This got a lot of attention this week. I guess it's novel.
But the song sucked, as did the supposed AI contribution. I couldn't quite figure out why everyone used this as the example of how it's all over and AI will take over the world of music.
I thought the song was pretty catchy. As a first example of what AI can do in this space, I'm excited at the possibilities -- especially as someone who writes music but can't sing like the Weeknd or rap like Drake. I look at it less as something that's going to replace songwriting and more as another tool.
> I couldn't quite figure out why everyone used this as the example of how it's all over and AI will take over the world of music.
The bar for "AI" gets lower every day. I still think the ChatGPT reaction is something of a marketing psyop. Beyond a couple of novelty searches, I haven't noticed anybody using it as a daily driver.
That's fine if you feel that way, but millions of people enjoyed it. The comments on reuploads of it are full of people gushing over it, asking why it was removed, and being surprised that AI was involved in it.
I believe a big reason this song blew up on social media was that it was misleadingly hyped as an "AI-generated song". I expect many people thought that AI composed and recorded the (reasonably idiomatic) music and lyrics.
For the record, here's my understanding of how this "AI-generated song" was made:
1. A human wrote the lyrics
2. A human recorded the beat
3. A human recorded themselves rapping
4. An "AI voice filter" made the recorded voice sound like Drake / the Weeknd.
IMO, the real tragedy here is not some threat to Drake (he’ll be fine) but that some dude who wrote an original song had to pretend it was written by AI to get it heard.
That's just how the music industry has become. As an indie musician myself, it's incredible hard to get heard. So while I don't agree with what was done in this case, I can understand why.
Is it though? This never been easier just to upload something and get some amount of listeners. If you lived a bit rural and was a poor teenager in the 60s you could forget ever getting anyone to listen to your music, if you even had the instruments to play it. Nobody is owed a big audience.
And now you can create, distribute, and market you song for free to a world wide audience. The weirdest stuff can get a thousand plays on soundcloud. Much more than you'll get from being played at any random bar or club. Also you would have to make a vinyl print of you song to provide it to a dj.
Thousands of soundcloud plays means one person one listen and this is over years. Many of those are short listens / skips. In a bar you could have a few hundred people all listening at once passively the entire song while associating with good times drinking.
Sharing a song with a group will create memories and stronger associations
Recording your music to any level of quality they would play on the radio back then was expensive. It was really time consuming to lay down tracks and edit them. Like manually cutting tape, etc. Today anyone can make a professional sounding recording at home.
Still possible with many small local radio stations. Mail in a CD with your shit and a genre label and odds are someone will listen to a track or two to decide if it has a place on their show.
This is what a pure meritocracy looks like. There's no more tricks, there's no more hacks, it's just a tight few metrics that you're either really really good at, or you aren't. It's the same with the rest of the country. Turns out a pure meritocracy leaves a lot of losers because there can only be so many winners.
It's never been easier to publish a blog available to billions of people that no one ever reads.
In the 60s you would travel to a music centre. A place where others performing music went. You leveraged a community. If someone in your community made it, it shone a light for everyone and created a scene. People helped each other make it.
That small community with those opportunities have been replaced by centralization without the community or scene multipliers.
>IMO, the real tragedy here is ... that some dude who wrote an original song had to pretend it was written by AI to get it heard.
Genuinely curious to hear why this is a tragedy. Does everyone who puts out anything original deserve to have it heard by a wider audience?
Edit: Asking as someone who has always been heavily involved in various aspects of music and has watched a massive increase in the volume of music being put out now that it's significantly easier to produce on your own. This is a very complicated topic.
To me it's pretty easy to grasp that it's disappointing the artist had to mask as two of the most successful artists of all time to get heard, while also understanding that it's the easier route to do that. It's easier to be a ghost writer than it is to push your own brand, that sort of situation.
I don't really understand the objection here. Must we couch everything with "of course in the grand scheme of the world and poverty and famine it isn't a big deal, but..."
Be grateful - at least you weren't born in an AI-controlled human farm created to inflict as much suffering onto humans as possible. Poverty and famine are only a reflection of your privilege as a free human.
That was a distracting side comment on my part. My real objection is that it's silly to characterize the artist as a victim. They had the opportunity to leverage other people's work, both technical and musical, to get eyeballs they likely never would have otherwise gotten.
To me, saying it's "disappointing that they had to" feels wrong almost to the point of being offensive.
Well...of course? If I pretend my blog post is written by Bill Gates, more people will read it. This is basic human nature.
We can try to prevent plagiarism and improve our detection of it, but there's nothing wrong or disappointing with the fact that people prefer to listen to Bill Gates than a random blogger.
> Well...of course? If I pretend my blog post is written by Bill Gates, more people will read it. This is basic human nature.
I think the tragedy is that the attention doesn't follow quality or value, it follows fame and gimmicks. In your example, Bill Gates's stupidest brain farts will get 1000x more positive attention that an objectively-better tour-de-force that almost anyone else posts.
> I think the tragedy is that the attention doesn't follow quality or value, it follows fame and gimmicks.
It is, but also... welcome to life. This is how it works. What determines success is very often more about things other than the quality of the product.
For evidence, just look at the most popular products at the store. Very rarely are they at the top of the quality scale.
My guess is some ppl get tired of the moaning about things you can’t change. So short circuiting the topic is avoiding the annoyance and maybe an antivirus to prevent one’s own thoughts from spiraling negatively.
Exactly that. I already went down that spiral many times and the result is nothing but sadness about the many injustices of the world. But ultimately it's out of my hand. It's out of your hand. The ball is in no one's court.
Focus on the better aspects of life. Take care of your loved ones. Do your best to improve your local community. Contribute to a better state of the world, no matter how small the contribution may be. Pray to a higher being if you believe in one. That's ultimately all you can do.
Attention cannot follow quality because you have to pay attention to determine quality. So what attention follows is a heuristic of quality: has this person exhibited quality in the past. That's not a tragedy or sad or anything. In fact, that's normal in a marketplace where you have to pay before you benefit. Rarely, something novel will arise and another heuristic of quality will be present: virality.
I think this is actually perfect and there is no tragedy.
These heuristics for quality being discussed and alluded to, especially over time, suck far worse than just sticking to personal recommendations and keeping your tastes disconnected the internet. It is not perfect; culture on the internet is just bad these day across many dimensions and getting worse. Next I expect to see some try to equate profitable with quality.
No. It has nothing to do with the Internet. It is simply a property that you cannot describe a thing accurately without observing the thing. You can only use heuristics. One such heuristic is personal recommendations (which I would place in the viral transmission mode). Another is creator.
I'm not disputing that claim of yours, but you're missing the point. I'm saying the heuristics for quality you presented, when taken in the context of our internet, tend to press culture towards shit quality, or as some parent comment mentioned "fame and gimmicks."
For the record, the type of personal recommendation I'm talking about cannot be viral, in any useful context for the word. I'm talking about recommendations delivered one on one and as an ongoing conversation over many moons; Not recommendations that I see in my Facebook Feed as a Facebook Post from my Facebook Friends to their Facebook Followers. These mediums are insufficient and grow worse, IMO.
It's really hard to evaluate everything by yourself. If you wanted to listen to some good music, you'd find something you like much faster by scrolling through the top 1000 most listened songs than a random shuffle of the maybe billions of songs that have ever existed. "A lot of other people like this" is a pretty good filter for things you'll probably like given how similar we all are (especially when we're stewing in the same culture).
The venue it's published on also matters... If music is published (for profit) on official services like iTunes or Spotify and it isn't exactly clear that it's AI generated content, then there is a certain level of misrepresentation involved in it all...
If the music was posted to somewhere that it's normal to expect non-serious content (like Reddit or social media) but still properly credited as Ai generated on a non-profit basis (standard practice of course), then that would have been far less offensive.
> We can try to prevent plagiarism and improve our detection of it, but there's nothing wrong or disappointing with the fact that people prefer to listen to Bill Gates than a random blogger.
Except when Mr. Gates and the random blogger are both out of their areas of expertise, people still believe Mr. Gates because he was successful elsewhere instead of applying proper skepticism.
Except they have good reason to apply less skepticism because they know Mr. Gates has a lot more to lose if he's caught making things up or lying or messing with folks on the record.
Whereas random bloggers are fly-by-night operations in comparison.
> Does everyone who puts out anything original deserve to have it heard by a wider audience?
Yes! This is a failure of capitalism coalescing and centralizing under fewer and fewer big profit centers.
There are many, many creators putting out work as good (if not better) that deserve discovery. (NB: not everything is worthy of "discovery", like some early experimental learning work, but there's definitely undiscovered well-polished work out there.) Culture can be made by the masses, not by the few, but profit sharks don't let that happen.
Just to clarify, I agree with the general sentiment of your comment.
My concern, or confusion, or lack of understanding, or ignorance, or whatever you wanna call it, is less about the mechanisms of capitalism controlling distribution and moreso the sheer volume of music available. I posit that there is more well-polished music being released on a daily basis than people are capable of listening to, or that we even mathematically have the available time, as a whole, to fit into our lives. Even if we shifted the controls of distribution and such, there seems to be far more content than there are ears to hear the content, and I just don't get how we can ever expect everything worth it's weight in notes to be heard.
I understand the desire, as an artist myself, to be heard, but I also know that I'm one of millions of other people feeling the same way. If everyone's creating (and there are definitely far more creators than ever before), who's listening?
No, you're not. You're dealing with principles of communism here. They're inherently nonsensical.
Note the standard of justice being used. It's not, "impersonating someone to use them as a stepping stone is wrong, no matter who does it, or who is the victim".
The standard is, the rich and powerful don't "deserve" equal protection from the law, the real victim is the artist who isn't being recognized because the industry isn't playing "fair". So it's ok if the underdog breaks the law, because the world is stacked against him.
Yeah, and basket ball isn't fair because I'm fat and uncoordinated, and math isn't fair because I'm stupid and don't want to study.
Ironically, these were the same people that said, "Nobody deserves a platform to spread the 'wrong' ideas." when it came to censorship on Twitter, and had the gall to call it "accountability" when people were banned for wrong think about vaccines.
If you're scratching your head and saying, "Wait, what am I missing? That makes no sense! Ten seconds ago it was the rich and powerful censoring the poor underdog, and that was bad. But now big pharma backed censorship is a good thing?" -- it's because you're the sane and logical person, and they're the ones without a consistent worldview.
> You're dealing with principles of communism here.
I don't know, the reasoning seems simpler than any school of thought can offer. Humans are bound to 14-16 waking hours per day. A large portion of those hours are taken up by obligations like work or chores. That leaves a set amount of time to consume media created by other humans. Even if more free time could be had, there will always be a hard biological limit of about 14 waking hours to consume things.
But more and more humans are being born and learning the arts, so combined with instant global communication they will inevitably produce and distribute way more than can be crammed into those hours. Look at a random streamer's Let's Play history for an example. If they play a longer JRPG then they can leave a trail of many hours-long videos per week depending on how long their sessions last. Sometimes that's to the exclusion of every other game in existence. It's a bit sobering if you think about it.
I read a book called The Plenitude by Rich Gold who discusses some of these topics. He doesn't go as far to give a fully workable solution to the problem but one of his suggestions is to challenge creators to limit themselves to 5 "big creations" over their entire life, and focus on honing those creations to the greatest extent. It's an interesting book and I'd recommend it.
>I read a book called The Plenitude by Rich Gold who discusses some of these topics. He doesn't go as far to give a fully workable solution to the problem but one of his suggestions is to challenge creators to limit themselves to 5 "big creations" over their entire life, and focus on honing those creations to the greatest extent. It's an interesting book and I'd recommend it.
Ooh, this sounds interesting. Thanks for the suggestion!
While I like the spirit behind Rich Gold’s suggestion here, I think he may be overlooking the fact that for a large number of artists the practice of quickly iterating and “shipping” songs is a fruitful one. Not every creation is going to be solid gold, but overall this mindset is a valid one for generating work with artistic value.
They "deserve" discovery? Ever since playback devices for recorded audio have become widely available there have been far more talented musicians than listeners could make any use of. An almost infinite oversupply, that's the price musicians pay, if they want or not, for the technical/physical ability to be heard by more people than they can fit in a room, and even at times when they don't actually play. The world does not owe attention to a musician, no matter how talented. If a musician is getting heard, good for them, if not, that's life.
This is totally backwards. Anyone can upload music and have it heard around the world. As another commenter stated, the problem for artists is oversupply. How do you propose capitalism or any economic system will make the situation better? What do "profit sharks" have to do with this?
> There are many, many creators putting out work as good (if not better) that deserve discovery.
With music, specifically, this has been one of the best impacts of the internet. As long as you avoid services that curate (streaming services, etc.), as a music lover there is a vast world of excellent music at your fingertips that you wouldn't ever have had the chance to hear otherwise.
As an artist, you can develop an audience from the pool of the whole world. You can make a decent living with a pool of a few thousand fans -- and finding a few thousand fans from the set of the Earth's population is a much more feasible thing than before because of the internet.
I'm someone that regularly releases original music and the small following that AI have genuinely likes the music, but it's an increasing hurdle to get new people just to listen, even when it's free, because of the absolute flood of phony music online. I don't resort to spending a lot of money on promo (which is the normally competitive move) because there is no guarantee that that money won't simply disappear without results, because these days platforms scam musicians as much as record labels do.
Not much online is more perilous and rife with scams than trying to make a name for yourself in the music industry. I think people should be given multiple opportunities to be heard by platforms, but now platforms don't do it unless artists pay them to be heard because of the shear volume of fraud accounts uploading fake and low effort music to streaming services.
TikTok shows everything from videos of people cooking eggs to people trimming cow hoofs.. Despite the company being built around music, very little choice is offered in terms of browsing music, and the platform regularly punishes unknown musicians by limiting their views and by imposing complex copyright control on original music to prevent it from being discovered. The majority of the music streamed on the platform funnels royalties for most of the videos on the platforms to already wealthy individuals, scammers, and to companies associated with music streaming.
I think that Independent musicians and labels are putting out music better than major industry now in most cases, yet because of that competition, major label and industry forces work to suppress indie music through lobbying platforms and government for control, and they also throw money around to make competition against them extra hard for musicians that aren't profitable. Tons of independent musicians are operating at a financial loss, and simply having a chance to be heard would make it worthwhile.
So yes, I think that everyone authentic should get as many chances to be heard, because it's an honor for platforms to get new authentic indie music from people that could upload it elsewhere instead, and platforms should be ritually meticulous in promoting all of the authentic independent artists on these platforms in support of music growth, because that's a far better prospect for long term success than just lining pockets of ancient and withering music label executives.
If he did everything he could to make it sound like Drake or The Weeknd musically and lyrically then it wasn't original and didn't deserve to get noticed much.
One less clone if a world of clones isn't a tragedy.
> IMO, the real tragedy here is not some threat to Drake (he’ll be fine)
Man, what a skewed version of justice this is. "He'll be fine." The REAL victim is this unnoticed artist? Give me a break. What he did impersonating another artist was wrong.
There's this perverse notion that has infused our society where it's ok to rob from a Target or Walmart because it doesn't really "hurt" them. "They have insurance. They'll be fine!"
That standard of justice will lead society to a very VERY bad place if you follow it.
Allowing massive conglomerates to exist has already led society to a VERY bad place already.
Anyways, besides that, impersonation for personal gain or to cause disadvantage is already criminal. If Drake wants to do something about that, he can.
I agree that it's not okay to rob the wealthy simply because they are wealthy.
But this case is more nuanced. A compelling argument can be made against the very idea of robbery of IP. I think the vast majority of "intellectual property" should not, in fact, be anyone's property, and therefore no one can be robbed of it.
Who is harmed and who benefits in the situation where Drake has a lock and key on his IP?
Consumers are harmed --- they clearly enjoy new Drake music regardless of whether or not it was written and sung by Drake, but this is not allowed.
The "AI artist" is harmed --- they could have created something valuable in the world, but this is not allowed.
Drake artificially benefits from a regulatory monopoly. Everyone is strictly worse off because of arbitrary, draconian rules.
A much better example is drug patents. Literally millions of lives could be saved --- and at a lower cost --- if patents didn't exist. Patents and intellectual property are just state-sanctioned monopolies where everyone is harmed and coerced to pay rent to some random asshole.
But is that true? If I were a gifted mimic, I couldn't go out and make a song called "Sad in Toronto", even if I was very clear about not actually being Drake?
A gifted mimic is not a mechanical reproduction. The right of publicity doesn’t protect you from people with similar appearances or voices, it does protect (to a certain extent) against uses of mechanical reproduction of your appearance or voice.
Because it is separate from copyright, US federal statutory fair use considerations don’t apply to this right (I’m not sure to what extent Constitutional fair use might apply to it; statutory fair use in copyright largely recapitulates what were found to be Constitutional limits on the copyright power stemming from the First Amendment, I don’t know if it has ever been considered the extent to which the First Amendment, applied ot the states through the 14th, applies similar restriction on the right of publicity, which is in the US, to the extent it exists, a state-law intellectual property right.)
The op is extremely certain that they have interpreted this law correctly, but the fact of the matter is that a case like the one you describe has never gone through US courts, so we do not know. We are likely to have a court case sometime in the next few years it seems, who knows how long it'll take to get an answer though.
There was a very important and similar case though where Bette Midler sued Ford over inappropriate use of "her" voice performed by an impersonator.
The context seems important in that case though. Ford had the impersonator sing a song that she herself released! (They had acquired copyright release for that part) So there's no reasonable claim that Bette Midler was not the one they were impersonating.
If you sang a song you wrote in a voice extremely similar to Drake's without mentioning Drake in any way, then that seems very different to me.
If it's clearly qualified that this is a fabrication of his voice, then what right does he have to control that? He's not being harmed. He's not being misrepresented.
I think he is being misrepresented in the song itself. People hearing it will think it's him. Whether or not that's called out in the marketing doesn't seem relevant.
As to harm, I don't know if he's being harmed or not -- but to my mind, that's not an important point. If, as I believe, it's wrong to produce works that show you doing and saying things you didn't do (without your permission), then it's wrong whether or not you were actually harmed by it. And what is harm, anyway? Being embarrassed is a kind of harm.
I think actually if there's an argument for restricting Drake-alikes, it needs to be said.
From my POV as long as you make it clear it isn't actually him, you should be able to do it. What's next, I can't do a parody of Drake either? Should we ban the Nixon-moonlanding-crash video?
Nobody has a unique voice anyway, I'm sure one of those 8B people sounds a lot like Drake. Does that other guy get to decide what songs Drake sings? Of course not, having a specific set of traits should not give you rights over it.
Does Arnel Pineda, who sounds a heck of a lot like Steve Perry, owe Steve a cut of whatever he makes with the band? (Aside from the fact that obviously there's contracts in place.) He's singing his ass off every time he's on stage, just like Steve did when he was the front man for Journey.
In the end this is an argument for the already powerful to stop the newcomer. There's no reason some dude shouldn't be able to do the same work as some other dude, as long as he doesn't pretend to actually be that other guy.
There are cases where using someone's voice can harm them, sure. That doesn't mean we should give everyone carte blanche jurisdiction over what can be done with their ideas or their voice.
I think the onus is on the speaker/singer to prove they are being materially harmed by a fabrication of their voice. In many cases, this could be simple: "That fabrication is being used to misrepresent my beliefs AND it's causing me harm." Fair enough. But is someone making a Drake song causing Drake harm? No. He could argue loss of income, but that would be shaky. He didn't come up with the lyrics. He didn't come up with the beat. He didn't even actually sing it. If it's so easy to rip off his voice to create a hit song, then is he actually still losing income? Why doesn't he come up with the hit song?
Yeah but identity theft can be dealt with without giving people rent-seeking monopolies. Everyone is worse off except Drake and the argument that he was materially harmed is very shaky at best.
> IMO, the real tragedy here is not some threat to Drake (he’ll be fine) but that some dude who wrote an original song had to pretend it was written by AI to get it heard.
The lyrics are semi-auto-biographical about The Weeknd and Drake. The guy who wrote this didn't date Selena Gomez. It's pretty uninteresting as a song if it's not from The Weeknd and Drake.
Not quite - the voices afiak are pure AI with the tuning, pitch, etc done in post. Essentially treating the vocal signature of someone as if it was an instrument. And pretty much everyone knew the beat and lyrics weren't AI generated, at least from what I've seen. This is going viral after there was a viral trend of people replacing songs with Kanye West AI, so people knew what was happening.
The impressive part is that the voices could be very convincing. That's what made it shareable.
just wait until the music industry elites start to recognize this...
This was predicted (as so many other things have) in anime...
Recall JEM? the soft-core-USA-version of anime?
but so many things have been predicted/projected through anime (art)... the cyberpunk future of holographic AI media pieces has been a long trope in anime...
at some point, youth will not care if its AI (EDM? YO!) and will just accept deepfake AI generated content along with AI generated sexually apealing rockstars...
What was not predicted in anime, but will become a reality is the personalization of the 'messenger' where we listen to the same song, but mine is personalized to me, and you, yours. so if, say, female, the song may be talored to being delivered to you differently than to male... however, that doesnt mean the delivery is based on gender... there will be subtle differences in timbre tone and rythm based on you....
> the personalization of the 'messenger' where we listen to the same song, but mine is personalized to me, and you
Dynamically generated, personalized content already exists in the form of video games. Yet it is not notable when an NPC remarks on my unique gear, achievements, etc.
Technology may eventually enable easier remixing of art to personal tastes. However, this has also existed as long as art itself.
What people look for in the artist is their uniquely curated tastes, human connection to the person themselves, and a sense of community with fellow fans.
A machine that creates everything for everyone can provide none of these. Its output will be products, not art. It will have customers, not fans.
> What people look for in the artist is their uniquely curated tastes, human connection to the person themselves, and a sense of community with fellow fans.
Some people do, some people don't. There's nothing obligating an anime/SNS art hoarder to think this way, and if all they ever do is consume the end product without giving a thought to how the artist laid down the individual strokes that made the picture or how the artist's circle of friends and life story influenced their style, there's no consequences. The artist and other people with a similar mindset are unable to criticize these consumers because their behavior is invisible. When AI artists started appearing with art featuring blatantly copied styles, they did finally become open to such criticism.
I have a feeling the nature of endless feeds with a practically infinite amount of art to consume has conditioned a lot of consumers to never have think about the human artist. They can just look at the stash in their Pictures folder and think "that's a cute X" divorced from the artistic context that's been integral in earlier art history. The consequences of this are only starting to become apparent right now. But I think if there were nothing like Stable Diffusion then one would be unable to tell apart the people who would be open to copying artists' styles with a hypothetical AI program to-be-invented in the future from people who wouldn't. They'd all just appear to be the same breed of fans. It was the invention of the technology and its spread that revealed those preferences.
> but will become a reality is the personalization of the 'messenger' where we listen to the same song, but mine is personalized to me, and you, yours.
Which would totally destroy one of the things that makes music powerful -- it provides a shared, deep experience. Not to mention that it all but eliminates another purpose for music: to communicate to people.
The more use cases I hear about AI, the more I think that economic devastation isn't the only thing we're flirting with. We're also flirting with destruction of society as we erase people and meaningful societal bonding.
Please note that, obviously I know what you mean (and what I mean).. and if it wasnt clear - I am NOT a fan of the dystopian-cyberpunk-future..
For fucks' sake, I am guilty of helping the world get to the current dystopian shitshow we have going on today (as is 99% of HN, but especially us from the 1980s)
Do you have any reason to believe people thought that? Everyone I talked to thought it was a human generated song with an autotune like drake ai filter on the lyrics.
Yes, many people assumed this. Here's the first example I found (scroll through the thread for many more): https://twitter.com/Russ_Wilcox/status/1647602468623114240 (Note that this guy was smart enough to ask the question - imagine how many people saw "AI-generated song" and just uncritically assumed a computer spat out the whole thing, chatGPT-style)
a few days EARLIER this was released in Israel by KAN (television chanel)..
I know you probably can't speak hebrew but here is a song made (vocals/lyrics/music) by AI, of two dead singers who are very famous in Israel (Zohar Argov and Ofra Haza).. It is convincing
> I believe a big reason this song blew up on social media was that it was misleadingly hyped as an "AI-generated song". I expect many people thought that AI composed and recorded the (reasonably idiomatic) music and lyrics.
IIRC, I read that in the 70s many people thought Popcorn by Hot Butter was also written by a computer, but it was just played on (analog) synthesizers.
I produce music but can’t sing to save my life. If I had access to an on-demand pitch perfect AI singer, I would likely release way more of my music to the public.
The thing with musicians are that there are so many talented musicians out there who can straight up show up and sight read a complex piece they never played before and put it on a record. Many who will work for free or close to it as a result of this saturation of talent. You could probably go on instagram and book a singer for your work by the afternoon.
That depends on if its faster to come up with the prompts to sing the song you are imagining in your head as a producer versus just hiring a cheap singer. Sometimes yes sometimes no I’d expect is the answer to this question, and given uncertainty some people do pay extra for certainty. Musicians pay extra for marginal quality gains too. If it was all just a race to the bottom all fender and gibsons would be made in China because no one cares, but thats not the case. American models that are sonically almost the same sometimes as a cheaper made in japan or thailand model command premiums due to the perceived quality.
The story is that there was a program, Racter, which allegedly created a book titled "The Policeman's Beard Is Half-Constructed" back in the 1980s. The book was released and was a hit primarily because of that AI connection, and a version of Racter was also sold. However, the Racter that got released could never have composed the book, leading me to this post:
> The least hypothesis is that the "good" version of Racter that supposedly composed the novel never existed, and that it was all a hoax made up by someone who wanted to bask in some publicity and sell the Racter that actually got released. As you say, the output is suspiciously good for something they insisted was accomplished on a home computer of the era and, oh no, they can't release the software that did it for Reasons... yeah, pull the other one, it's got bells on.
The tune should have never been uploaded to begin with... I remix artists, but use their actual vocals, and never release remixes that aren't properly cleared...
Putting out unauthorized work by others (for profit), especially when it emulates that they actually created the product, can harm their reputation. Putting out unauthorized work from musicians (for profit) often represents that the work was cleared... It's a move to either generate attention for the tool that they used, or for the emulation services they provide as "musicians", or even worse it could be a scam to make money off of putting out counterfeit music... I guess whoever was behind it (and now there will likely be even more of this) figures the legal issues that arise are worth the risk in seeking whatever profit may come from breaking common rules of music production such as this.
It's troubling to see this happen as an Indie musician amid the already overly complex scam environment within the music industry.
The legality of doing this depends on either the AI related lawsuits that have already been filed or legislative bodies answering the legal questions through new laws.
Copying a voice isn't banned, per se. You can do a perfect impersonation of someone without fear of violating any copyright.
Distributing recorded samples, however, is a clear copyright violation.
So, does AI merely make an impersonation or is it combining files that you have no right to distribute to form the voice? Are images in the Stable Diffusion dataset partially embedded in the outputs or not? Are the stories ChatGPT writes just weapons grade copyright infringement?
It all depends on how many rights we all attribute to the things we stuff into AI datasets. If AI models are ruled to be some kind of special case where the output doesn't violate the copyright if the training dataset, I think AI voice cloning should be completely legal.
I believe the music and movie industries will do what they can to get the law on their side, perhaps demanding royalties for every work of theirs in AI datasets. Maybe they'll demand payment per generated work. I believe Universal is already demanding their partners to stop blocking scrapers for use in AI.
I have no idea how a judge will rule on this. Based on the little work from professionals I've seen, it could go either way with current copyright laws.
> IMO, the real tragedy here is not some threat to Drake (he’ll be fine) but that some dude who wrote an original song had to pretend it was written by AI to get it heard.
I haven't listened to the song in it's entirety or very closely but an actual Drake or Weeknd song usually has much better production and songwriting. Not saying it's not good or that Drake songs are good but from my amateur ear there's a difference in quality.
> I can imagine some independent artists being more open to allowing AI imitation.
Stop with the bs tech optimism. Artists don't want this and rightfully so, no matter how big or small. The only people excited for this are tech-ies with no artistic talent happy to have a way to "disrupt" another industry, alongside an idle (but nervous) curiosity from consumers
Why would you think artists will be able to sell licenses at a sustainable level for their voices? The whole point of these generative AI models is to avoid paying for human labor so the industry will have to be dragged kicking and screaming to do so. We have already seen all of these projects actively avoid paying for their training data, either by just not caring about copywrite or by using only fully open source training data.
If the RIAA takes openAI, et al to court and wins then we may see licenses for this, but given the pittance that spotify pays now for streams, we should expect any license revenue to be utterly minimal for all but the biggest artists.
There are lots of very talented struggling singers and musicians around the world who make $0 with their skills.
They would be happy to get $1000 for signing away their amazing voice and music talent for a large corp who can spend $1 Billion collecting all kinds of voices.
If Mega Corp hires someone who can imitate Elvis and if they use it in their training data, it is completely fair game
Yeah and why do you think that a Mega Corp would be willing to spend $1k per musician when they could offer $100? This hypothetical Mega Corp will surely prefer getting 10 million samples instead of 1 million samples for their $1 billion. In this situation they have all the leverage and we can't rely on their largesse
And it is definitely better to have a super low chance of breaking through in the future and $0 in your pocket than to have no chance of breaking through ever and $1k in your pocket (gone in 1-2 months of rent). Especially considering for most of the really struggling artists, they won't get that deal at all and instead it will be $1k in the pocket of a studio musician.
Once again, for every artist that rejects the $1000 sign-away-for-lifetime offer there are a 1000 who will accept.
Also at the end of the day, every human needs to offer what is value to the society. So even though I'm an extremely good pen-twirler, if the society doesn't value that, I need to change bedsheets (which is more valuable) to make money.
So, I'm not sure why you need to protect "great voice" over "great pen-twirling" skills
Except that having a great singing voice involves lots of practice and maintenance and people dedicate their lives to it, and until now society has valued that enough for it to be economically viable for some. And now this new tech comes along and threatens to undercut that skill for the sake of capturing a market. What happens when a large swath of previously valued skills (including software engineering) are significantly automated, not on the order of decades (like what happened with auto manufacturing) but on the order of a couple years? What happens when there are nowhere near enough jobs to employ the current population? I don't see a mass movement for a UBI to be paired with AI driven job losses, and the only alternative is finding new work for people or having people starve in the street. Our society is terrible at finding adequate replacement work for people who have been automated out of a career. We didn't find replacement work for the auto industry and the "technology always generates new work for people" line is a canard in the face of AI taking over knowledge work. If generative AI fulfills the current expectations for it, there will be no possible job for a human to do that an AI can't do better.
We are sleepwalking into an economic, societal, and political catastrophe and all this forum gets excited about is how much faster and efficient (critically though, not better or more resilient) production systems can become.
If society doesn't owe you a living because your skills now have the same value as "great pen twirling", what you are arguing for is mass death.
Reality says we have more Job openings than ever and the lowest unemployment rate.
I can't find a decent plumber, lawn mower, handyman or a doctor. If the economic incentives pushes artists to acquire skills that are useful to society then we should accelerate that process rather than artificially propping up their skills with debt we can't service.
There will be a time for UBI, but that day hasn't come (Unemployment and Job Openings will be a clear indicator) of that.
Imagine if we propped up textile workers or blacksmith or shoe repairer just like we are planning to prop up artists, the world would be much worse.
Today anyone born in a Western Democracy has access to basic needs -- food, shelter, knowledge without having to do anything.
> Today anyone born in a Western Democracy has access to basic needs -- food, shelter, knowledge without having to do anything.
I'm sorry but this is just not true. The US has ~600k homeless people, the UK had over 2 million people make use of food banks last year (up from ~25k in 2008), and almost all Western nations are seeing metrics around quality of life decline. This isn't AI's doing of course, but we are not in a worry-free position here. Western European countries have a good/decent social safety net, but the Anglo-sphere is far less generous to its poor.
> I can't find a decent plumber, lawn mower, handyman or a doctor. If the economic incentives pushes artists to acquire skills that are useful to society then we should accelerate that process rather than artificially propping up their skills with debt we can't service.
Retraining in midlife is no easy task, and especially so for doctors. Becoming a doctor entails taking on massive amounts of medical school debt and extremely brutal hours until you hit residency. It's a path that is only reasonable for the dedicated and the young, not for a non medical professional like a musician or software engineer in their mid 30s and over.
As for plumbers, yard work, house work, electricians, etc, yes I agree those won't be easily automated away and they don't have the financial up front cost of a medical profession. But there are >25 million office workers (https://www.bls.gov/emp/tables/emp-by-detailed-occupation.ht... between the Management occupations, Business and financial operations occupations, and Computer and mathematical occupations buckets) in this country who are at short term risk of being automated away. If GPT-[X] fulfills the hype, then let's say half of those jobs go away quickly due to a 100% gain in per worker productivity. That becomes 13 million unemployed people and an unemployment rate over 11% (2008 peaked at 10.6%). That becomes a major recession or depression rather quickly.
And as always for people in positions of privilege telling those under the boot to just change their profession, you first. Quit your cushy software job and help fix the labor shortage you see in plumbing.
And this is why normal people hate the tech industry deep in their guts. All we do nowadays is immiserate people by replacing their jobs with shitty but cheap alternatives that reduce the amounts of human labor. When you have your own software job replaced by GPT-[X] and find yourself unable to pay your mortgage, remember how you treated those who had this done to them before it happened to you.
Most software engineers love using GPT-[X] in their work. Copilot is literally for programmers. There are tons of artists using AI tools in their work for the same reasons; it makes them more productive.
I'm 100 certain we'll soon listen to (and later watch) home made virtual performances by deceased artists and actors, with also huge legal implications.
Just suppose someone manages to create a very convincing new song with Elvis and Kurt singing along Amy Winehouse, then releases it without naming the involved artists, so that except fans one couldn't tell who they are with 100% certainty, could lawyers have enough grounds for a take down request and possibly a lawsuit?
That was with his estate's permission, though. To the OP's example I actually think it's going to be much more common for this stuff to happen with permission if the surviving relatives are money-hungry enough. New Elvis album in a few years.
Why not have Tupac explain the reasons for black voters to vote republican? I'm tired of seeing America's great cities turned to waste from Democrats and think Tupac could help.
AI vs the RIAA is going to be quite the fight. They're likely to argue that any algorithmic system which uses one of their works as input and produces something that sounds like it as an output is a derivative work.
Not really a massive surprise that artists are more comfortable with algorithms recommending their work to listeners (and paying them for the listens) than algorithms converting their voice into an audio filter so that songs can be marketed as having their vocals without them being paid to sing on it...
I mean, if the whole thing is that the filter makes you sound like Drake, you are objectively using Drake’s likeness for commercial benefit without his permission.
IANAL, but that’s about as straightforward of a copyright violation as it gets. Doesn’t matter if it’s AI or not.
But as far as generative AI? That’s a way more interesting question. If I train a GAN to create original new songs, is it “sampling” songs in its training data, or is it “learning” from them and then transforming that information into something new? I would say the latter, but I’m sure the RIAA disagrees.
Again, super duper not a lawyer, but my understanding is that everyone has copyright protection over their own voice/face/name.
It’s not like a trademark, which you have to file for. It’s assumed protection.
Edit: after some quick googling, I think it’s inaccurate to call this “copyright” — “personality rights” or “publicity laws” are probably what protect people in this case? But overall, yes, appropriating someone else’s name, image, or likeness is illegal.
This case is the closest the US has seen, but several features of it make it different than AI voices, most notably that an impersonator was hired to sing an artist's own song, so there's no room to say "it's just a coincidence".
It's astonishing to me that some people are so incredibly anti-AI that they will side with the RIAA of all organizations. People who are fine with piracy now start acting as if copyright and IP are the most important things ever.
Couldn't the same be said for your interpretation of this issue? I don't care much about the RIAA, but I don't particularly want AI impersonation to become commonplace and legitimized. That doesn't mean I "side" with excessive copyright enforcement and so on.
Your dislike of the RIAA leads you to believe that someone who is critical of this AI song must be siding with them, which is not really correct.
Pirates were downloading music for their own personal use. RIAA overstepped by suing them individually.
This OTOH is taking existing IP, creating a derivative product, and then uploading it to their Spotify profile, for monetary gain. I think most people can make that distinction.
Is a voice that sounds similar to Drake’s really copyrightable? What if the voice is 20hz higher than Drake’s?
I don’t think you can enforce copyrights on someone sounding like someone else. You obviously can’t claim that it’s Drake, but I’m sure I can release a rap song where I rap like Drake (as many artists already do).
It's copyrightable if it runs Drake's voice through software and then generates an imitation. It's not if he personally listened to a lot of Drake and recorded his own vocal imitation with off the shelf instruments.
This is a PR stunt, it will never go to court anyway. But if it did, the way to have done this legally would have been the Death Row Records approach when Snoop Dogg left them in 1996. They simply signed a rapper who sounded like Snoop, gave called him Top Dogg, and had him perform guest verses for other label acts. He appears on 2pac's "All About U", recorded at a time when both Dr. Dre and Snoop had left the label.
So if I'm personally inspired by a Drake song I heard and use that memory and inspiration to create music in a similar style to Drake, I'm a criminal? Why does the kinds of tools/software used to create the end product matter at all?
But the circumstances there make it different from a lot of AI voice work since in her case there could be no claim that sounding like her was a coincidence since Ford had an impersonator sing one of her songs (which they did properly secure copyright to use).
Producing counterfeits is a completely different issue from piracy and it isn't surprising that people would have different opinions about them despite some lawyers having bundled them together under the label of "Intellectual Property". I frequently saw people getting upset about ripoffs long before AI.
Counterfeits are bad only because the quality was usually much lower than the original. But if counterfeits / copies are of equal quality, I should have the option to buy it as long as it is labelled as such
Most people who are pro-piracy just want noncommercial sharing[0] to be exempted from commercial regulation. The song in question is being sold commercially; it's on a streaming service that pays you per stream.
To put this another way: about a year ago the RIAA went after HitPiece, an NFT scam that was trying to sell NFTs of other people's songs. The amount of cognitive dissonance I had when I realized "I am fully in support of the music labels crushing these guys" was immense.
What you need to take away from this is:
- There are no good guys, so blindly buying into one group is a bad idea. And also even the 'bad guys' might be right once or twice. The RIAA is very much the organization that (legally) stole music from musicians, but that doesn't mean they can't be victims of a crime as well.
- Even your principles and priors might need reevaluation in the light of new evidence. Piracy services were empathetic back when the entertainment industry was dead set on not selling their work through the Internet. They stop being as empathetic when they become profitable entities.
[0] The RIAA's argument back then was that this wasn't really noncommercial because you could just go up the value chain to find someone who had benefited. Sure, you weren't selling the song, but the platform you were sharing it on could sell access to you...
It seems you're in the team sports model of politics. In other words, ideas are not evaluated based on whether they are good or bad towards some principle or goal. They are evaluated based on who supports it. Everything the other team wants must be bad, and must be opposed solely because it comes from that team.
Kind of ironic here as Drake has pretty consistently stolen flows from his contemporaries. There are Drake tracks where he 1:1 takes flows from X, Baby Keem, Kodak... I'm sure the list goes on.
The podcast Switched on Pop had a fascinating episode[1] last year called 'Invasion of the Vibe Snatchers.' They cite a number of instances where people are assigned songwriting credits on tracks simply because the new song is heavily inspired by an existing song or body of work. To be clear, the new songs are not covers, they don't contain any samples and there's no concrete element of the song that you can point to as being copied. However, taken as a whole, it's clear that the new song is similar to the old one.
I bring this up, because it seems to me that if we live in a legal environment where a song's "vibe" is now de facto copyrightable, there is no way anyone will get away with cloning someone's voice (outside of parody.)
In that case the artist is protecting themselves from possible issues, much like Weird Al gets permission from everyone he parodies even though he doesn't have to.
Intuitively it feels logical that this would be challenged, but what do I know about copyright law. Surely cloning a famous artist's voice and musical style to piggyback off of their popularity would be some kind of infringement? And it's not like you can pretend to be loosely "inspired" by them or so, you're definitely intentionally faking their singing and voice.
Beyond this specific occurrence here, I think impersonating someone else's voice or appearance should generally be forbidden, except for say satirical use cases of course (which should come with some disclaimer).
Claiming intent hardly seems fair when it was advertised as being an AI impersonation, no? If they intended to deceive then surely they would have claimed it wasn't AI.
As a friend in the industry pointed out, regardless of the legality of releasing the actual song, this will be a huge boon for producers/songwriters. Imagine being able to produce a song and send an artist a demo of themselves. If they like it, they can immediately jump in and re-record it.
I simply cannot understand how you can have this take. This is the same bizarre wishful thinking that infested this forum during the crypto bubble. Just because you have a hip new tech doesn't mean artists will go against their material interests to engage with it and be won over.
What you are describing is a work loop where the artist is being pushed to the outside, and in the medium/long term this will make artist's working lives much more precarious. And just like with NFT scams, most artists saw the impact to their material interests much more clearly than the tech bro set imagined they would. It is a lot more likely that producers will simply cut artists out of the loop and produce mountains of bland but acceptable crap at near 0 cost, instead of keeping an artist on retainer to re-record/edit songs.
I think the parent comment is referring to pitches sent to successful pop artists who write none of their own lyrics or melodies and take no part in the development of the music beyond recording their vocals. Their personality, looks, and voice are their valuable contribution, not their songwriting/production abilities. They’re always looking for hit songs to jump on and record, not write themselves.
The main reason(s) that this was not impressive were
- Drake is VERY easy to imitate. He's little more than a Lil Wayne impression himself.
- "The Weeknd" here definitely sounded like an imitation of the Weeknd, I'm fairly certain that if I wasn't told in advance, I would be like "something is way off, bad day at the studio? Not actually him?"
The Weeknd is so damn unique that I actually like his voice more than his songs. Starboy is a masterpiece, but I can listen to him sing songs that I know I normally wouldn't like if anyone else was singing them. That said, I actually kinda liked this song, but could tell it wasn't really The Weeknd.
The implications of this kind of technology are wild. Imagine seeing a tribute show where the lead singer runs their voice through a voice changer that matches whatever artist they’re covering.
That will just sound like lip syncing unless the cover band undergoes some heavy cosmetic surgery. The fun of a cover band is watching a rendition of a song where the band are being themselves.
Why not skip the cosmetic surgery and just use a projected, AI mapped face swapper? I do think you would still hear the variation from the live performer, e.g. their rhythm, their breath, their vocal fluctuations, simply with the AI voice changer software slapped on top of it.
I’m not sure how fun the concert would be – you’re right about that – but tech like this is definitely has the potential to transform many facets of the entertainment industry.
For a cover band sure, but a "tribute show" I associate more with trying to exactly match an experience no longer available (the tribute-ees are dead or aged out in some way).
People aren't going to the Van Halen Tribute Show because they're really excited to see John Smith's creative reinterpretation of Eddie's guitar solos.
Just like what happened with AI imagegen and the artists complaining about their styles being stolen will happen with this. After a few months of toying with simply using other artists specific styles, people will start to use the tools to do original works instead, that are not pure clones.
In this case, what is interesting is not that someone can (and did) produce a "fake Drake" song but that you could synthesize a singing voice of your choice and use it to produce songs without finding a new singer or training your own voice in the desired direction. I'm sure someone will make a prompt-based AI voice where you can just say "nasal annoying out of tune male voice" or something and control it that way and you can keep tweaking until you get what you want.
End result is of course even more (perhaps debatable) high quality content but with less star-appeal. We already have that on Spotify etc, there are loads of "troll" companies having hundreds of virtual fake artists releasing albums and tracks already. This will turbo-boost that..
However just like with AI images, it also enables someone with less talent of drawing or singing to do creative works. That has to be pretty valuable as well.
High quality music is made by humans, not machines. The number of successful Drake ripoffs will be roughly the same as the number of successful Metallica ripoffs. One or two will flourish, the rest will be forgotten, if heard at all.
This is not really true anymore in the streaming economy. On Spotify, you earn a LOT of money if your song ends up in any of the generic playlists that people listen to while going to the gym our out running or in a hotel reception etc. You definitely don't need to do hits to get on these lists and earn money. Sure, they won't be remembered, but they will be heard! This reflects a change in how people consume music I guess.
> "The training of generative AI using our artists’ music [...] begs the question as to which side of history all stakeholders in the music ecosystem want to be on: the side of artists, fans and human creative expression, or on the side of deep fakes, fraud and denying artists their due compensation."
Ironically it's been shown that the fans can also be on the side of AI. Some of them will happily download or train a LoRA based on their favorite artist and use it with Stable Diffusion, often to generate images the original artist would never draw (because of limited commission slots/doesn't take commissions at all, lack of experience/interest or explicitly stating they will not take commissions for certain genres).
Someone made a scarcity-reduction machine and my industry of commodifying sound waves will be harmed, let's shame anyone who engages in the reduction of scarcity so we can keep our copywritten sound waves at a high market value.
There's been a new media for music every ten years for the past fifty or so. Records, cassette's, CD's, mp3's, streaming. I've been doing the thought exercise with some of my more creative and/or musically oriented friends recently to try to determine what comes next (if anything).
The only conclusion we're able to come to is some kind of procedurally-generated music that is more specifically catered to each listener. I'm not surprised this wave is starting to break and it's a great time for artists and legal teams to figure out how to work with this technology if there's any possible way to do so.
UMG is taking a straightforward instance of impersonation (it was advertised as Drake and The Weeknd if I recall correctly and the song contained references to the singers' lives from the first person) and they are trying to snatch everything. Which is exactly what should be expected and watched for from the greedy music execs.
They are going further than saying "take this song down" and are trying to go after training on any copyrighted data. It's a massive snatch and grab and a bit of a bait and switch too.
Straightforward impersonation that presents an AI voice as any specific human shouldn't be permitted, and US law is pretty clear on that I think. But an AI vocals system that works like stable diffusion where you ask for a voice with certain vocal characteristics that has been trained on copyrighted music? That's a VERY different thing and is much murkier. IMO it should be allowed, but what's clear is that the law doesn't know if it's allowed yet (in the US anyway).
Couldn't someone simply train a model on a singer who has a similar sound to a famous artist, and then get the rights from that person at a more affordable price? Obviously, it's important not to use the real artist's name to prevent any misunderstandings.
212 comments
[ 4.7 ms ] story [ 202 ms ] threadBut the song sucked, as did the supposed AI contribution. I couldn't quite figure out why everyone used this as the example of how it's all over and AI will take over the world of music.
If all AI stuff was 500x worse than it is currently that would still be a big deal.
Tech does not stand still, with any rate of improvement this stuff will become insanely good quality.
The bar for "AI" gets lower every day. I still think the ChatGPT reaction is something of a marketing psyop. Beyond a couple of novelty searches, I haven't noticed anybody using it as a daily driver.
For the record, here's my understanding of how this "AI-generated song" was made:
1. A human wrote the lyrics
2. A human recorded the beat
3. A human recorded themselves rapping
4. An "AI voice filter" made the recorded voice sound like Drake / the Weeknd.
IMO, the real tragedy here is not some threat to Drake (he’ll be fine) but that some dude who wrote an original song had to pretend it was written by AI to get it heard.
With the sheer volume of online music and lack of people to bribe or sleep with, it’s very difficult for a new entrant to get known.
Sharing a song with a group will create memories and stronger associations
In the 60s you would travel to a music centre. A place where others performing music went. You leveraged a community. If someone in your community made it, it shone a light for everyone and created a scene. People helped each other make it.
That small community with those opportunities have been replaced by centralization without the community or scene multipliers.
Genuinely curious to hear why this is a tragedy. Does everyone who puts out anything original deserve to have it heard by a wider audience?
Edit: Asking as someone who has always been heavily involved in various aspects of music and has watched a massive increase in the volume of music being put out now that it's significantly easier to produce on your own. This is a very complicated topic.
To me, saying it's "disappointing that they had to" feels wrong almost to the point of being offensive.
We can try to prevent plagiarism and improve our detection of it, but there's nothing wrong or disappointing with the fact that people prefer to listen to Bill Gates than a random blogger.
I think the tragedy is that the attention doesn't follow quality or value, it follows fame and gimmicks. In your example, Bill Gates's stupidest brain farts will get 1000x more positive attention that an objectively-better tour-de-force that almost anyone else posts.
It is, but also... welcome to life. This is how it works. What determines success is very often more about things other than the quality of the product.
For evidence, just look at the most popular products at the store. Very rarely are they at the top of the quality scale.
> It is, but also... welcome to life. This is how it works.
So what? That it works that way doesn't make it any less of a tragedy* or any more fair.
* The word chosen by someone else upthread.
Focus on the better aspects of life. Take care of your loved ones. Do your best to improve your local community. Contribute to a better state of the world, no matter how small the contribution may be. Pray to a higher being if you believe in one. That's ultimately all you can do.
I think this is actually perfect and there is no tragedy.
For the record, the type of personal recommendation I'm talking about cannot be viral, in any useful context for the word. I'm talking about recommendations delivered one on one and as an ongoing conversation over many moons; Not recommendations that I see in my Facebook Feed as a Facebook Post from my Facebook Friends to their Facebook Followers. These mediums are insufficient and grow worse, IMO.
If the music was posted to somewhere that it's normal to expect non-serious content (like Reddit or social media) but still properly credited as Ai generated on a non-profit basis (standard practice of course), then that would have been far less offensive.
Except when Mr. Gates and the random blogger are both out of their areas of expertise, people still believe Mr. Gates because he was successful elsewhere instead of applying proper skepticism.
Whereas random bloggers are fly-by-night operations in comparison.
AI and art is a hot topic at the moment, and this artist found a way to capitalize on the zeitgeist.
Yes! This is a failure of capitalism coalescing and centralizing under fewer and fewer big profit centers.
There are many, many creators putting out work as good (if not better) that deserve discovery. (NB: not everything is worthy of "discovery", like some early experimental learning work, but there's definitely undiscovered well-polished work out there.) Culture can be made by the masses, not by the few, but profit sharks don't let that happen.
My concern, or confusion, or lack of understanding, or ignorance, or whatever you wanna call it, is less about the mechanisms of capitalism controlling distribution and moreso the sheer volume of music available. I posit that there is more well-polished music being released on a daily basis than people are capable of listening to, or that we even mathematically have the available time, as a whole, to fit into our lives. Even if we shifted the controls of distribution and such, there seems to be far more content than there are ears to hear the content, and I just don't get how we can ever expect everything worth it's weight in notes to be heard.
I understand the desire, as an artist myself, to be heard, but I also know that I'm one of millions of other people feeling the same way. If everyone's creating (and there are definitely far more creators than ever before), who's listening?
Am I crazy? Am I missing something?
Note the standard of justice being used. It's not, "impersonating someone to use them as a stepping stone is wrong, no matter who does it, or who is the victim".
The standard is, the rich and powerful don't "deserve" equal protection from the law, the real victim is the artist who isn't being recognized because the industry isn't playing "fair". So it's ok if the underdog breaks the law, because the world is stacked against him.
Yeah, and basket ball isn't fair because I'm fat and uncoordinated, and math isn't fair because I'm stupid and don't want to study.
Ironically, these were the same people that said, "Nobody deserves a platform to spread the 'wrong' ideas." when it came to censorship on Twitter, and had the gall to call it "accountability" when people were banned for wrong think about vaccines.
If you're scratching your head and saying, "Wait, what am I missing? That makes no sense! Ten seconds ago it was the rich and powerful censoring the poor underdog, and that was bad. But now big pharma backed censorship is a good thing?" -- it's because you're the sane and logical person, and they're the ones without a consistent worldview.
I don't know, the reasoning seems simpler than any school of thought can offer. Humans are bound to 14-16 waking hours per day. A large portion of those hours are taken up by obligations like work or chores. That leaves a set amount of time to consume media created by other humans. Even if more free time could be had, there will always be a hard biological limit of about 14 waking hours to consume things.
But more and more humans are being born and learning the arts, so combined with instant global communication they will inevitably produce and distribute way more than can be crammed into those hours. Look at a random streamer's Let's Play history for an example. If they play a longer JRPG then they can leave a trail of many hours-long videos per week depending on how long their sessions last. Sometimes that's to the exclusion of every other game in existence. It's a bit sobering if you think about it.
I read a book called The Plenitude by Rich Gold who discusses some of these topics. He doesn't go as far to give a fully workable solution to the problem but one of his suggestions is to challenge creators to limit themselves to 5 "big creations" over their entire life, and focus on honing those creations to the greatest extent. It's an interesting book and I'd recommend it.
Ooh, this sounds interesting. Thanks for the suggestion!
With music, specifically, this has been one of the best impacts of the internet. As long as you avoid services that curate (streaming services, etc.), as a music lover there is a vast world of excellent music at your fingertips that you wouldn't ever have had the chance to hear otherwise.
As an artist, you can develop an audience from the pool of the whole world. You can make a decent living with a pool of a few thousand fans -- and finding a few thousand fans from the set of the Earth's population is a much more feasible thing than before because of the internet.
Not much online is more perilous and rife with scams than trying to make a name for yourself in the music industry. I think people should be given multiple opportunities to be heard by platforms, but now platforms don't do it unless artists pay them to be heard because of the shear volume of fraud accounts uploading fake and low effort music to streaming services.
TikTok shows everything from videos of people cooking eggs to people trimming cow hoofs.. Despite the company being built around music, very little choice is offered in terms of browsing music, and the platform regularly punishes unknown musicians by limiting their views and by imposing complex copyright control on original music to prevent it from being discovered. The majority of the music streamed on the platform funnels royalties for most of the videos on the platforms to already wealthy individuals, scammers, and to companies associated with music streaming.
I think that Independent musicians and labels are putting out music better than major industry now in most cases, yet because of that competition, major label and industry forces work to suppress indie music through lobbying platforms and government for control, and they also throw money around to make competition against them extra hard for musicians that aren't profitable. Tons of independent musicians are operating at a financial loss, and simply having a chance to be heard would make it worthwhile.
So yes, I think that everyone authentic should get as many chances to be heard, because it's an honor for platforms to get new authentic indie music from people that could upload it elsewhere instead, and platforms should be ritually meticulous in promoting all of the authentic independent artists on these platforms in support of music growth, because that's a far better prospect for long term success than just lining pockets of ancient and withering music label executives.
One less clone if a world of clones isn't a tragedy.
Man, what a skewed version of justice this is. "He'll be fine." The REAL victim is this unnoticed artist? Give me a break. What he did impersonating another artist was wrong.
There's this perverse notion that has infused our society where it's ok to rob from a Target or Walmart because it doesn't really "hurt" them. "They have insurance. They'll be fine!"
That standard of justice will lead society to a very VERY bad place if you follow it.
Anyways, besides that, impersonation for personal gain or to cause disadvantage is already criminal. If Drake wants to do something about that, he can.
But this case is more nuanced. A compelling argument can be made against the very idea of robbery of IP. I think the vast majority of "intellectual property" should not, in fact, be anyone's property, and therefore no one can be robbed of it.
Who is harmed and who benefits in the situation where Drake has a lock and key on his IP?
Consumers are harmed --- they clearly enjoy new Drake music regardless of whether or not it was written and sung by Drake, but this is not allowed.
The "AI artist" is harmed --- they could have created something valuable in the world, but this is not allowed.
Drake artificially benefits from a regulatory monopoly. Everyone is strictly worse off because of arbitrary, draconian rules.
A much better example is drug patents. Literally millions of lives could be saved --- and at a lower cost --- if patents didn't exist. Patents and intellectual property are just state-sanctioned monopolies where everyone is harmed and coerced to pay rent to some random asshole.
Drake has a right to decide what his voice says, and have a monopoly on it.
The AI artist is not "harmed" because they can't use Drake's voice.
I can't believe this needs to be said.
Because it is separate from copyright, US federal statutory fair use considerations don’t apply to this right (I’m not sure to what extent Constitutional fair use might apply to it; statutory fair use in copyright largely recapitulates what were found to be Constitutional limits on the copyright power stemming from the First Amendment, I don’t know if it has ever been considered the extent to which the First Amendment, applied ot the states through the 14th, applies similar restriction on the right of publicity, which is in the US, to the extent it exists, a state-law intellectual property right.)
There was a very important and similar case though where Bette Midler sued Ford over inappropriate use of "her" voice performed by an impersonator.
https://en.m.wikipedia.org/wiki/Midler_v._Ford_Motor_Co.
The context seems important in that case though. Ford had the impersonator sing a song that she herself released! (They had acquired copyright release for that part) So there's no reasonable claim that Bette Midler was not the one they were impersonating.
If you sang a song you wrote in a voice extremely similar to Drake's without mentioning Drake in any way, then that seems very different to me.
If it's clearly qualified that this is a fabrication of his voice, then what right does he have to control that? He's not being harmed. He's not being misrepresented.
As to harm, I don't know if he's being harmed or not -- but to my mind, that's not an important point. If, as I believe, it's wrong to produce works that show you doing and saying things you didn't do (without your permission), then it's wrong whether or not you were actually harmed by it. And what is harm, anyway? Being embarrassed is a kind of harm.
From my POV as long as you make it clear it isn't actually him, you should be able to do it. What's next, I can't do a parody of Drake either? Should we ban the Nixon-moonlanding-crash video?
Nobody has a unique voice anyway, I'm sure one of those 8B people sounds a lot like Drake. Does that other guy get to decide what songs Drake sings? Of course not, having a specific set of traits should not give you rights over it.
Does Arnel Pineda, who sounds a heck of a lot like Steve Perry, owe Steve a cut of whatever he makes with the band? (Aside from the fact that obviously there's contracts in place.) He's singing his ass off every time he's on stage, just like Steve did when he was the front man for Journey.
In the end this is an argument for the already powerful to stop the newcomer. There's no reason some dude shouldn't be able to do the same work as some other dude, as long as he doesn't pretend to actually be that other guy.
I think the onus is on the speaker/singer to prove they are being materially harmed by a fabrication of their voice. In many cases, this could be simple: "That fabrication is being used to misrepresent my beliefs AND it's causing me harm." Fair enough. But is someone making a Drake song causing Drake harm? No. He could argue loss of income, but that would be shaky. He didn't come up with the lyrics. He didn't come up with the beat. He didn't even actually sing it. If it's so easy to rip off his voice to create a hit song, then is he actually still losing income? Why doesn't he come up with the hit song?
I agree 100%. It's a bit disturbing that people seem to think it was a lesser wrong just because the victims are wealthy and famous.
Learn to spot the DARVO tactic: Deny, accuse, reverse victim and offender.
The lyrics are semi-auto-biographical about The Weeknd and Drake. The guy who wrote this didn't date Selena Gomez. It's pretty uninteresting as a song if it's not from The Weeknd and Drake.
The impressive part is that the voices could be very convincing. That's what made it shareable.
That's what I thought at first too, but the available info suggests that the AI "autotune" was done on a human reference vocal. (cf https://twitter.com/rpnickson/status/1647603043851026433)
This was predicted (as so many other things have) in anime...
Recall JEM? the soft-core-USA-version of anime?
but so many things have been predicted/projected through anime (art)... the cyberpunk future of holographic AI media pieces has been a long trope in anime...
at some point, youth will not care if its AI (EDM? YO!) and will just accept deepfake AI generated content along with AI generated sexually apealing rockstars...
What was not predicted in anime, but will become a reality is the personalization of the 'messenger' where we listen to the same song, but mine is personalized to me, and you, yours. so if, say, female, the song may be talored to being delivered to you differently than to male... however, that doesnt mean the delivery is based on gender... there will be subtle differences in timbre tone and rythm based on you....
Dynamically generated, personalized content already exists in the form of video games. Yet it is not notable when an NPC remarks on my unique gear, achievements, etc.
Technology may eventually enable easier remixing of art to personal tastes. However, this has also existed as long as art itself.
What people look for in the artist is their uniquely curated tastes, human connection to the person themselves, and a sense of community with fellow fans.
A machine that creates everything for everyone can provide none of these. Its output will be products, not art. It will have customers, not fans.
Some people do, some people don't. There's nothing obligating an anime/SNS art hoarder to think this way, and if all they ever do is consume the end product without giving a thought to how the artist laid down the individual strokes that made the picture or how the artist's circle of friends and life story influenced their style, there's no consequences. The artist and other people with a similar mindset are unable to criticize these consumers because their behavior is invisible. When AI artists started appearing with art featuring blatantly copied styles, they did finally become open to such criticism.
I have a feeling the nature of endless feeds with a practically infinite amount of art to consume has conditioned a lot of consumers to never have think about the human artist. They can just look at the stash in their Pictures folder and think "that's a cute X" divorced from the artistic context that's been integral in earlier art history. The consequences of this are only starting to become apparent right now. But I think if there were nothing like Stable Diffusion then one would be unable to tell apart the people who would be open to copying artists' styles with a hypothetical AI program to-be-invented in the future from people who wouldn't. They'd all just appear to be the same breed of fans. It was the invention of the technology and its spread that revealed those preferences.
Which would totally destroy one of the things that makes music powerful -- it provides a shared, deep experience. Not to mention that it all but eliminates another purpose for music: to communicate to people.
The more use cases I hear about AI, the more I think that economic devastation isn't the only thing we're flirting with. We're also flirting with destruction of society as we erase people and meaningful societal bonding.
For fucks' sake, I am guilty of helping the world get to the current dystopian shitshow we have going on today (as is 99% of HN, but especially us from the 1980s)
So take my stuff tongue and cheek.
As Above So Below.
I think a lot of really weird shit is going to happen in the next decade.
a few days EARLIER this was released in Israel by KAN (television chanel)..
I know you probably can't speak hebrew but here is a song made (vocals/lyrics/music) by AI, of two dead singers who are very famous in Israel (Zohar Argov and Ofra Haza).. It is convincing
https://youtu.be/7ND1Pw6QD_0
I think if it wasn't in hebrew language(and to be honest, Israeli) it would have been on HackerNews or at least on reddit somewhere.
There is also some really good ones of Curt Cobain. The AI group is a Tel Aviv startup calle Aiode.
IIRC, I read that in the 70s many people thought Popcorn by Hot Butter was also written by a computer, but it was just played on (analog) synthesizers.
https://en.wikipedia.org/wiki/Popcorn_(instrumental)
https://www.youtube.com/watch?v=YK3ZP6frAMc
I just ask you don't forget about the little people when you become famous
Just don't actually have it sing in a pitch-perfect manner. That's uninteresting.
https://en.wikipedia.org/wiki/Racter
The story is that there was a program, Racter, which allegedly created a book titled "The Policeman's Beard Is Half-Constructed" back in the 1980s. The book was released and was a hit primarily because of that AI connection, and a version of Racter was also sold. However, the Racter that got released could never have composed the book, leading me to this post:
https://news.ycombinator.com/item?id=34575252
> The least hypothesis is that the "good" version of Racter that supposedly composed the novel never existed, and that it was all a hoax made up by someone who wanted to bask in some publicity and sell the Racter that actually got released. As you say, the output is suspiciously good for something they insisted was accomplished on a home computer of the era and, oh no, they can't release the software that did it for Reasons... yeah, pull the other one, it's got bells on.
Putting out unauthorized work by others (for profit), especially when it emulates that they actually created the product, can harm their reputation. Putting out unauthorized work from musicians (for profit) often represents that the work was cleared... It's a move to either generate attention for the tool that they used, or for the emulation services they provide as "musicians", or even worse it could be a scam to make money off of putting out counterfeit music... I guess whoever was behind it (and now there will likely be even more of this) figures the legal issues that arise are worth the risk in seeking whatever profit may come from breaking common rules of music production such as this.
It's troubling to see this happen as an Indie musician amid the already overly complex scam environment within the music industry.
Copying a voice isn't banned, per se. You can do a perfect impersonation of someone without fear of violating any copyright.
Distributing recorded samples, however, is a clear copyright violation.
So, does AI merely make an impersonation or is it combining files that you have no right to distribute to form the voice? Are images in the Stable Diffusion dataset partially embedded in the outputs or not? Are the stories ChatGPT writes just weapons grade copyright infringement?
It all depends on how many rights we all attribute to the things we stuff into AI datasets. If AI models are ruled to be some kind of special case where the output doesn't violate the copyright if the training dataset, I think AI voice cloning should be completely legal.
I believe the music and movie industries will do what they can to get the law on their side, perhaps demanding royalties for every work of theirs in AI datasets. Maybe they'll demand payment per generated work. I believe Universal is already demanding their partners to stop blocking scrapers for use in AI.
I have no idea how a judge will rule on this. Based on the little work from professionals I've seen, it could go either way with current copyright laws.
I haven't listened to the song in it's entirety or very closely but an actual Drake or Weeknd song usually has much better production and songwriting. Not saying it's not good or that Drake songs are good but from my amateur ear there's a difference in quality.
I can imagine some independent artists being more open to allowing AI imitation.
Stop with the bs tech optimism. Artists don't want this and rightfully so, no matter how big or small. The only people excited for this are tech-ies with no artistic talent happy to have a way to "disrupt" another industry, alongside an idle (but nervous) curiosity from consumers
I could see Drake using it for demos/reference tracks though since he has ghost writers
I'm sure eventually some artists will sell licenses to AI models of their voices. It's not a "real" feature since AI will never be an exact replica.
If the RIAA takes openAI, et al to court and wins then we may see licenses for this, but given the pittance that spotify pays now for streams, we should expect any license revenue to be utterly minimal for all but the biggest artists.
They would be happy to get $1000 for signing away their amazing voice and music talent for a large corp who can spend $1 Billion collecting all kinds of voices.
If Mega Corp hires someone who can imitate Elvis and if they use it in their training data, it is completely fair game
And it is definitely better to have a super low chance of breaking through in the future and $0 in your pocket than to have no chance of breaking through ever and $1k in your pocket (gone in 1-2 months of rent). Especially considering for most of the really struggling artists, they won't get that deal at all and instead it will be $1k in the pocket of a studio musician.
Also at the end of the day, every human needs to offer what is value to the society. So even though I'm an extremely good pen-twirler, if the society doesn't value that, I need to change bedsheets (which is more valuable) to make money.
So, I'm not sure why you need to protect "great voice" over "great pen-twirling" skills
We are sleepwalking into an economic, societal, and political catastrophe and all this forum gets excited about is how much faster and efficient (critically though, not better or more resilient) production systems can become.
If society doesn't owe you a living because your skills now have the same value as "great pen twirling", what you are arguing for is mass death.
I can't find a decent plumber, lawn mower, handyman or a doctor. If the economic incentives pushes artists to acquire skills that are useful to society then we should accelerate that process rather than artificially propping up their skills with debt we can't service.
There will be a time for UBI, but that day hasn't come (Unemployment and Job Openings will be a clear indicator) of that.
Imagine if we propped up textile workers or blacksmith or shoe repairer just like we are planning to prop up artists, the world would be much worse.
Today anyone born in a Western Democracy has access to basic needs -- food, shelter, knowledge without having to do anything.
I'm sorry but this is just not true. The US has ~600k homeless people, the UK had over 2 million people make use of food banks last year (up from ~25k in 2008), and almost all Western nations are seeing metrics around quality of life decline. This isn't AI's doing of course, but we are not in a worry-free position here. Western European countries have a good/decent social safety net, but the Anglo-sphere is far less generous to its poor.
> I can't find a decent plumber, lawn mower, handyman or a doctor. If the economic incentives pushes artists to acquire skills that are useful to society then we should accelerate that process rather than artificially propping up their skills with debt we can't service.
Retraining in midlife is no easy task, and especially so for doctors. Becoming a doctor entails taking on massive amounts of medical school debt and extremely brutal hours until you hit residency. It's a path that is only reasonable for the dedicated and the young, not for a non medical professional like a musician or software engineer in their mid 30s and over.
As for plumbers, yard work, house work, electricians, etc, yes I agree those won't be easily automated away and they don't have the financial up front cost of a medical profession. But there are >25 million office workers (https://www.bls.gov/emp/tables/emp-by-detailed-occupation.ht... between the Management occupations, Business and financial operations occupations, and Computer and mathematical occupations buckets) in this country who are at short term risk of being automated away. If GPT-[X] fulfills the hype, then let's say half of those jobs go away quickly due to a 100% gain in per worker productivity. That becomes 13 million unemployed people and an unemployment rate over 11% (2008 peaked at 10.6%). That becomes a major recession or depression rather quickly.
And as always for people in positions of privilege telling those under the boot to just change their profession, you first. Quit your cushy software job and help fix the labor shortage you see in plumbing.
It's as good as a right as an artist
Most software engineers love using GPT-[X] in their work. Copilot is literally for programmers. There are tons of artists using AI tools in their work for the same reasons; it makes them more productive.
If you take great joy in typing
public class MyClassImpl implements MyClass
then you are perfectly free to keep doing so.
This dude was super excited that he worked an Eminem deepfake into his performance: https://youtu.be/40lOIZLUACU
And much simpler, too; I don't think anyone would regard Spotify's Discover Weekly as artificial intelligence.
Even if the AI is only used for lyrics as a voice filter, it's still quite close to the heart of the work.
Then you have its market effect, where original artists will face a heavy economic impact from the infringement (if that's what the courts rule).
That's just my prediction, but I have limited experience in the area.
IANAL, but that’s about as straightforward of a copyright violation as it gets. Doesn’t matter if it’s AI or not.
But as far as generative AI? That’s a way more interesting question. If I train a GAN to create original new songs, is it “sampling” songs in its training data, or is it “learning” from them and then transforming that information into something new? I would say the latter, but I’m sure the RIAA disagrees.
It’s not like a trademark, which you have to file for. It’s assumed protection.
Edit: after some quick googling, I think it’s inaccurate to call this “copyright” — “personality rights” or “publicity laws” are probably what protect people in this case? But overall, yes, appropriating someone else’s name, image, or likeness is illegal.
They MIGHT be protected under other laws.
This case is the closest the US has seen, but several features of it make it different than AI voices, most notably that an impersonator was hired to sing an artist's own song, so there's no room to say "it's just a coincidence".
https://en.m.wikipedia.org/wiki/Midler_v._Ford_Motor_Co.
The dude made the beat, lyrics, vocals. He just changed his voice.
People who refuse a position because they've found themselves with strange bedfellows are one of the primary reasons things are so fucked these days.
Your dislike of the RIAA leads you to believe that someone who is critical of this AI song must be siding with them, which is not really correct.
This OTOH is taking existing IP, creating a derivative product, and then uploading it to their Spotify profile, for monetary gain. I think most people can make that distinction.
I don’t think you can enforce copyrights on someone sounding like someone else. You obviously can’t claim that it’s Drake, but I’m sure I can release a rap song where I rap like Drake (as many artists already do).
This is a PR stunt, it will never go to court anyway. But if it did, the way to have done this legally would have been the Death Row Records approach when Snoop Dogg left them in 1996. They simply signed a rapper who sounded like Snoop, gave called him Top Dogg, and had him perform guest verses for other label acts. He appears on 2pac's "All About U", recorded at a time when both Dr. Dre and Snoop had left the label.
What they *might* be is protected by other laws. The closest case the US has seen is this Bette Midler one: https://en.m.wikipedia.org/wiki/Midler_v._Ford_Motor_Co.
But the circumstances there make it different from a lot of AI voice work since in her case there could be no claim that sounding like her was a coincidence since Ford had an impersonator sing one of her songs (which they did properly secure copyright to use).
To put this another way: about a year ago the RIAA went after HitPiece, an NFT scam that was trying to sell NFTs of other people's songs. The amount of cognitive dissonance I had when I realized "I am fully in support of the music labels crushing these guys" was immense.
What you need to take away from this is:
- There are no good guys, so blindly buying into one group is a bad idea. And also even the 'bad guys' might be right once or twice. The RIAA is very much the organization that (legally) stole music from musicians, but that doesn't mean they can't be victims of a crime as well.
- Even your principles and priors might need reevaluation in the light of new evidence. Piracy services were empathetic back when the entertainment industry was dead set on not selling their work through the Internet. They stop being as empathetic when they become profitable entities.
[0] The RIAA's argument back then was that this wasn't really noncommercial because you could just go up the value chain to find someone who had benefited. Sure, you weren't selling the song, but the platform you were sharing it on could sell access to you...
Reverse engineering someone's talent and identity and using it to create new work under their brand and identity should be a crime
Hell, take it one step further. Non-consentual deepfakes of any kind should be a crime.
I bring this up, because it seems to me that if we live in a legal environment where a song's "vibe" is now de facto copyrightable, there is no way anyone will get away with cloning someone's voice (outside of parody.)
1. https://switchedonpop.com/episodes/invasion-of-the-vibe-snat...
Beyond this specific occurrence here, I think impersonating someone else's voice or appearance should generally be forbidden, except for say satirical use cases of course (which should come with some disclaimer).
Let's just call a spade a spade. This is impersonation with the intent to deceive.
What you are describing is a work loop where the artist is being pushed to the outside, and in the medium/long term this will make artist's working lives much more precarious. And just like with NFT scams, most artists saw the impact to their material interests much more clearly than the tech bro set imagined they would. It is a lot more likely that producers will simply cut artists out of the loop and produce mountains of bland but acceptable crap at near 0 cost, instead of keeping an artist on retainer to re-record/edit songs.
Kan and Aidode(israeli AI music generation startup) made a duet with two dead singers:
Ofra Haza + Zohar Argov https://youtu.be/7ND1Pw6QD_0
got 200k play in a few days, not bad for hebrew only. Very convincing.
Completely lost in non hebrew speaking world (99% of the world), and just as impressive I think.
The main reason(s) that this was not impressive were
- Drake is VERY easy to imitate. He's little more than a Lil Wayne impression himself.
- "The Weeknd" here definitely sounded like an imitation of the Weeknd, I'm fairly certain that if I wasn't told in advance, I would be like "something is way off, bad day at the studio? Not actually him?"
Also, Starboy? Funny, I FAR prefer everything before that.
https://m.youtube.com/watch?v=y7r6PAkFRfU&feature=youtu.be
The implications of this kind of technology are wild. Imagine seeing a tribute show where the lead singer runs their voice through a voice changer that matches whatever artist they’re covering.
I’m not sure how fun the concert would be – you’re right about that – but tech like this is definitely has the potential to transform many facets of the entertainment industry.
People aren't going to the Van Halen Tribute Show because they're really excited to see John Smith's creative reinterpretation of Eddie's guitar solos.
In this case, what is interesting is not that someone can (and did) produce a "fake Drake" song but that you could synthesize a singing voice of your choice and use it to produce songs without finding a new singer or training your own voice in the desired direction. I'm sure someone will make a prompt-based AI voice where you can just say "nasal annoying out of tune male voice" or something and control it that way and you can keep tweaking until you get what you want.
End result is of course even more (perhaps debatable) high quality content but with less star-appeal. We already have that on Spotify etc, there are loads of "troll" companies having hundreds of virtual fake artists releasing albums and tracks already. This will turbo-boost that..
However just like with AI images, it also enables someone with less talent of drawing or singing to do creative works. That has to be pretty valuable as well.
That's how hip hop came into being, with the advent of samplers and drum machines.
Ironically it's been shown that the fans can also be on the side of AI. Some of them will happily download or train a LoRA based on their favorite artist and use it with Stable Diffusion, often to generate images the original artist would never draw (because of limited commission slots/doesn't take commissions at all, lack of experience/interest or explicitly stating they will not take commissions for certain genres).
The only conclusion we're able to come to is some kind of procedurally-generated music that is more specifically catered to each listener. I'm not surprised this wave is starting to break and it's a great time for artists and legal teams to figure out how to work with this technology if there's any possible way to do so.
How would this not represent the death of music? Music is much more than a sequence of pretty sounds.
They are going further than saying "take this song down" and are trying to go after training on any copyrighted data. It's a massive snatch and grab and a bit of a bait and switch too.
Straightforward impersonation that presents an AI voice as any specific human shouldn't be permitted, and US law is pretty clear on that I think. But an AI vocals system that works like stable diffusion where you ask for a voice with certain vocal characteristics that has been trained on copyrighted music? That's a VERY different thing and is much murkier. IMO it should be allowed, but what's clear is that the law doesn't know if it's allowed yet (in the US anyway).
https://www.youtube.com/watch?v=_Y0U4hmHt44