Unpopular Opinions

[Beananium]

My unpopular opinion:
Using AI for the MV or the instrumental isnt that bad

Not really an unpopular opinion on my part, and I do apologize in advance for the pretentious sounding presentation of this personal point, but I feel that one of the best elements of Vocaloid in all aspects is human expression.

It's accessible software that gives voices to those that would otherwise have trouble with getting their talent and heart heard. If it weren't for seeing what stories other people told with songs using those voicebanks and the fan community's inside jokes and portrayals with the characters themselves, I may have never gotten passionate about creative writing and art to the extent that I have.

I may have never found an outlet to express the feelings I have by imagining these characters in a variety of different ways. I may have not laughed out loud getting more absurd with these characters and how I see them. I may have not met people that became some of my best friends over that common ground. And I definitely wouldve never found pieces of art that moved me emotionally to the point of finally coming with terms with a feeling that was close to killing me.

The best songs and content using Vocaloid and it's characters is art in its purest form. A form of free expression using these tools.

And A.I generation in any part of expressing those feelings spits in the face of art. Even if you suck at, say, illustration, and need to slap a picture onto a song you labored for weeks of near constant all-nighters, that picture will not be done with any understanding of what you need to convey. You're not giving your vision the dignity it deserves. It cheapens the entire work. How would anyone want to feel anything when they see something so fake slapped in it to make it appetizing? Something silly will seem fake, and something powerful will seem weak.

A much better option is to either tough it out and make something within your skill-level; wowaka made some of the most iconic Miku songs of all times, and the song art is simple abstract images in white and grey. And it serves the songs well and have become instantly identifiable in their own right. Or commission someone for art and talk to them; tell them your vision, and talk about what you want out of it. You'll give someone a job, and you may get interest from their end. Their work can even drive in more people to that song.

And A.I music makes me wonder why someone would even bother making a Vocaloid song.

Generative A.I is something I personally have very little tolerance for in Vocal Synth creation, and I think anyone using it shouldn't be given any attention. Leave them to either realize their mistake or linger in unpopularity.

 

[IUnknown]

I only write translyrics and tune, so covers are my limit. If one day I decided to do an original Vocaloid song, then AI would be a way to bridge my gaps in knowledge regarding art and melody. (NOT saying I would do it, im just saying it could be a way) Just look at how nice Liar Macaron turned out with an obviously AI-generated instrumental.

And A.I generation in any part of expressing those feelings spits in the face of art

Any part? So if I got lazy one day, decided a section of the song would have 0 manual tuning, and let SynthV/Vocaloid6's (terrible) AI auto-pitch do the work for me, then am I spitting on art too?

Personally, if there is at least one human element driven by genuine creative intent, then it's good enough for me. I don't think that an AI background, for example, poisons the well

 

[lIlI]

I only write translyrics and tune, so covers are my limit. If one day I decided to do an original Vocaloid song, then AI would be a way to bridge my gaps in knowledge regarding art and melody. (NOT saying I would do it, im just saying it could be a way) Just look at how nice Liar Macaron turned out with an obviously AI-generated instrumental.

I think the main, important concern, is how most of those AI tools are trained from stolen content taken without consent or compensation for the artists, unlike Synthesizer V where the singers are paid and agreed to the use of their likeness. For example, Liar Macaron scraped from a specific existing Vocaloid artist to create an imitation of their work. This type of unchecked exploitation discourages artists from sharing their work publicly, which sucks for everyone.

On a more philosophical note: flaws are a key component to the originality and authenticity of artistic works. So many musicians' earlier, rougher songs are beloved for their rawness and intimacy: a huge amount of the most unique artistic expressions are born from the openness and experimentation of inexperience. It's in how artists tackle those challenges that their style and personal voice is formed.

What qualifies as an imperfection is too subjective to label, with lots of fans complaining that artists lose their spark when they become more polished. Using AI to smooth away the edges feels like a misunderstanding of art's intricacies. If AI were used to make something more 'perfect' it would severely diminish my interest in the final work.

What makes art meaningful to me is that its our oldest method of communication. Even if a chatbot is perfectly agreeable and never makes a grammatical error, I get a lot more out of socialising with people. That's why the concept of AI perfectly imitating humans is for many, nightmarish, not exciting. We loved the human connection achieved through art; AI makes it hard to discern if you're connecting with a person, or an imitation of one. I think that's the reason why the majority of music fans would rather listen to something made by people but rough around the edges, than polished by a bot.

How these conversations interact with artificial voices is an interesting one. Personally, I view vocal synths as equivalent to the electronic instruments that have existed since the 80s, and see using autotuning as no different from incorporating an unedited sine wave or raw guitar sample pack - perfectly fine, as it's just one instrument in the arrangement. I don't demand every sample or sound be edited. People draw the line in different places, and some prefer only to listen to live instrumentalists or manual tuning: I understand the value in that too, and my most unpopular opinion is that I do get, on some level, those who prefer real vocalists for the type of personal expression they can add. Others dislike listening to pop songs with too many writers: the line of human authenticity has always been up for debate in music. But one thing's for sure: everyone values it highly.

 

[IUnknown]

I think the main, important concern, is how most of those AI tools are trained from stolen content taken without consent or compensation for the artists

I don't think this will be a lasting argument. Even if an AI like SUNO dropped tomorrow with 100% consensual training data, it would still be disliked by the exact same people. Let's not kid ourselves here -- the main reason GenAI is hated is because it (most of the time) acts as a lazy shortcut.

note: flaws are a key component to the originality and authenticity of artistic works

Using AI somewhere in your workflow doesn't make your work perfect; it just outputs a statistical average. If flaws are such a virtue, then I guess we should start telling people to skip SV Teto and stick to her UTAU voice since it is much less "perfect"?

We loved the human connection achieved through art

Okay, so if someone spends an entire month working on a Vocaloid song and puts badly made MS Paint art as the background, then it's wholesome chungus 100 reddit gold, but if they use an AI image, the song is suddenly devoid of any human emotion?

 

[Alphonse]

Okay, so if someone spends an entire month working on a Vocaloid song and puts badly made MS Paint art as the background, then it's wholesome chungus 100 reddit gold, but if they use an AI image, the song is suddenly devoid of any human emotion?

No one said the song is devoid of any human emotion, but the image sure is.

 

[IUnknown]

No one said the song is devoid of any human emotion, but the image sure is.

So the image has no human emotion but the song is fine. Cool. Now apply that same logic to a song with AI instrumental -- the instrumental has no human emotion by your standard, but there's still a human writing, tuning, and making creative decisions around it. So its OK by me.
The crux of my argument is that even if you use Gen AI in some parts of your workflow, its not the end of the world, as I said here:

Personally, if there is at least one human element driven by genuine creative intent, then it's good enough for me.

 

[Bookworm2]

There're also the environmental impacts of AI usage, but nobody wants to talk about that, and neither do I. But the biggest part to me is that if you're using AI, you aren't really making anything. Use it for the instrumental? That's like half the song that you're not making, and at that point it isn't really "your" song, it's the AI's song. You didn't make it, something else made it. Same with other AI generated parts, especially vocals. If you use AI for the vocals, what's even the point of making it?

 

[IUnknown]

Use it for the instrumental? That's like half the song that you're not making, and at that point it isn't really "your" song, it's the AI's song.

Imagine going up to Oxi and telling them 'Hey your Kasane Territory song isnt yours because you used a Touhou Instrumental'. Would that make any sense?

 

[Bookworm2]

Imagine going up to Oxi and telling them 'Hey your Kasane Territory song isnt yours because you used a Touhou Instrumental'. Would that make any sense?

In this case, yeah I'm in the wrong. I think I misstated my point, sorry. But I am neither smart enough nor verbose enough to try and restate my point, so I'll just leave it with this. With Kasane Territory, I'd imagine that Oxi heard the original song, thought that they could build off of it yet make the different enough to make the song their own, and did so. But with AI, you don't really do that. Can I explain myself? No, not really. Does it make sense to me? Yes. Bye.

 

[kaede00582]

I feel like my thoughts on AI will get me attacked from both pro-AI and anti-AI sides.

First of all, there are clear ethical issues with AI as it exists right now. I strongly dislike when people who use AI to generate an entire song claim they are real musicians for giving a prompt to an AI model. I hate it even more when people hide that their "work" is completely AI-generated. That's no different to copying someone else's homework, hoping you will get credit for work you didn't do. It has also hurt the music landscape. Multi-billion dollar companies are training their AI models on art which they have taken from real artists without consent or compensation, and people looking for ~passive income~ are flooding Spotify (and other music platforms) with mindless AI generated slop, taking views and money away from real artists. All of this is incredibly unfair to people who truly love music, have real talent, and are trying to make a living from their craft.

But the thing is, AI is probably being used a lot more than you think by 'non-AI' artists. Audio processing VSTs/plugins are often powered by AI nowadays, especially for complex tasks such as background noise removal, responsive real time EQ, or simulating analog audio equipment. On the composing side, there are AI VSTs which generate MIDI chord progressions or drum beats to kick start the music writing process. What makes those AI VSTs different from Suno is that they can't generate a finished song for you. All they can do is make your creative process more efficient and your final results more polished. That's very important for professional audio technicians who are often under time pressure to deliver high quality results to clients. Also, it gives indie artists and music producers higher-quality tools to use, without replacing the artists themselves.

Japanese vocal synth producers are already using AI as tools for their songs. Brainrot - Tokyo Manaka ft Kasane Teto uses AI-generated backgrounds in the MV. It's about liminal spaces and, uh, brainrot, so AI slop images fit the theme, and those images only there to complement Teto, who is animated by a real person. kikuo has also used AI to create some of his concert visuals. When I went to see him live, the vast majority of images/animations on screen were made by the illustrator si_ku who he has worked with for a long time. However, there was one song where kikuo was clearly experimenting with the uncanny nature of AI generated animation to complement the creepy vibes of his music as an artistic choice, which I thought worked well. (IIRC he has completely stopped using AI now after feedback from his audience.)

kikuo and Tokyo Manaka are real artists who have original creative vision and AI is just one tool in a much larger creative process for them. I think that is very different to someone having every second of a song generated for them by AI, and I wish more people would make a distinction between those contexts.

 

[WacoWacko39]

I feel like my thoughts on AI will get me attacked from both pro-AI and anti-AI sides.

First of all, there are clear ethical issues with AI as it exists right now. I strongly dislike when people who use AI to generate an entire song claim they are real musicians for giving a prompt to an AI model. I hate it even more when people hide that their "work" is completely AI-generated. That's no different to copying someone else's homework, hoping you will get credit for work you didn't do. It has also hurt the music landscape. Multi-billion dollar companies are training their AI models on art which they have taken from real artists without consent or compensation, and people looking for ~passive income~ are flooding Spotify (and other music platforms) with mindless AI generated slop, taking views and money away from real artists. All of this is incredibly unfair to people who truly love music, have real talent, and are trying to make a living from their craft.

But the thing is, AI is probably being used a lot more than you think by 'non-AI' artists. Audio processing VSTs/plugins are often powered by AI nowadays, especially for complex tasks such as background noise removal, responsive real time EQ, or simulating analog audio equipment. On the composing side, there are AI VSTs which generate MIDI chord progressions or drum beats to kick start the music writing process. What makes those AI VSTs different from Suno is that they can't generate a finished song for you. All they can do is make your creative process more efficient and your final results more polished. That's very important for professional audio technicians who are often under time pressure to deliver high quality results to clients. Also, it gives indie artists and music producers higher-quality tools to use, without replacing the artists themselves.

Japanese vocal synth producers are already using AI as tools for their songs. Brainrot - Tokyo Manaka ft Kasane Teto uses AI-generated backgrounds in the MV. It's about liminal spaces and, uh, brainrot, so AI slop images fit the theme, and those images only there to complement Teto, who is animated by a real person. kikuo has also used AI to create some of his concert visuals. When I went to see him live, the vast majority of images/animations on screen were made by the illustrator si_ku who he has worked with for a long time. However, there was one song where kikuo was clearly experimenting with the uncanny nature of AI generated animation to complement the creepy vibes of his music as an artistic choice, which I thought worked well. (IIRC he has completely stopped using AI now after feedback from his audience.)

kikuo and Tokyo Manaka are real artists who have original creative vision and AI is just one tool in a much larger creative process for them. I think that is very different to someone having every second of a song generated for them by AI, and I wish more people would make a distinction between those contexts.

I take no issue with you're first paragraph, in fact I generally agree.

Regarding the second though, I develop differing opinions. In respect to ai being used to make the end result better, and removing 'busy-work' (background noise removal is an example you used), I may yield the point that it's mostly* benign and helpful. But that mostly* I used refers to a far more general concern of mine using ai, which is: how convenient should everything be? That question is somewhat to the side of this, so I'll take up that torture-stake later and keep on point.

But any infringing upon the creative process (MIDI cords progressions and drum beats, as you mentioned), these may not be these may not be 'moral felonies', like those who generate entire works, but they may be 'moral misdemeanors'. All work involves a level of frustration, wood working for example is joked as to be 90% sanding 10% everything else. I am not a wood workers, so I can't confirm, but I do know no one enjoys sanding. Removing the toil or frustration of a work you inherently cheapen it, at least to a degree. And this may be particularly true of a creative work like music; I believe you have to sand your own wood in this regard.

Your third paragraph brings up mixing ai art with man-made music. I have little to say here: I find these to be two different sins, that should be handled somewhat separately. As long as the poster discloses it. I disapprove of using ai are, but as they're not passing themselves off as artists, I can only fault them for a personal failing not a creative one.

To return to my 'torture-stake'. I think life is very convenient for a great many people, (I'm speaking in regards to those who live in the U.S.A. and countries of similar living stander, I'm not making broad statements here.) and ai making it more convenient is I believe somewhat immoral. Regarding it being uses to find answers and information: I think it has only short term gains and long term detriment. Knowledge on most anything is reasonably available and abundant. If you find yourself unwilling to take the time to 'learn something right', I find myself wondering if you deserve that information. I been told it also has negative effects on short/long term memory, but this is second-hand, so I'll move on from that point. I firmly believe ai cheapens information, which is criminal. My opinion extends to the effect it has on music, by taking out the 'busy-work', though not as strongly or with as much conviction, but there is a point where if you make work so easy to accomplish, that you no longer deserve the work.

P.S. This is my first time putting these opinions to writing, I believe nor did I proof read, I must apologize if at any point my writing seems overly stilted or disjointed.

 

[IUnknown]

Removing the toil or frustration of a work you inherently cheapen it, at least to a degree

Maybe Dreamtonics got inspiration from you and decided to make their editor less intuitive in the SynthV2 release because 'harder = more genuine work' lol.

I get what you're saying about effort, but I wouldn't use 'struggle' alone to measure art's validity. People always assume the technological barrier of entry they grew up with is the 'correct' one, and anything that makes it easier is a sin.

You're arguing that people need to 'sand their own wood' to earn the music. But traditional musicians said the exact same thing in the 80s when synthesizers and drum machines came out. They said pressing a key to get a string sound removed the 'toil' of learning the violin, and therefore it was cheap and soulless.

 

[lIlI]

Maybe Dreamtonics got inspiration from you and decided to make their editor less intuitive in the SynthV2 release because 'harder = more genuine work' lol.

I get what you're saying about effort, but I wouldn't use 'struggle' alone to measure art's validity. People always assume the technological barrier of entry they grew up with is the 'correct' one, and anything that makes it easier is a sin.

You're arguing that people need to 'sand their own wood' to earn the music. But traditional musicians said the exact same thing in the 80s when synthesizers and drum machines came out. They said pressing a key to get a string sound removed the 'toil' of learning the violin, and therefore it was cheap and soulless.

For myself, the removal of effort isn't significant, but the removal of creative decision makes AI art worthless. Some of my favourite art is very simple, but hugely impactful because of the meaning and intent behind it. I think AI users see their prompts as decisions, and underestimate the value of the thousands of tiny choices that go into the act of creation.

Before AI art generators, there were more amazing artists than I could ever hope to discover, more incredible music than I had time to listen to, and free tools to create my own were all available to me.

After AI, RAM prices are higher.

 

[IUnknown]

but the removal of creative decision makes AI art worthless

If you want to debate whether AI use can still qualify as art, then sure.

In my opinion, for something to be classified as art, it has to involve either craftsmanship or resonance.
So a painting is art, but Fountain by Duchamp (which was simply a urinal on display) is also art due to the resonance it had in art theory. Notice that it required virtually no effort to produce, as it was a readymade object, yet it went down as one of the most important art pieces in 20th-century art.
Why can’t a hastily generated AI artwork achieve the same?

Now, onto a more serious argument:
The main idea is that AI can (and is) used in ways that don’t completely automate thinking and craftsmanship, but rather do the heavy lifting for you.
I think AI VocalSynth software are a very good example of this. Their AI engine are end-to-end and do a lot of the heavy lifting in making a voice sound realistic. some even generate the notes for you from an acapella. Yet it’s not “AI slop,” as you can still manually tune it -- human intervention is still present.
So the question now is: how far can you push it in drawing, instrumental work, and lyric creation? According to some people in this thread, any use of it is a sin. I simply have more tolerance for the use of AI before I would call the entire work “slop.” Hence my original quote:

"Using AI for the MV or the instrumental isnt THAT bad"

 

[bibbs]

things that we call "art" also tend to have a very granular level of detail. when you prompt ai to make something it will choose a note or word just because the model is designed to spit out what is most statistically plausible to come next. but for something by a human, a single note or word can hold a huge amount of signficance. obviously, in practice artists work together, you commission an editor and an artist for a music video to your song and leave it to their judgement. delegating like that has a similar purpose as using ai, but what does ai do better than a human artist?

in theory.. ai is a lot cheaper than commissioning a real artist, but i wonder about that. we're already facing the cost of ram and ssd prices and these ai companies are subsidized by venture capitalists and even the government right now. these companies are running at *very* sleep losses using a very similar business model as social media companies: get everybody using it and figure out how to make money later. between the costs of what ai will look like when the vc money dries up, the cost of having to live with data centers draining the water supply and hogging the power grid, the effects on other markets like ram, gpus, and ssds, and the privacy concerns, is it actually more inexpensive than commissioning an artist, or is that just an illusion?

or maybe they can make money the sam altman way.. but gl with that


here's some quotes that i was responding to, but it all got kind of mixed together in my response

There're also the environmental impacts of AI usage, but nobody wants to talk about that, and neither do I. But the biggest part to me is that if you're using AI, you aren't really making anything. Use it for the instrumental? That's like half the song that you're not making, and at that point it isn't really "your" song, it's the AI's song. You didn't make it, something else made it. Same with other AI generated parts, especially vocals. If you use AI for the vocals, what's even the point of making it?

I get what you're saying about effort, but I wouldn't use 'struggle' alone to measure art's validity. People always assume the technological barrier of entry they grew up with is the 'correct' one, and anything that makes it easier is a sin.

You're arguing that people need to 'sand their own wood' to earn the music. But traditional musicians said the exact same thing in the 80s when synthesizers and drum machines came out. They said pressing a key to get a string sound removed the 'toil' of learning the violin, and therefore it was cheap and soulless.

For myself, the removal of effort isn't significant, but the removal of creative decision makes AI art worthless. Some of my favourite art is very simple, but hugely impactful because of the meaning and intent behind it. I think AI users see their prompts as decisions, and underestimate the value of the thousands of tiny choices that go into the act of creation.

Before AI art generators, there were more amazing artists than I could ever hope to discover, more incredible music than I had time to listen to, and free tools to create my own were all available to me.

After AI, RAM prices are higher.

 

[Beananium]

When most Vocaloid producers and artists are called out for using AI, its always generative AI. Something like ChatGPT calls back to those data centers that are making life more expensive to all and more unhealthy and dangerous for people living nearby them. Meanwhile, most forms of AI thats localized to a program or not on a webpage often does not call to those same data centers.

I also agree that some features of a program using artificial intelligence would not detract from usage in a piece. There are some tedious tasks that most are able to do, but can easily be automated with them with no control over the output being lost.

Using an optimization feature built into Synth V so the exported vocals don't have choppy breaks when a new section ends and starts, and asking Grok to generate a picture of Miku to slap onto your song about Hawk Tuah are not equally bad. It gets to the point where that usage is used as a substitution for a product that requires skills honed by experience, rather then optimizing a element that has only one way to do it to cut down on time.

 

[WacoWacko39]

Maybe Dreamtonics got inspiration from you and decided to make their editor less intuitive in the SynthV2 release because 'harder = more genuine work' lol.

I get what you're saying about effort, but I wouldn't use 'struggle' alone to measure art's validity. People always assume the technological barrier of entry they grew up with is the 'correct' one, and anything that makes it easier is a sin.

You're arguing that people need to 'sand their own wood' to earn the music. But traditional musicians said the exact same thing in the 80s when synthesizers and drum machines came out. They said pressing a key to get a string sound removed the 'toil' of learning the violin, and therefore it was cheap and soulless.

I don't know who Dreamtonics is :P.

Maybe you're right about the technology part regarding me (I'm driving cars from the Clinton years XD). But there's a difference in those of the 80s you mentioned and what is being discussed here. They were against easing the creation portion, I am against easing the creative portion. Firstly I must state where I stand, I do not completely approve nor condemn those who use ai to "do the heavy lifting for you" as you said in in post number #2,414. A person who uses ai in this *creation* portion' should receive little or no scrutiny. But some one who uses it in the *creative* portion should be heavily scrutinized.

The creative portion is what should reflect the creator, and if they don't care enough to make it why should I? Music is meant to be a creative form isn't? And if they only used it for a portion, well they included it with the rest, deeming it appropriate. In a sense (not to quote scripture) a little levean ferments the whole batch.

How easy the creation process should be allowed to be respectfully lowered, is an interesting question that I don't have the answer to. But the creative should be left in the hands of man as much as reasonably possible.