming it's gonna do all the work... Well, I like to add my personal touch including my own vocals/guitar work/keys/percussion, etc...
Talent is king.For a poor lyricist AI is absolutely necessary and can make a talented song writer be seen. With out it they may be lost in the dark.
We're going to see how artist ends up using it in the future. I see the younger generations taking more advantage of it than those that have been in the industry longer
Gonna somewhat disagree with Jesse in terms of licensing and AI training. AI training is just a digital version of a human artist listening to a track and learning to play covers and turning that into experience. Does the human artist have to pay a licensing to learn a song or learn to play in someone's style? Not really. At most they might have a Spotify account or perhaps bought some tabs or sheet music. I mean you got talented artists like Anthony Vincent over at 10 second songs. Do you think he paid a license for all of the impressions he does? Doubt it. From a philosophical point of view here, why should it require a license to learn digitally versus a human learning simply because some perceive it as an existential threat?
I mean Krystle does pose the same question to him. But I think the point is moot. For example - cover bands. They trained and learned a bunch of music. They get hired for a bar or some other gig. They're making money for that gig - a profit. And it's NOT at all transformative. They aren't paying licenses either. Or what about all the cover bands on UA-cam. They might not be making ad money from UA-cam but I can guarantee you a lot of them are making money from Patreon from fans. That's profit.
Sync has been broken since the 90s. Services like Taxi and Broadjam are absolute scams. There are 1000s of artists with dozens or 100s of tracks each for every singular need for a placement.
The pay is abysmal unless you are the go to person for a studio or famous enough your name brings in viewers. That’s why so many famous chart tracks are show horned into shows, movies, and commercials.
Blaming Ai sounds like a terrible case of survivor bias from someone mad their gravy train is heading out.
I am singed non exclusively to three Nashville film sync libraries and have had several Amazon and Netflix placements. So no, I’m not bitter. I just understand the actual reality of what’s happening.
Boo AI 👎
🤣🤣🤣🤣🤣🤣
crazy fun thing with udio: try typing "reversed" at the end of your prompts. 🤯
it also lets you straight up type the instruments you want to hear. lol this is so fun. also you can type shit like "reverb" and recordings to get more "record/sample" vibes. you can just type "sampling" too.
Why do you need AI if really talented and experienced producers can come with an original fresh track in 1h
That red lobster ad was made by some tone deaf marketing executive. It literally sounds like hold music when you call the water company or something.
Red lobster will go bankrupt if they're down to that.
I think the best thing about AI music is how it will help real artists produce albums far quicker. Imagine not having to wait another ten years for a Tool album? I am sure Maynard, Adam and Dan wouldn't want to use AI to help them make Tool albums; I am simply using this as an example.
AI is the artist then
Human brain also uses training data right from the day of birth, with this analogy nothing on this face of the earth is original 😂
It's not about being original. It's about respect
@@ajat3202 Yes, I respect all humans who put their brain in the development of AI😂
Are you single?
i see the video that he made of udio and i think that he is a crying looser , he needs to go to other place as photography , art , he needs to get out of the box is like a crying baby
@@laif9857 🤣🤣🤣🤣 I think he is delulu. Sync music is dead
Forget SUNO, forget UDIO. 11labs is coming from Poland