Don't make too much popcorn, though, this is probably the labels pushing for a settlement where they get access to data or models in exchange for some set fee per work trained on.
Good guess, but there is a reason why they want a jury trial - polling shows 4 in 5 ppl support the labels. Why, i have no real idea, but that is a potential freight train coming. Jury trials that turn on legal matters are a dangerous tangle to get into, especially if the other side has more expensive counsel.
At this moment, there is no investor thinking, "I should invest in a record label," so we already know that the lawsuit will not proceed. Suno and Udio will make record labels disappear, or they will have to cede copyright to Suno and Udio to survive in some way. Because even if Suno did not use copyrighted material as the record labels claim, the ability to create thousands of songs per second makes it impossible for record labels with their artists to compete with even 0.1% of the market. Right now, there are companies, even advertising campaigns, already using AI-created music at zero cost.
Thanks for the look at the lawsuit. Was cool seeing/hearing the actual bits of music referenced in the suit. I'd be interested in your opinion on what I've done with my lyrics/melodies in conjunction with Udio. I've been a serious writer for almost 20yrs. I'm a novelist (four), short story writer (lots for family/friends), and a writer of 5-600 songs/poems (mostly for family and friends). I'm old, health's not good. Decided I wanted to hear my songs in more than my crappy a cappella. Spent the last two years going to bars, open mics, smaller gigs, talking with a lot of really nice, talented people and trying to get someone to collaborate with me and my lyrics, while also sending stuff to more recognized artists and their reps. Nothing. Not spending my last few years trying to learn to play music, dying and never hearing my stuff in a more robust world of sound. So, after purposely staying away from AI (as an artist in the world of writing, I'm conflicted), I jumped into Udio a couple weeks ago. I take a lot of time, 20-30 prompts to find thirty seconds of something that comes close to the world I've created in my lyrics/songs and what bit of melody I came up with a cappella: the feel, sound, taste, emotion, plot, multi-layered meanings, etc. I've been stunned by some of the resulting music, as have friends who've been tortured with my a cappella versions the last 6 years. Most of my lyrics/melodies find a comfortable home in anything from folk to country, old time blues to electric blues, Chicago/London 60's blues. I can hear some of what the app was trained on, but because they're my lyrics, they are pretty unique. I think? Again, would love an opinion on if you think Udio is spitting out obviously derivative music. Edit* I took all but one of my 160 a cappella songs private on my U Tube when I decided to use AI. Cheers! David
Hi David, thanks for your comment and story. I think these tools are a fantastic use case for your situation. You are now able to hear creations of the musical ideas in your head and can now experience the joy of creating music like many others are blessed with. While you personally are not composing or creating any music, your lyrics are being given musical life thanks to these tools, just like crutches or other medical devices help the physically disabled. I love that you are able to do this, please post your music so others can hear your work! :)
@@YoPaulieMusic Thanks for the reply, Paul. Honestly, probably won't ever be comfortable using AI. But, as you point out, much like the hundreds of millions using things like Chat GPT for writing, those "medical devices" are hopefully helping millions of "physically disabled" writers find some new and exiting meaning to the thoughts in their heads. My stuffs up on U Tube. Cheers!
At this moment, there is no investor thinking, "I should invest in a record label," so we already know that the lawsuit will not proceed. Suno and Udio will make record labels disappear, or they will have to cede copyright to Suno and Udio to survive in some way. Because even if Suno did not use copyrighted material as the record labels claim, the ability to create thousands of songs per second makes it impossible for record labels with their artists to compete with even 0.1% of the market. Right now, there are companies, even advertising campaigns, already using AI-created music at zero cost.
"Udio will make record labels disappear, or they will have to cede copyright to Suno and Udio to survive in some way." They have stems and they can be used to make a much better large model music AI that lets you dig into songs. More artist friendly.
You are starting with copyrighted lyrics. The Suno song is already not yours clearly. If you use your own lyrics like I do, you own the lyrics without any argument.
The lawsuit is the first step towards copyright extinction; the record labels have already lost, and the only thing they are doing is publicly passing the torch to Suno and Udio. By filing the lawsuit, they gave Suno and Udio international publicity. If there were any doubts from people about these two, the record labels have brought them to the forefront. The lawsuit is really irrelevant, as nothing will come of it. At the end of the day, 2024 is when the great shift towards AI and the AI economy is taking place. Everything is going to change, and copyright will become public domain. Just as there is no copyright in AI art, it will not exist in any artistic aspect of the future.
Personally, I don't see any legal difference between training humans and training AI. No one creates art from a vacuum, they all learn from past masters and successful artists. Copyright has an explicit exemption for education. As for outlawing creating music that learns from a hit and duplicates the style, wouldn't that put all the bar bands out of business?
TBH, some record companies’ man created (copyrighted) songs are also sound like other copyrighted song ( lyric and melody) because Man also use other songs as reference.
I think the debate whether or not a photo (or recording) is real is an interesting one. It illustrates how much of reality is based on imagination, and how policing our imagination is wrapped up in a billion dollar industry
Thanks for this video. I asked the question online: Can song styles be copyrighted? No, styles of music, like genres, cannot be copyrighted. This is because of several limitations on copyright, including fair use, the idea/expression dichotomy, and the fact that not all original expression is copyrightable. For example, many bands in the same genre have similar sounds, and genres are defined by stylistic similarities.
Copyright will cease to exist; remember that AI art cannot be copyrighted, and everything that currently has copyright will lose it and will have to pass into the public domain. Record labels will disappear because their business model will become obsolete when AI reaches full expansion in the arts.
You think they've trained the neural network to not sample, but if that was the case the voice they created would exist separate from the lyrics. Will it have Frank Sinatra singing delta Blues? No. These are not exact copies, but are no different than a cover band doing their version, which is also copyrighted for commercial use.
The other thing is... since when has it been illegal for a tool to create a cover song? If you sing a song EXACTLY like the original and comp it all together in your DAW, the music industry can't sue the daw creator... lol If you distribute it, THEN you need to get the cover license. But other than that, it isn't illegal to produce....
The problem in your example is the AI is the one putting the parts together, making those decisions of harmony, melody tempo and rhythm, not you. Copyright requires human authorship. They cannot assign rights to an algorithm's output. This is no different than going to a drive-thru McDonald's and ordering a big Mac without pickles. You did not create the original recipe and you did not create the actual product, you merely ordered a variation of someone else's efforts.
I tend to not agree with Suno on that particular prospect. Parody won't hurt financial bottom line of the original art. Educational content won't hurt financial bottom line of the original art. But if you train a machine on sync music and then offer that machine for free, it will hurt earnings of authors of that sync music. Hence there's nothing fair about this use. You want a model capable of creating metal music? License tracks from metal bands or make your own metal music. Everything else is stealing, sorry.
Suno also says in the agreement that you sign that you are not allowed to put in lyrics of already copyrighted material. So, obviously this rule is being broken by the record companies since they logged in with an account and proceeded to put in copyrighted material. hmmm.. try putting in original expression of who you are and it will create original music around your words.
@@carcolevan7102 in the examples shown, there were no lyrics input, however the AI "created" copywritten lyrics, the same words in the same order, surprisingly using the same rhythm or melody. What a coincidence that the AI would imagine the same words, melody and voice.
@@tonyr.4778 It says on page 16 (paragraph 52) of the brief "Plaintiffs specified the lyrics for the output, so as to more easily surface the underlying melodic or rhythmic similarities with specific Copyrighted Recordings." In Paragraph 54 (discussing the "Johnny B. Goode" examples), it says, "Using the prompt, “1950s rock and roll, rhythm & blues, 12 bar blues, rockabilly, energetic male vocalist, singer guitarist” and the lyrics from the original, ...'" So these are certainly violations of the Suno terms of service, and at least in these cases, the Suno AI did not generate the copyrighted lyrics.
Thank you for your input. You are one with a clear thinking mind on the topic. I have been with UDIO for about two months now, and you cannot duplicate a singer's voice or their music exactly - who would really want to? And anyone that's worked with AI knows it's tricky to get a particular sound or image. And, like you said, any artist or performer or singer has learned from copyrighted material. To me, it's just a joke that they are going to these kinds of extremes in trying to stop AI music generators. Maybe, just maybe, it will push them to a new level of creativity of music and sound.
When I went to two music schools, we were given material to learn. Rarely was there examples of authentic sheet music of commercial music being taught. Most music had to be cleared for use or was designed by the school. Why the AI companies cut corners to train without proper consent? Fair use is an ambiguous concept when applied to teaching. The results are, of course, going to be a copy of what the AI learned. Distributing the output of its music to the public should have been stopped. In all fairness. It is a new and potentially innovative technology, I would use my discretion passing judgment against the AI companies. Perhaps it music is just a demonstration of its abilities. Not for sale, but for show.
The legal argument that a tool that can generate pre-existing works is therefore breaching copyright is absolutely laughable. A guy about 15 years ago or more already did a website that hosted every melody in existence through a generative recursive algorithm. Same shit as the Monkey typewriter website.
Funny the labels aren’t suing for reproduction, Suno is still micro sampling, whether it’s model generated output or not. Suno does not play any instruments. So the only way Suno can create a track is reproduce learnt recording data and piece together a new track. Surely the labels would investigate this angle, Suno would have to prove it’s ai is actuallt playing the instruments and recording the output..
No. You are not a maschine. You are human. Stop this "we are also inspired by listening, we learn the same way" crap. We have laws in place to make a fair play for HUMANS, and we agreed upon certain rules for humans. Maschine doesn't have any rights and no one cares how the maschine works and learns. It will take you years to learn to play or sing like your favorite artists, and then if you make a song that's too similar and infringes copyright, you will also get sued. The scalability of these maschines are beyond anything possible for any human and you cannot compare or treat it the same.
@@jvlbme yes you can. Again, you are human and it's a fair game. We shouldn't apply the same principle to the maschine in arts. It's absolutely not fair, and it only leads to extreme devaluation and flood of bad, fake music. Very distopian indeed. Why don't we put robots to compete in marathon vs humans? They are currently learning to walk and move similarly to how a human child learns it. They will surely be faster and more efficient.
@@jvlbme oh it's definitely not better. And it's not musical evolution in any way, shape or form. These generators don't do anything human cannot do or what hasn't been done before. They just do lots of average stuff that has already been done before, just very fast. It's not a new tool like electric guitar or synth was. I'm very worried about the future of music if people can't see the difference. It won't improve music in any way, just completely devalue it to the point where it becomes completely worthless due to extreme saturation, because no skill is required. Why would you want to automate arts and human expression? It's not some factory where you need to output million products per day, it shouldn't be about that. And no, I cannot seriously consider prompting genre and mood + some lyrics and getting the full song out of it as an human expression. It's a slot maschine, you don't know what you're gonna get, it's not your melody and you didn't come up with it.
It also doesn't help how many people are intentionally trying to poison the well with regard to AI because they see it as competition. Just SUPER dishonest takes and trying to frame it as "plagiarism with some extra steps" to get audiences to judge the music before they even hear it. Of course they won't win. As people find more music they like that was created with AI, they will slowly get over the dirty tactics.... but it will take a while for that to happen now.
Don't make too much popcorn, though, this is probably the labels pushing for a settlement where they get access to data or models in exchange for some set fee per work trained on.
Good guess, but there is a reason why they want a jury trial - polling shows 4 in 5 ppl support the labels. Why, i have no real idea, but that is a potential freight train coming. Jury trials that turn on legal matters are a dangerous tangle to get into, especially if the other side has more expensive counsel.
At this moment, there is no investor thinking, "I should invest in a record label," so we already know that the lawsuit will not proceed. Suno and Udio will make record labels disappear, or they will have to cede copyright to Suno and Udio to survive in some way.
Because even if Suno did not use copyrighted material as the record labels claim, the ability to create thousands of songs per second makes it impossible for record labels with their artists to compete with even 0.1% of the market.
Right now, there are companies, even advertising campaigns, already using AI-created music at zero cost.
Thanks for the look at the lawsuit. Was cool seeing/hearing the actual bits of music referenced in the suit.
I'd be interested in your opinion on what I've done with my lyrics/melodies in conjunction with Udio. I've been a serious writer for almost 20yrs.
I'm a novelist (four), short story writer (lots for family/friends), and a writer of 5-600 songs/poems (mostly for family and friends). I'm old, health's not good. Decided I wanted to hear my songs in more than my crappy a cappella. Spent the last two years going to bars, open mics, smaller gigs, talking with a lot of really nice, talented people and trying to get someone to collaborate with me and my lyrics, while also sending stuff to more recognized artists and their reps. Nothing.
Not spending my last few years trying to learn to play music, dying and never hearing my stuff in a more robust world of sound. So, after purposely staying away from AI (as an artist in the world of writing, I'm conflicted), I jumped into Udio a couple weeks ago.
I take a lot of time, 20-30 prompts to find thirty seconds of something that comes close to the world I've created in my lyrics/songs and what bit of melody I came up with a cappella: the feel, sound, taste, emotion, plot, multi-layered meanings, etc. I've been stunned by some of the resulting music, as have friends who've been tortured with my a cappella versions the last 6 years.
Most of my lyrics/melodies find a comfortable home in anything from folk to country, old time blues to electric blues, Chicago/London 60's blues. I can hear some of what the app was trained on, but because they're my lyrics, they are pretty unique. I think? Again, would love an opinion on if you think Udio is spitting out obviously derivative music.
Edit* I took all but one of my 160 a cappella songs private on my U Tube when I decided to use AI.
Cheers! David
Hi David, thanks for your comment and story. I think these tools are a fantastic use case for your situation. You are now able to hear creations of the musical ideas in your head and can now experience the joy of creating music like many others are blessed with. While you personally are not composing or creating any music, your lyrics are being given musical life thanks to these tools, just like crutches or other medical devices help the physically disabled. I love that you are able to do this, please post your music so others can hear your work! :)
@@YoPaulieMusic Thanks for the reply, Paul. Honestly, probably won't ever be comfortable using AI. But, as you point out, much like the hundreds of millions using things like Chat GPT for writing, those "medical devices" are hopefully helping millions of "physically disabled" writers find some new and exiting meaning to the thoughts in their heads. My stuffs up on U Tube. Cheers!
At this moment, there is no investor thinking, "I should invest in a record label," so we already know that the lawsuit will not proceed. Suno and Udio will make record labels disappear, or they will have to cede copyright to Suno and Udio to survive in some way.
Because even if Suno did not use copyrighted material as the record labels claim, the ability to create thousands of songs per second makes it impossible for record labels with their artists to compete with even 0.1% of the market.
Right now, there are companies, even advertising campaigns, already using AI-created music at zero cost.
"Udio will make record labels disappear, or they will have to cede copyright to Suno and Udio to survive in some way."
They have stems and they can be used to make a much better large model music AI that lets you dig into songs. More artist friendly.
You are starting with copyrighted lyrics. The Suno song is already not yours clearly. If you use your own lyrics like I do, you own the lyrics without any argument.
The lawsuit is the first step towards copyright extinction; the record labels have already lost, and the only thing they are doing is publicly passing the torch to Suno and Udio. By filing the lawsuit, they gave Suno and Udio international publicity. If there were any doubts from people about these two, the record labels have brought them to the forefront. The lawsuit is really irrelevant, as nothing will come of it. At the end of the day, 2024 is when the great shift towards AI and the AI economy is taking place. Everything is going to change, and copyright will become public domain. Just as there is no copyright in AI art, it will not exist in any artistic aspect of the future.
Personally, I don't see any legal difference between training humans and training AI. No one creates art from a vacuum, they all learn from past masters and successful artists. Copyright has an explicit exemption for education. As for outlawing creating music that learns from a hit and duplicates the style, wouldn't that put all the bar bands out of business?
TBH, some record companies’ man created (copyrighted) songs are also sound like other copyrighted song ( lyric and melody) because Man also use other songs as reference.
I think the debate whether or not a photo (or recording) is real is an interesting one. It illustrates how much of reality is based on imagination, and how policing our imagination is wrapped up in a billion dollar industry
Thanks for this video. I asked the question online: Can song styles be copyrighted? No, styles of music, like genres, cannot be copyrighted. This is because of several limitations on copyright, including fair use, the idea/expression dichotomy, and the fact that not all original expression is copyrightable. For example, many bands in the same genre have similar sounds, and genres are defined by stylistic similarities.
Copyright will cease to exist; remember that AI art cannot be copyrighted, and everything that currently has copyright will lose it and will have to pass into the public domain. Record labels will disappear because their business model will become obsolete when AI reaches full expansion in the arts.
You think they've trained the neural network to not sample, but if that was the case the voice they created would exist separate from the lyrics. Will it have Frank Sinatra singing delta Blues? No. These are not exact copies, but are no different than a cover band doing their version, which is also copyrighted for commercial use.
I wonder how many of those examples of copyrighted song someone else did were cherry picked. And how long it took.
The other thing is... since when has it been illegal for a tool to create a cover song? If you sing a song EXACTLY like the original and comp it all together in your DAW, the music industry can't sue the daw creator... lol
If you distribute it, THEN you need to get the cover license. But other than that, it isn't illegal to produce....
The problem in your example is the AI is the one putting the parts together, making those decisions of harmony, melody tempo and rhythm, not you. Copyright requires human authorship. They cannot assign rights to an algorithm's output. This is no different than going to a drive-thru McDonald's and ordering a big Mac without pickles. You did not create the original recipe and you did not create the actual product, you merely ordered a variation of someone else's efforts.
They recently removed these audio files from the platform.
for like 170(maybe a bit more) songs it did take me like 230 pages in total of generated crap :D(the picked worked out songs included)
I tend to not agree with Suno on that particular prospect. Parody won't hurt financial bottom line of the original art. Educational content won't hurt financial bottom line of the original art. But if you train a machine on sync music and then offer that machine for free, it will hurt earnings of authors of that sync music. Hence there's nothing fair about this use. You want a model capable of creating metal music? License tracks from metal bands or make your own metal music. Everything else is stealing, sorry.
I thought the whole deal with Suno was that the devs were also musicians so they could train the thing on their own, disappointing
Suno also says in the agreement that you sign that you are not allowed to put in lyrics of already copyrighted material. So, obviously this rule is being broken by the record companies since they logged in with an account and proceeded to put in copyrighted material. hmmm.. try putting in original expression of who you are and it will create original music around your words.
I came here to post this point. Glad to see someone beat me to it.
@@carcolevan7102 in the examples shown, there were no lyrics input, however the AI "created" copywritten lyrics, the same words in the same order, surprisingly using the same rhythm or melody. What a coincidence that the AI would imagine the same words, melody and voice.
@@tonyr.4778 It says on page 16 (paragraph 52) of the brief "Plaintiffs specified the lyrics for the output, so as to more easily
surface the underlying melodic or rhythmic similarities with specific Copyrighted Recordings."
In Paragraph 54 (discussing the "Johnny B. Goode" examples), it says, "Using the
prompt, “1950s rock and roll, rhythm & blues, 12 bar blues, rockabilly, energetic male vocalist,
singer guitarist” and the lyrics from the original, ...'"
So these are certainly violations of the Suno terms of service, and at least in these cases, the Suno AI did not generate the copyrighted lyrics.
Thank you for your input. You are one with a clear thinking mind on the topic. I have been with UDIO for about two months now, and you cannot duplicate a singer's voice or their music exactly - who would really want to? And anyone that's worked with AI knows it's tricky to get a particular sound or image. And, like you said, any artist or performer or singer has learned from copyrighted material. To me, it's just a joke that they are going to these kinds of extremes in trying to stop AI music generators. Maybe, just maybe, it will push them to a new level of creativity of music and sound.
When I went to two music schools, we were given material to learn. Rarely was there examples of authentic sheet music of commercial music being taught. Most music had to be cleared for use or was designed by the school.
Why the AI companies cut corners to train without proper consent? Fair use is an ambiguous concept when applied to teaching.
The results are, of course, going to be a copy of what the AI learned. Distributing the output of its music to the public should have been stopped.
In all fairness. It is a new and potentially innovative technology, I would use my discretion passing judgment against the AI companies. Perhaps it music is just a demonstration of its abilities. Not for sale, but for show.
The legal argument that a tool that can generate pre-existing works is therefore breaching copyright is absolutely laughable. A guy about 15 years ago or more already did a website that hosted every melody in existence through a generative recursive algorithm.
Same shit as the Monkey typewriter website.
Funny the labels aren’t suing for reproduction, Suno is still micro sampling, whether it’s model generated output or not. Suno does not play any instruments. So the only way Suno can create a track is reproduce learnt recording data and piece together a new track. Surely the labels would investigate this angle, Suno would have to prove it’s ai is actuallt playing the instruments and recording the output..
So the music industry wants us to listen to their music, but if we do they could sue us for copyright infringement..?
No. You are not a maschine. You are human. Stop this "we are also inspired by listening, we learn the same way" crap. We have laws in place to make a fair play for HUMANS, and we agreed upon certain rules for humans. Maschine doesn't have any rights and no one cares how the maschine works and learns. It will take you years to learn to play or sing like your favorite artists, and then if you make a song that's too similar and infringes copyright, you will also get sued. The scalability of these maschines are beyond anything possible for any human and you cannot compare or treat it the same.
@@homeproject4108 What is your point? We are not allowed to learn too fast?
@@jvlbme yes you can. Again, you are human and it's a fair game. We shouldn't apply the same principle to the maschine in arts. It's absolutely not fair, and it only leads to extreme devaluation and flood of bad, fake music. Very distopian indeed. Why don't we put robots to compete in marathon vs humans? They are currently learning to walk and move similarly to how a human child learns it. They will surely be faster and more efficient.
@@homeproject4108 So you are just generally against doing things better?
@@jvlbme oh it's definitely not better. And it's not musical evolution in any way, shape or form. These generators don't do anything human cannot do or what hasn't been done before. They just do lots of average stuff that has already been done before, just very fast. It's not a new tool like electric guitar or synth was. I'm very worried about the future of music if people can't see the difference. It won't improve music in any way, just completely devalue it to the point where it becomes completely worthless due to extreme saturation, because no skill is required. Why would you want to automate arts and human expression? It's not some factory where you need to output million products per day, it shouldn't be about that. And no, I cannot seriously consider prompting genre and mood + some lyrics and getting the full song out of it as an human expression. It's a slot maschine, you don't know what you're gonna get, it's not your melody and you didn't come up with it.
It also doesn't help how many people are intentionally trying to poison the well with regard to AI because they see it as competition. Just SUPER dishonest takes and trying to frame it as "plagiarism with some extra steps" to get audiences to judge the music before they even hear it.
Of course they won't win. As people find more music they like that was created with AI, they will slowly get over the dirty tactics.... but it will take a while for that to happen now.