The Problem w/ Suno & Udio AI Music
Vložit
- čas přidán 14. 04. 2024
- Free TV/Film Sync Licensing Course:
www.syncmymusic.com/freecourse
Get Professional Feedback On Your Music:
www.profeedback.me
Music Library Directory (300+ Companies):
www.syncmymusic.com/musicdire...
My Music Library Recommendation Service:
www.syncedge.me
Video Tutorials On Creating Licensable Music For TV:
www.syncacademy.me
SMM Podcast:
www.syncmymusic.com/podcast
Sync My Music Main Site:
www.syncmymusic.com - Hudba
‘Easier to ask for forgiveness than permission’ seems to be the approach they are taking
Cheers again for the informative video Jesse - hot topic indeed! Here’s a good one - just saw an ad for ‘Music AI Generator’ - quote “No musical experience or knowledge is required to use Music AI Generator”. Nuff said 😂
Is there any organization speaking in our behalf in regards of AI in the music industry? The PROs? The big libraries (BMI, Warner…)? Is there anyone getting in front of people in power to let theme know music creators even exists outside the big celebrities?
I haven’t heard of any cohesive effort, but I would surprised if the PMA didn’t come out with a statement about all this soon.
very good points
Good points 👏 value your data 👌
IMO I think, A.I could be a great companion for your creativity in music production.
Thats the approach we've taken at TwoShot
Validated. Especially for generating Top-Lines. They may not be perfect, but you can get close and in turn, perhaps create better top lines. On the other hand, we can generate music and USE it as a learning tool (especially for beginners) on structure and the inner workings of arrangement. You have to think differently, because this stuff is going nowhere soon. It’s going to be one giant loophole. The people abusing it now will end up in litigation years down the road. Slippery slope. USE A.I and don’t let it use you. It’s a tool. NOTHING more. PERIOD
@@86crush Well said.
@@twoshotappyou sure about that? From the looks of it, you guys are directly ripping tracks, which doesn’t contribute to music making…All you’ll end up with is lazy production and future litigation after you get taken down. Nice effort, but you’re gonna end up on the dark side of the force.
@@RAM_845 I personally have used to to fill in the blanks for a good rhyme scheme lyric or structure. I’ve studied these things, but still really stink at it 😂 so for me, it’s at the -very- least…entertaining. That’s all it should really be, as the cream, will always rise to the top! Music is a LIFETIME of learning. AI is a weirdly-shapen speed bump. 😂
great info
I wish AI was more in the Band In A Box sphere where you can write your chords and mix and match actual performances of instruments to come up with unique songs and sounds. I think of it like studio session players in a box. They have a generative AI built in based on single instrument performances, and their Real Tracks are very impressive. They group them into styles, but you can freely mix and match based on the chords or notes you enter. This is the kind of d of AI that would be useful to composers, and involves, you know, actual composing. But my bet is that in the end, politicians and the big players in the music industry will sell out composers and just plod on without giving a shit about what happens to the real humans this affects.
AIVA is a little more like this
Yes Udio is like this. If you prompt to mix in styles you can have Beyonce singing on a Salsa, with a goth Metal bridge. But if you prompt it exactly in their style, you'll get exactly their style. So it's all up to the prompter to mix or not.
Don't worry this will be there within 1 or 2 years
Just tried Suno. They are indeed making full tracks with vocals and it's... horrifying for the music industry.
Indeed, but since they started doing so many vocals using autotune etc, I think they have lost the high artistic moral ground.
Also, think about it….eventually everyone will be able to spot AI tracks. They ALL have a certain timbre, like a really low quality mp3 from the early 2000’s. It will be really cool for about 5 minutes…then it just feels icky and wrong.
It's definitely a novel fun to mess with at the moment, but as it gets better, who knows where this will go.
@@getkraken8064 oh I totally agree
@@YoungBlaze Very true, good call.
Im glad you put this video out. I thought i was the only one hearing the similarities. So of the style i like sounds like a country singer that is no longer with us. Strange ya think...
Hey Jesse,
I wonder if we will see established Production Music Libraries ease back on recruiting new composers in anticipation of the impact of the new A.I. landscape?
Not sure about that. Right now I’m trying to help libraries understand the landscape so they can make the right choice for their catalogue and composers.
My question though, what method would be used in the courts or otherwise to prove that someone used or did not use ai to create their music?
Covering that in a video next week!
@@SyncMyMusic looking forward 🫡
It's gonna be on our shoulders to prove how we produced the song without the likeness of the popular artist. If I have a friend that sounds exactly like Eminem, then I'm sure I can get by. But If I don't... then what's stopping Eminem from suing me that my song sounds exactly like him...
The burden of proof is on the accuser.
@@8bitninja64 that’s why there’s the discovery process in lawsuits.
Thank you for your videos... it's heart warming
I agree 100% Jesse, NO ONE-TIME FEES! And i think it should be a least a 75/25 split for us. Like you said, the AI code is literally NOTHING without quality content to learn from.
Your point about that theses AI music models are only as good as the content they have been trained on is a really great point. The irony is that most of these models need high quality music that have been produced by humans. That is pretty interesting and a bit of a mind bender.
more horrifying when you understand that they've ALREADY trained themselves and now have no need to train on potentially copyrighted material. they have exited the henhouse with all of the chickens
I tried it a lot. Many of the songs that I wrote in swedish sounds exactly like some of the real artists. Not only the song but the whole production.
But is it the same song or melody? That is what matters to copyright questions.
There are million of tribute bands doing music in the style of... for ages. To exactly sound like is not an infringement.
@@samthesomniator Agree. It's not the same melody or the chords of the original artist. It sounds like they just made a new track with the same singer and the same style and instruments. With the current law that is not a copyright crime. The artist will not like it but still, not a crime... yet.
It might help those artists to write better songs and extend their careers. Most bands struggle after a couple of albums.
Was wondering what your take would be. The moment has arrived, the one button = song Ai models are here.
And yet the sun still rose this morning.
I've done some testing on both Udio and Suno, and my conclusion is that the tech is very impressive, but the music on the whole is very inconsistent, and the audio quality is still very poor. Whenever the AI "nailed" a cue, it seemed to me as if it had simply lifted a cue note-for-note from well-known stock music tracks, particularly "epic orchestral" prompts would return the exact same epic orchestral cue that we've all heard a million times from 2 Steps From Hell, for example. It was so similar that it might as well have just been a search engine for stock music, because it didn't feel to me as is the AI had done any improvisation at all, merely just ripped the track note for note and regurgitated it at a lower audio quality, and not even locked to tempo. And for the "worse" cues, the AI seemed to really meander with its improvisations, adding extra beats to bars where it doesn't make sense, and adding extra bars to chord progressions that don't make sense. It would give the "impression" of "good" music in short bursts (10 or 20 seconds) and then it would all fall apart quickly after that, with instruments dropping out randomly, or the chord progression would loop incessantly, or there would be no variation in the melody, or any other number of off-putting creative choices that a typical composer wouldn't make. On the whole, completely unusable for your average composer, and not as good of an option for a typical customer as even basic stock music tracks. That said, while the instrumentals are quite poor IMO, the lyric/voice generation technology, however, is very impressive. The way they capture the cadence and intonation of a singing voice, and not just a talking voice that's being pitched up and down, is incredible.
Can you share the links to your 'epic orchestral' experiments? I tried to create something in the style of TSFH and it failed miserably.
@@snarf1504 Try these words in your prompts in Suno: Medieval, castle, epic orchestral, epic strings and brass, fast string ostinato, heavy epic percussion, suspenseful, building up to a climax. Some combination of these words consistently returned some very familiar instrumentals that sounded very much like 2 Steps From Hell, or at least a stock music version of them. Suno's AI seems to handle more popular, well-established types of stock music tracks like that, but is inconsistent when it comes to certain other genres or types of tracks.
@@snarf1504Use Udio, Suno is bad for orchestral
@@Andrew-qq8fbThanks, I tried. It spits out generic Audiojungle-level 'epic' music, not close to TSFH stuff. But anyway I agree with the rest of what you said.
If the AI art generation companies haven't had too much legal problems. I don't think music will either since it's even more nebulous. I don't want to say that, since it's my career, but here we go.
The music business has a much stronger history of fighting for copyright than the art world, but yes we'll see what happens!
Legitimate question. I never gave Google the authority to scrape my website so they could improve their search engine value. What is the difference between that and scraping web data for an LLM?
That’s a mutually beneficial relationship. They scrape your website and profit off the ad revenue. In exchange you get more web traffic to sell your own products or ad services.
You can prevent google from scraping your website, google "robots.txt"
@@SyncMyMusic
This isn't entirely correct. Google actually had to change the way image search worked, for instance, because it became apparent that people were just grabbing the images off of the search page and not going to the scraped sites, thus driving down traffic and driving sites out of business. Google realized if they killed the sites they were pulling from, they'd have nothing to show people on search.
Basically, any animal without a natural predator will eat itself to death. AI experts are already worrying about what happens when there's no new data left to mine and the models start being trained on AI output.
The question is are production companies going to be interested in getting into this potentially nasty copyright fight? I'd imagine a lot of them would say no, some might be yes. Still a massive risk vs reward issue for them.
So what if Udio creates a new model trained only on the million of user generated AI tracks already made? So not trained on any copyrighted work, but trained on work generated previously that was trained on copyrighted work. I have to think this is their backup plan and why they left it free to use for so long; so they could get a massive library to train from
I've been using Suno for providing the samples for lyrics I've written, all of the tracks I've created through Suno I have written 100% of the lyrics and have logged the writing process using notepad/notepad++ using version numbers, demo "X" and remix of "X"
The quality of the sample generations varies and I've spent significant time altering the lyrics and grammar to produce results similar to what I'm looking for
To make a full length track you require to "stitch" samples together and quality can take huge dives the more "stitches" put together and this can be a varied and time expensive process that takes a fair amount of patience
I have been playing with the thought of taking the concepts I've written to bands I know or a production company and paying them to produce the tracks
I absolutely agree that musicians, sound techs and other talented humans are required for the AI to produce high/higher quality samples/tracks
At the end of the day there's nothing really quite like the real thing, being in the same room when the instruments are being played, the texture in the vocals, it's a full experience and I won't help to contribute to the downfall of that, if anything AI should be a tool to supplement the already existing talent, not detract from it
As of early winter '23, several music libraries intended to license their tracks for LLM training. At this point, I think it is really only fair to assume Suno/Udio did their training above board, and cast the shadow on the libraries who sold/licensed their data as soon as they could - no?
I don’t assume anyone is above board, especially companies who don’t disclose where they got their data from. Curious how you learned some libraries are allowing llm training?
@@SyncMyMusic - Again, we need to understand what is meant when we talk about "data". AI doesn't scrape or copy tracks - it learns from the music, the genre, the rythm, etc. the same way musicians do. The Udio and Sudo databases simply are not large enough to contain even a fraction of the entire songs they examine. In many of the largest graphic models, for example, the number of images examined works out to one pixel of imagery for every image viewed. Now you can see why courts have not ruled in favor of artists yet. There is no direct copying going on - or even possible.
@@TheoCage data can mean a recorded file of audio (mp3, wav, etc). AI models need some kind of digital, tangible medium to train on.
@@SyncMyMusic - I agree that they need content to train on, but there is a common misconception that these models copy everything they see and then refer back to it when generating a response. That is not how AI works in these cases. The model is 'learning' from the material it is exposed to - almost exactly the way that humans learn music from their exposure to different artists and genres etc. To say that AI shouldn't learn from established music would be the same as saying that musicians and songwriters cannot learn from the music of others or would need to attribute all of their influences.
@@TheoCage We agree on that. I know AI models aren't copying or storing the data they train on. Watch my most recent video for an explanation for why humans and AI's aren't doing the same thing when they learn and get inspired from other music...
We need to be responsible in shaping out how the future of music would be! Let's do our bit. Otherwise, we'd regret it for eternity for not taking any effort on our part. We music creators are the only one who can decide how valuable we are! If we do not take action, no one will take it for us and we will become an outdated commodity. Jesse, perhaps we can 'try' to do something collectively? Could be something simple to make ourselves heard, like hashtags with constant tweets or something to build up some trend to our favor!
Love your spirit in this Zack! Yeah I'm not sure all that can be done collectively, but I just focus on what I can do - post videos here and try to educate my subscribers and Library owners to the best of my ability. I'm sure you'll have unique talents and skills that can help the cause as well!
In the future of music there should be creative freedom, we should be able to imagine any song and make it appear in the world, I could think of the next banger that would have appeared in Michael Jackson's Thriller Album and everyone else could listen to it!
The battle that is being fought is against this technology by people that are afraid of being replaced and having to do something else because the common man will be able to create whatever they want with AI and for free, Music production is being democratized and it's great and it's a shame people are focusing about what other people shouldn't be allowed to do.
Like a great chef being afraid nobody will go to their restaurant because they can make delicious food at home, and wanting for it to be illegal so people can't do it and have to visit their place.
That is what you are doing, you are on the wrong side.
@@SyncMyMusic Of course, and makes sense. My point is, let’s bring our individual input, but use like 1 or 2 hashtags or any similar idea that can help us be more consistent and under one trend. You keep on doing what you’re doing but use such a hashtag that others can also use in whatever platform they found themselves active on! I’m sure with our limited time we can still be more focused and organised to make the impact. If you want, you can throw this idea in a video or poll, I’ll also do my best to make sure we are heard.
@@Ovyron sounds like you don’t create music for a living?
@@SyncMyMusic Sounds like people creating music for a living could make great use of this new musical technology? To get inspired by AI creations? To cut chunks of what is generated into their projects to save time? To make songs much faster because they can polish one with great potential instead of having to start from scratch? Imagine when people get the power of making full length movies with hollywood quality at home thanks to AI and they finally are able to tell their stories the way they want, but people that "create movies for a living" want to make it illegal. Let's cooperate instead of competing, I already know people that create music to profit from it, and they started with the release of Udio, so I can only assume the old guard is just afraid to be defeated by people using this new technology while they refuse to benefit from it, and want it banned.
If ai companies get shut down will it stop ai music being made/used commercially? My prediction is that companies will stop using the term "ai" swapping it out with something new and boring.
Now they've added this new feature with uploading your own audio, and now they don't have to wave off questions about what data they've trained on , they'll just answer that it's on user uploaded material. Unfortunately there is no technical way to verify this, this applies to all neural networks, it's basically a black box. I think technology is so far ahead of the legal system that it's already an unsolvable problem. I don't think we'll see any lawsuits because it would be impossible to technically prove that they trained on specific artists or songs. More specifically I don't think we'll see lawsuits against SUNO/UDIO, but may well see lawsuits against their users who received the generated tracks in full confidence that they could commercially use them and it wouldn't be plagiarized by any artists because they directly took it from the generative AI algorithms rather than plagiarizing it themselves manually
It’s scary how good the songs are, I just checked Suno out and the quality of the tracks is really good, I studied at the Sync Academy and the amount of talent you have at the academy is huge! There are so many talented people at the academy who work hard on their tracks that deserve to get their music placed, I agree, AI is great if the intention is to work with us the composers to help enhance our tracks, but it’s not right if it’s just ripping everyone off, i would believe the music library’s wouldn’t accept pure AI tracks due to the issue at hand and surly most companies wouldn’t take the risk of using it, normally I run my comments through Chat GPT to make sure they make sense l, I dare put this into chat GPT, it would probably delete the message due to the content 🤣
I opted in years ago when asked and I'm so glad I did , and so does my bank account. Remember the old saying " Some $ is better than no money " ... well it still holds true. I don't let the unknown get in the way of (my) big picture which is "to make a living doing what I love to do and that's Music" . I just go with the flow ...... 😉Just my 2 cents . NOW if you asked me if this whole A,I music thing is good or A.I in general and my answer would be "NO", get rid of it now. The only people who use Music AI are ones who can't create on their own or are lazy and prefer to lie and cheat the system and the people. Think hard and let's be honest about this topic.🤔🤔
BRO panickinG while talking about UDIO AI
It don't matter who they are training off. They are not violating copyrights. I did a bunch of songs on one of the sites. I wrote all the lyrics. Took what the A.I. gave me to the DAW and arranged, mixed, added harmonies etc; . I used google
song recognition App and not one of those songs were recognized. I tested a bunch of songs from Artist releases and every single one was recognized Immediately. The problem I have is that the companes have what is basically your masters on their servers. From personal experience, if you do something good, they are not going to let you delete your music even though as a paying user, you supposedly have all rights. I won't use A.I. again unless somebody creates an app you can use on your own computer. Great lyrics would be just as good a song with humans as A>I
I think people will pay good money for an App like Udio, that can be used like other DAWS without signing to a service
Do you have copyrighted music that generates income?
just had a go with Udio.. phew.. my stomach just dropped to my feet, terrifying stuff. This is the best I've heard yet and definitely library music ready.
With that being said - I definitely heard Nicki Minaj, XXX Tentacion and possibly even Blink 182 in there... I don't see this site lasting very long, something is definitely fishy.
Im here to do whatever I can for this battle and I wont go down without a fight. But this has definitely left me pretty shaken up.
they are 100% using copyrighted material despite saying they “replace the artist names” with genres once you put them in. I for one have generated songs with exact replicas of pitbull, rihanna, and beyoncé’s voices-it really can’t be a coincidence
But really, is Nicki Minaj or XXX Tentacion actually Nicki or XXX, or are they half Autotune? It cuts both ways.
@@getkraken8064 umm no, it really doesn't...
I visited the Suno subreddit and saw a thread posted by someone of their full album of songs they had Suno AI create and had already posted it to Spotify. Honestly, it's so aggravating, because the songs are so bland and very 'AI generated'. Bugs me that this person is already going full in and cashing in on AI generated music. I imagine this issue will grow in the future.
I assume they haven't read the terms and conditions on Suno. SUno claims ownership of everything and does not allow commercial use of anything created on Suno. Looking forward to Suno suing.
@@davidord2934 That's true for non-subscribers, not for those who pay to use Suno.
It can't be any worse than the mainstream music out there.
Beyond legal challenge, some of my friends who produce for Netflix etc. have indivated that film music directors would rather hire real musicians due to the complications of trying to create music that syncs with film. They are not likely to try to fiddle with umpteen prompts to sync music the film. It’s just too onerous.
By the timeany laws are enacted or court cases dealt with these will already be the standard and will be able to train on themselves. Stock music is already pretty much dead.
Not necessarily true. The law recently proposed to force AI companies to show their data sources would be applied to retroactive companies as well. So yes they might be in business for a year or 2 before a bill gets passed, but in the long-run we don't know how the law will affect this all.
@@SyncMyMusic A year or 2 in AI terms is like 20 years in human terms. LOL. Look at Midjourney - last year it was like a 10 year old drawing, now it's more than pro ready. Visual artists are taking them to court, but it's too late. These companies embed themselves in the population like Uber, then by the time they go to court their user base is so large it's too late to go back - especially with the public backlash. Metallica learned that the hard way.
@@blurtam188 Napster "killed" the music industry and somehow we managed to become profitable in the new paradigm. Nope, won't buy the "its too late" narrative.
The more days go by copyright means less and less, and every piece of software will be worth 0 dollars sometime in the future
I look at it as a starter track. Get an idea first of what you wanna do. Because for one, It's never gonna be perfect, so you can simply re-arrange, re-compose, a better version ! Thats 100💯 legal if you are making it after the fact. Like re-recording a well known song for an advertisement.
I agree with the partnership model of 50/50 share in the revenue. I use AI to give me Ideas when I am creating.
a real awaken nighmare suno music experience TuneCore DistroKid 😝named DannyHO ,who realease this 42th albums in 2 month lol
AI composing will obviously go the same way as AI art. It won't be copyrightable. meaning if we publish a song within udio/suno, anyone can just take it, add a kick, and claim it as their own. Be careful what you share!
udio is amazing
There's a reason why these softwares are still in beta. They know they are playing with fire with copyrighted data. Like AI art, give it a year for the courts to slam their wrists. If Blurred lines had to pay $5million to Marvin Gaye, what kind of lawsuits are gonna happen when your song sounds exactly like a popular singer?
Give it a year for the court to slap the wrist? How about giving the courts five years! That estate of Marvin Gaye's suing Ed Sheeran dragged on and on and on forever. The first suit should've only lasted five minutes before the whole thing was dropped. No other suits should've been brought. Usually, jurers don't have a grasp of the complexities of music, let alone AI technology. Someone has to show them how dangerous and fraudulent this technology can be. Ed Sheeran took his guitar into the courtroom and played a lot of similar songs that had the same chord progression that he had with his song thinking out loud. Only then did the jury get it.
@@jessmithmusician i think song styles, and chord progressions cannot be copyrighted, but the likeness of a singers' voice can be. Yet what can the famous singer do? Is ai fair use? Many cover artists can sound like their favorite artists.
First there was this discussion among the visual artists, then among the writers... now it's the musicians' turn...the musicians probably have the strongest lobby and can at least perform live.....as an independent artist and musician, who has been publishing art (i.e. data) on the internet for 20 years, I have no hope of receiving compensation from AI companies.
I have decided to flee forward and use AI (I will never pay for it! ...and I try to use it as ethical as possible).
It's interesting to watch how I mutate from an artist to an art director...
....it all boils down to a love-hate relationship...
Because AI is based on all of us's data, it should be available to everyone for free
Happy colored Greetinx
im kinda in a middle point about the morality of using the public data to train models, on one hand its public and you can do anything with it, everyone can see it everyone can use it but on the other hand they are profitting off of this bussiness so im not sure. but if i had to choose i would say its morally okay
why is audio quality sooooooooooo bad
For me or for Suno/Udio?
I checked it out it doesn't seem like it can come up with fresh new musical ideas. Or is it just me?😅
Generation isn't creation.
@@redwithblackstripes Generation: "The production or CREATION of something." Thats literally the definition...
Learn how to prompt. It can 100% come up with original/creative ideas. Don't just ask for "an upbeat rock song" or something like that. Be creative and it will be creative.
@@overlords2722it generates variations of what it knows. Whether that is 'creative' is up for debate.
@@WillyJunior I know how to compose and produce by myself fine i don't need to ask anyone to do it for me i'd suggest people learn that instead, pretty useful if your going to charge people for your music.I know generative syntax models and paradigmatic classification as well as the next ai, only difference is i don't need anyone to prompt my neural network or infringe anyone's copyright to turn them into actual music. Generation is making something from something else creation is making something that did not exist before, you got to go beyond typing "what is the definition of" in google on this one.Its possible to generate something without it being a creation you do that on the toilet seat everyday, different prompt, same shit. Learn to get creative with lunch i guess.
When AI reaches AGI They will proclaim it a being with rights. They will say the AI was inspired by other artists just like humans are.
The current models do not put out results that are perfect. You still need to do some mixing and mastering to get a really good sound. I usually accompany the AI with my own live music too. To me, it's not much different than hiring a musician to play our music or to play something random. I've certainly heard a lot of known singers in the outputs, some from way back. Not exactly the same, but the influence is obvious. Same for some singers though. I suspect the AI randomly pulls from top hits in various genres, and tries to mimic the sound, more than a particular artist.
I only use it for background music, but if I had a full studio with all the bells and whistles, I could definitely make it sound professional.
Consider, producers, singers and musicians also have to train on current and past music examples and theory.
The difference though is the output capabilities of AI models. Got a video on that coming this Monday. Thanks for the comment!
I don't know about anyone else, but AI images creep me out.
The music turns my stomach.
if u go to Suno, download a song, reproduce it and perform it with your voice u bypass the copyright issue. U can't be tracked and sued, is suno runnning content id on the millions of songs it generates each day and comparing them to social media and streaming platform videos and music for matches? imagine doing that plus doing that with songs that sound similar, it gets very fuzzy real fast. There's no copyright cuz they can't prove shit. and u can't copyright a melody. Only thing u should worry about is the voice, which u can bypass if u perform the song yourself.
From 1984 by George Orwell
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.” In April of 2016, “A New Rembrandt” was revealed (1). The painting, like the songs of a versificator, was made by machines. In August of 2016, Music Business Worldwide (2) accused Spotify of “creating fake artists.” What is a fake artist? Can music be fake?
The world of 1984 is a grim place. Members of the “Party” have access to resources based on their rank. The rest of society are called “Proles.” The term is short for the “proletarian” and refers to the working class. The Proles make up the majority of society, and so the Party provides them with various sources of entertainment to keep them from getting too restless.
I agree, BUT this is the situation at the moment, where AI is trained on data… the big problem by saying “no harm done…for now” is that we are feeling relatively“safe” and completely disregarding that the big companies are investing billions to go further and very soon. Wait for AGI or ASI, and everything will change again… when truly creative digital minds doing like or better than humans will be out there, it will be the end for us, in every human field, unfortunately… that’s the direction, unfortunately.
My question is: why? Why are these companies willing to put something out here that is easier, cheaper, but it will heavily cloud our future? Is anybody clearly stating that our art being replaced by technology is not the future we want? We are just watching companies putting out there digital products, that ironically WE are paying, that are going in the direction that they want, for money… We are buying and selling our future for what? Again… Why?
There is nothing we can do to compete against billions put into this technology, other than waiting, acknowledging , surviving... and that’s is even more sad
Practically we don't need a powerful Mac or PC as now we can make music with our phones and web-based AI. Even the Computer manufacturers gonna lose their profits. It's gonna change industry massively, not just the musicians hit by AI. No one knows what's going to happen.
I don't get why the copyright is lagging so much. Transformer based models are essentially a very clever form of compression, where the output is intentionally "loose" enough so the decocded results are not 100% accurate to the originals, but rather combined together in random ways. If you look at it that way, the act of encoding a song into a set of neural weights is essentially ... well ... copying. It's not the Suno's/Udio's output that breaches the copyright. It's the creation of those models that does. I have no idea how is this hard for legislators to figure out.
Sounds like a process very similar to the way musicians learn from previous musical art. If I write a song that is not a direct copy, but inspired by and based on learning I did from repeatedly listening to another artist - is that copyright infringement? Sounds vaguely like a thought crime.
@@TheoCage But ideally, you (or someone) paid so you can listen to that music. They did not. That's the difference.
@@farley333 - From a pure copyright perspective, it would be instructive to look at legal precedent. In HathiTrust, the courts agreed that gathering data such as book content from the Internet, was not a copyright infringement if the final use of the data was instructive or transformational. The intention of US copyright law is not to restrict "fair use" of content, whether it is paid for or not. Google also won an important case where they gathered the content of millions of books to create a searchable database. They were never aske to pay the creators of that content and when you do a quote search on Google you make use of that data. So there clearly is precedent for AI use of music or art.
Compression is probably the wrong word-I think abstraction is more accurate-the data is summarized into a sort of statistical distribution-that’s why you can’t really go back and retrieve the source data from the model.
75/25 percentage all the way! Forget the 50-50 percent cut. I totally agree, AI models may be impressive, but they are worthless and nothing without any data to train off of.
Although I understand your concern we really need to be careful with stuff like saying "this is just ed sheeran, blink-182, etc" .
If we go so far to say a style or vibe of a piece of music is attributable to one artist it gives the largest artist a monopoly on music and the excuse to take legal action on smaller artist for just sounding like them even worsening the Marvin Gaye/ Blur Lines lawsuit conclusion.
Or they can just claim AI was used to sound like them even without evidence which creates a legal headache for smaller up and coming artist worst than the AI problem.
Great point. I def don’t think artists can or should ever claim a “style” but my point was to explain that I’m hearing very familiar artists that MAY have contributed to the training of these AI models without consent or compensation.
@@SyncMyMusic Wow I just saw Weaverbeats video about this and he played examples on specifically UDIO and I didn't know how literal it was when you were saying they are legitimately just taking the whole voice of established artist, they literally have Future (the rapper's) voice and many other artist I completely understand now what you mean haha.
I used SUNO before but I think SUNO tries to generally imitates a style of music broadly where as UDIO is literally just taking people's voices .
Music licensing is almost done. Everybody who needs stock/sync music will go for AI. Game over.
The battle is just getting started.
@@SyncMyMusic True and it's gonna get though. Als releasing songs new songs will never be the same. It will take a bit longer but I am a bit worried about it. For sure I have to change my strategy now. The old days are gone.
I love Suno AI because I no longer have to struggle with home recording, sound engineering, mixing, singing etc to only come up with tracks not quite high enough quality. I am playing Bernie Taupin to AI's Elton John, as writing lyrics is my thing. I am amazed at the song generation of Suno AI, it does take some skill at prompting and putting song pieces together, but the final product sound far better than anything I could personally accomplish. Some of my old songs have new life and it's great to hear them performed so well. But I realize I am in a minority, most people want to be Elton and I am happy to be Bernie. Good topical video, thanks.
Sounds like an excuse cuz you cant put the effort in to me...
@@alicanozturk1360 I am one guy. I can't do everything. Sounds like sour grapes to me.
AI is creating new stuff based on experience. That is basically the same way how we do it. If someone grows up isolated only hearing Britney Spears music, that person will not write a song that sounds like Metallica if you ask that person to write a song. Music develops over time (like art and everything) and humans work based on inspiration. AI doesn't work differently. It is just scary because it is much more precise in what belongs together and faster than our brain. Due to CPU power there are no limits. Like an unlimited brain. The scary stuff is yet to come. But trying to stop it won't work. Professionals just need to get better. Doing photography cannot just satisfy anymore because you got some bokeh in the picture. Programmers will need to do more specialized stuff when users can click together their own apps and lawyers or medical doctors also will have to change. There is a reason humans develop technology... EDIT: One more thing: What is interesting about this is that in the past tech. was replacing more physical jobs, but now I think the shout out is so loud because now it will challenge the people with the "thinking" jobs (includes me...). But give it some time and things will sort out.
the revenue sharing sounds good but what if they trained of of youtube, does Google get to share in my generations?
Not sure I follow. How does Google play into your CR rights?
@@SyncMyMusic if the model was trained on youtube and Google owns these videos will they ask for a share?
The labels will license their catalogue that they directly control, or will create their own AI for their own use. Blocking them from learning off copyrighted music (the same way every human musician did and didn't pay for that use) only hurts the regular person, the corporations won't feel the difference.
But say it happens, and you refuse to license your stuff to them, and their output gets popular? It doesn't sound like you, so popular music moves away from you. Smart business move...
I wrote my lyrics and I put them in udio. I did spend 2 days in a row for each song. Does that constitute enough as human interaction?
Not sure! Try registering it with the copyright office with your explanation and see what happens!
Here’s a FACT! Every music producer gets their production ideas from listening to other artists.. called SAMPLING btw.. so that makes everyone a copycat
alright no one should be harassing you at all, but my view is as someone who is basically scratching at the top of making my own music and trying to turn into a career, seeing this concerns me because yes its not good now but what happens in a year or two when it can have a proper song and also we know how corporations operate. I know im dooming here also no hate on you at all, just voicing my opinion.
this stuff might me kinda end the idea of a career or trying to get harder because its a case of it might take me a month to make a song or a few songs when an ai can produce 10 songs in 1 hour.
Ai can't replace hatsune miku lol xD
Everything is a copy of a copy of a copy. Doubt, this will be much of an issue. If their is profit to be had these companies will blast the doors open without much difficulty imo.
Thing is, what is a human artist without input of high quality data? 🤔 Your brain does also not produce in vacuum. Do you have to pay people that inspired your organic neuronal network aka brain? Would be the logical consequence
That’s a common argument I’ve heard. Will release a video Monday challenging that assumption.
Not to be combative but does fair use come in to this anywhere?
Watch some of my more recent videos for a better assessment of my opinion on fair use.
The quality is crap to my ears :(
Man…i think what you say makes sense, but it could be utopia..Why would one need permission from production music library to train AI if its already available to play online on the web? If someone asks where do you train from, AI owners can say “from free stock music”. Who is going to prove them wrong? Audio jungle, Pond5…they all gave the data away for $, and they also have high quality music there. Less, but there is.
This is very tricky. Imagine you have a company of 3000 composers who constantly write music matching some style for music that is available online. This is what AI is doing basically…and is just copying a style, people do that all the time - this is not copyright infringement. What can you do against it?
Check out my video tomorrow. There's a new law being proposed that could potentially solve that very problem you've accurately described.
Looking forward to it!
I don't see a difference to a composer who has employees who come up with ideas, like Hans Zimmer f.e.. exactly the same, from workföw and even to legal issues. No composer would let his employee do the whole work and put his name under it. But they are using their basic ideas as a team. Now with AI, anyone can have assistants! :-) like laptops who made producing affordable for anyone a decade ago.
People keep using the term "scraping" when they talk about the process of AI examining images or songs or writing on the Internet (or from various databases like Shutterstock.) Scraping is a term that goes way back to the early days of the Net. But Ai doesn't 'scrap' data anymore than humans scrape fine art when they visit an art gallery or listen to music for inspiration. That's the reason why not one single copyright case against AI has been successful yet. Judges have learned to understand the difference between learning and copying. I doubt anyone, no matter how long they work at it, can duplicate a known piece of music using Udio. The same way that AI art generators have not recreated an exact copy of a given piece of art (I know, I have generated over 25,000 images on MidJourney so far).
Artists and writers and musicians have a right to be concerned about AI stepping into their space, but that tech is clearly inevitable - and we are still at the infancy of what artifical intelligence can create.
Copyright is on the way out.
So is the sun, but let's fight one battle at a time.
But W.I.L.I.A.M supports them..." This makes me think 🤔 This guy is not just an artist; he's also the "Director of Creative Innovation" at Intel, a multinational technology company that designs and manufactures semiconductors, microprocessors, and other computing-related products. Intel is also deeply involved in AI technology. Could they benefit from his data and data that he can deliver from elsewhere?!
Doesn't matter who supports any of these companies. We all need to think for ourselves 👍
W.I.L.I.A.M is a producer that actually also uses other people's songs to make them his own. So of course he supports this, he's been at it for years.
@@SyncMyMusic Absolutely!
@@Toxicflu I do not agree on this. That he is featured doesn't mean he takes the ownership
u guys don't understand and thats why you're in this situation, you naively believed that human creativity was some magical thing, its not, its computation, novelty is computation, its equations in our head that we don't understand. I've known this was happening for years and prepared myself, I'm a music producer btw. keep coping and saying the ai can't truly be creative. maybe you will learn when its too late
Speaking for myself, I'm fully aware making music is a "magical" thing. It's an art form full of complex data points that now cpus are able to do as well. It's obvious AI is creative as well. What we're talking about is requiring the HUMANS who created these AI models to get permission to use our data to allow their AI models to generate creative output.
@@SyncMyMusic the reason that’s tricky is because when an artist makes a song you don’t go back and ask him if he had permission to listen to all the music that inspired him, even if he paid for some of that music or pirated some which is very common. so u can’t make that argument about the ai just cuz it can listen to a million songs in a minute or something. copyright only comes in if the ai produces something that is exactly a rip off of something else, every other thing is fair game. plenty of songs share the same progressions and similar melodic motifs. the final product being compared to pieces of already existing music is where all that will be decided.
@@olabassey3142 The difference between humans an AI is in the output capability. Tomorrow's video will shed some light on why that argument has some major weaknesses.
@@SyncMyMusic will look forward to it
You’re kidding yourself.
Which part?
If you had cancer and a new robot was available that operates better than any human being, you wouldn't be here complaining that a surgeon lost his job. So stop hypocrisy. Yes progress makes some professions obsolete, roll up your sleeves and learn prompt enginering or shut up forever and don't complain that as of today we computer nerds make the money from music.
No one is complaining about lost jobs here. I'm fully predicting and aware our industry might be fully disrupted in the coming 5 years. We're talking about protecting our data which we feel needs permission to train on. Did you watch the video?
@@SyncMyMusic I admit I only managed to watch half the video, then I had to stop. I'm sorry but this type of reasoning isn't in my veins, I can't stand it, I feel bad hearing things that I can't share even as a joke. If I am creating a fake, a work that I say is made by that artist, whose name, fame, style and everything else I am exploiting, I agree that it is incorrect and that it must be, as indeed it is, illegal, but imitating is a completely different story. Claiming that you are prohibited from emulating the work of others, or even worse, drawing inspiration from the work of others is something that I personally consider horrible. Did you invent the buildup? The drop? The guitar solo? And then according to your logic you should be prohibited from creating your buildups, your drops, your guitar solos. And just because an artificial intelligence is more capable than a man of learning from existing works: 1) it doesn't mean that humans don't do the same, no one invents anything, it's always just contamination, evolution, improvements of pre-existing things; 2) in every civilized country in the world there is the presumption of innocence, until it is proven that I am committing a crime I cannot be accused of anything. And why should a human musician hear other music be okay with it, but if he does an artificial musician it should automatically be accused of violating copyright laws? And if you make your music public, who gives you the right to establish that a particular category of listeners should be prohibited from listening your works? It's true no one talks about stolen work in your video (maybe, I repeat I only managed to tolerate half of it) but the concept is the same, you complain that some non-human also enjoys your content which you also decided to make accessible to anyone and that by listening to these works of yours this non-human becomes better than you at doing your work and you say that this must be illegal, while I, a human, only because I have no musical ability to imitate you, I can listen, study, try to imitate, all of your work just because as a fact I am harmless. Laws against plagiarism exist, why do they have to be different and depending on whether it is a man or a machine? If this isn't hypocrisy, tell me what it would be.
Anyway, thank you very much for replying to my initial message. I hope for a civil discussion, I know that I am not the holder of any truth at all, I only express my opinion and respect that of others.
@@AntonioSorrentini How quickly can you and I consume music to learn from vs. an AI model? How much output can you and I create vs. an AI model? We're talking about very different levels of productivity here, one of which can massively disrupt and jeopardize human creativity as a whole. Laws are created to serve people, not machines. So while the initial "training" is similar between people and AI, the implications for AI generated music is obviously something we need to be thoughtful about.
what if these companies hire pro or semi pro talented artists and musicians to submit recordings just for their a.i. database. this could be a "win win" situation for both parties because now maybe artist that struggle to get commercial attention now have a new way to be compensated monetarily. personaly i hate this a.i. bull. spit!i think it's a real punch in the stomach for those of us that have worked hard at our craft . it discredits those who have perfected their individuality and creativeness. these a.i. models make me feel like im being slapped in the face by a group of really smart dudes that figured out a way to produce music that is essentially stolen, glorified song mashups. this is just a continuation of the structure of todays society filled with absent minded thieves, careless cheaters, and in cognizant phonies,
You'll get like 20 cents per song lol.
These models draw from millions of songs.
Music is about to get devalued immensely quickly.
Yeah, bro, I want to talk about the problem with you. Don't you remember the story of King Canute? Name me one artist who didn't learn and steal from those who went before-just one. It's okay. I can see you will become obsolete in the very near future, while anybody with sense would be using AI to move forward.
i feel like you don't know what you are talking about ;D
Thanks for sharing your feelings. Any facts or specific points you want to let me know about?
@@SyncMyMusic Thank you for responding ! And reading my message carefully. Can you make a video about why it's hard to label stealing in music really as "stealing ". Because as humans we need to build upon our human heritage , put common language called music 🎵 . ?
It would be cool. ❤
Last 2 decades artists already sound like AOL Chat Buddy and look no different. Nothing of value will be lost
Yeah, once everything became autotuned to death it already sounds like it's computer generated - including the voices.
It's too late, they don't need your permission or want your music. This is just massive cope.
If you're already giving up, that's fine and it's your choice to do so. But I'm not done with the battle yet.
@@SyncMyMusic I've not given up. I've just come to terms with reality - AI will make better music than humans. I'm already using Suno and other AI engines to embrace and and adapt to the changing landscape. The days of making albums for libraries is over, evolve or become extinct.
@@Silver_Town_Music those sites won’t give you a copyright for the music they create. They say they will, but the courts say otherwise. Just make sure you’re aware of this before you release AI music publicly.
@@SyncMyMusic I'd argue otherwise. I agree that the courts have said that, but it depends on the process. If you're putting in lyrics, and doing a collaboration with the AI, back and forth, picking sounds you like, not just picking the first song it produces for that song, I actually do not think it is a lot different than coordinating with musicians who get the vibe you're going for, start playing and then you add more and more to the song. I think this is a tool that is here to stay, and rather I think that the process will always still be a collaboration between person with idea, artist, and the technology. Of course there will be people who put in the most minimal prompt and get some music for their CZcams video. Though I also suspect that too, will have some protection, as in mechanical production, and if they did some back and forth with the AI, choosing variations. I think it will become impossible to differentiate this from collaboration with humans. I think there will be desperation for a while, and then most newer artists will embrace the much wider creativity it will enable for their art. To denounce it and insist it's a bad thing now, is just fighting against the winds of change. I do think there could be regressive legislation sparked by large industries, but I do not think that will benefit artists in the long-run. Old catalogs and big companies will always get their way. I do not think it will take anything away from artists with catalogs of songs. But more music of course is more competition. And establishments hate more competition. I think AI will create opportunities for artists, in the long run, to be independent of larger companies and THAT will definitely shake the music industry, though those companies might also do what you suggest, buy in, and that would be bad for the artists and the technology, I think, in the long-run. But traditional thinking will attempt to stagnate creativity and create comfort for a while, no doubt. I think there is going to be plenty of new opportunity for a broader range of artists, if this technology is allowed to flourish. I think a far sighted label would collaborate with these tools to find new songs, new catalog and new artists, for the long-run. Artist's songs could be licensed for revenue sharing, not imposed on artists. Your model suggests it's GOOD for large companies to come in and impose themselves and steal the revenue for all future music. That seems quite draconian and not really as pro artist as you might think it sounds. I think that would be awful and all about killing the technology and opportunities for creativity. Also, painters hated photography. I think this is the new photography. The theater industry hated the idea of movies, no doubt, at some point. Then they saw the opportunity for both. Live music, and artists who actually can perform the magic of music, are not going anywhere. More likely than not, they will take the songs of new writers using AI, and do even more amazing things with it. And like art, I am also not sure that learning to play music by listening to it, whether AI or human, is entirely different. Sounding like an artist is one thing, but, for instance, Udio and Suno don't allow you to type in an artists name. I tried to write in a birthday song for a family celebration. One of the birthday celebrants was named "Trina".... but there is an artist named Trina. So it would not accept creating a song about "Trina". It's pretty strict.
@@legalsage just be careful. I understand your argument for thinking you’ll own the music, but I’m not sure the law will back you up.
Wow. People are so butt-hurt over this technology! 😂🤣 You are fighting against the future. It's like when people complained about Jazz music, because it wasn't classical enough. Or about mp3s for a variety of reasons. Learn to UTILIZE the technology to improve your craft.
Did you watch the video?
@@SyncMyMusic Yeah, just finished it, but I saw where you were going. The thing about it is, the AI is just LISTENING. Is that illegal? It's creativity, sure, is based on what it hears, but then, so is yours! Of course! It's kind-hearted of you to try to represent "All Music Creators", but explain to me how you have the right to demand payment for Ella Fitzgerald, or The Stray Cats, or Mozart, for that matter. I mean, that ship sailed long ago, and the data's out there. There may be legal owners of certain catalogs of work who may want to opt out of all of this, but don't forget the downside: you're killing the influence of your music for the future. If you truly want NO illegal use of any of your works, keep them locked up in a cabinet in your home, and never release them, because God forbid someone should hear them and be influenced by them.
I do see where you are coming from. I thought it was a crock of s*** when Apple leveraged the storage capacity of the iPod to help people - let's face it - steal a LOT of music and carry it around with them. The "compromise" solution (remember) was that tracks would cost $1, instead of buying an album for $15 with one good track on it. That solution worked OK, but Apple still monetized a technology that wasn't especially friendly to record companies and artists.
But music always evolves. Nor does it keep you from making music exactly how you want to. Continue using your Neve console and your 3M 79 1/2" 4-track tape recorder, your giant patch-bays and Neumann mics! Nothing stopping you if you enjoy it!! But regardless, music distribution, and now production has been democratized. There will still be ways to make money in music. I just may not be the SAME ways as in the past. Best of luck!
@@SyncMyMusicwanna keep ur job? be creative then, musicians should just stop making these cheap lazy music and go back to be creative.
pop and electronic music nowdays is just the same note spammed hundreds of times with little difference here and there, not to sound like a boomer but in the 2000s it had actual melodies yet.
@@jktech2117 Yes, I notice these AI generated songs are very bland, 'stock' sounding 4 chord tunes, for the most part. Not a lot of wildly creative tunes. At least not yet.
@@johnpinion8033... Actually artists and musicians evolve with life. And it reflects thru their music. But music is still only music. What is in question here is the production of music. And it boils down to whether you're a purist as a producer who has worked on your craft and enjoy the process or if you're well....kind of lazy, inexperienced or trying to catch a genie in a bottle and make a quick hit. And good luck with that. Because purist or quick fix producer they're both still capable of producing some BS. AI included 😂
If we want to release a song in Udio, and the vocals sound obviously like someone popular. It's our gamble that we don't get sued by the the popular singer , unless we alter the singer's vocals.
Udio wants us to clear the copyright with the data owners... Which is complete BS since we are never guaranteed what copyright material Udio actually used.
but at the same time, it's obvious my latest song sounds exactly like Beyonce...
In the fine print Udio writes:
------------------------------------------------------------------------------
6.3.4
By posting or submitting Your Content through the Services, you represent and warrant that you have, or have obtained, all rights, licenses, consents, permissions, power and/or authority necessary to post or submit Your Content and to grant the rights granted herein for Your Content. You agree that Your Content will not contain material subject to copyright or other proprietary rights owned or controlled by any third party, unless you have the necessary permission or are otherwise legally entitled to post the material and to grant us the license described above.
Interesting how they expect the user to obtain all rights from CR owners, but they don't disclose that THEY obtained all CR rights from the music they trained on. 🤔