If AI replaces musicians, does the entire plugin industry die with them?

[VitaminD]

Ambient, Techno, Bubblegum Kawaii 8-bit nintendopop are all heavily reliant on digital pathways with minimally complex patterns (in general). I'm pretty sure even today an AI model could generate hundreds of believable tracks in these type of genres and no one would bat an eye. They're all largely derivative anyways so that is easy work for AI which only does derivative work.

Lyria3 produced this with the input "Bubblegum Kawaii 8-bit nintendopop song mp3"

You're all welcome.

[Tubeman]

Ambient, Techno, Bubblegum Kawaii 8-bit nintendopop are all heavily reliant on digital pathways with minimally complex patterns (in general). I'm pretty sure even today an AI model could generate hundreds of believable tracks in these type of genres and no one would bat an eye. They're all largely derivative anyways so that is easy work for AI which only does derivative work.

Lyria3 produced this with the input "Bubblegum Kawaii 8-bit nintendopop song mp3"

You're all welcome.

That really sounds nothing like 8-bit music.

[Bunny_boy]

I used the same term and the song produced is just Disney pop bollocks.

[VitaminD]

Ambient, Techno, Bubblegum Kawaii 8-bit nintendopop are all heavily reliant on digital pathways with minimally complex patterns (in general). I'm pretty sure even today an AI model could generate hundreds of believable tracks in these type of genres and no one would bat an eye. They're all largely derivative anyways so that is easy work for AI which only does derivative work.

Lyria3 produced this with the input "Bubblegum Kawaii 8-bit nintendopop song mp3"

You're all welcome.

That really sounds nothing like 8-bit music.

It does not.

But it does sound like what I'd expect Bubblegum Kawaii nintendopop to sound like. Now it is in your mind. muah ha haaaa..

One other limitation I see of that model, is it doesn't seem to be able to create single instrument solos outright. At least not with the free non-logged in version. It makes 30 second clips that includes percussion even if you tell it directly no drums or no percussion, just (for example) a bass solo.

I kind of question what it fed off of to generate what it returns too. Will it be too close to something else copyrighted that I'm going to get strikes or lawsuits?

[havran]

If AI really does replace a large part of music creation, who is actually going to buy plugins anymore?

At the moment I'm more worried about the whole economy being disrupted by AI (including employment, with no obvious public-policy scaffolding in place), not just the music-plugin sector.

Who will still have a job and spare money to support plugin devs a year from now?

How many people will be stuck with houses they can no longer afford, but which they cannot sell?

How many people are going to participate in Black Friday sales if they think they might get laid off just before Christmas (okay, sure, that might be specific to the IT sector and maybe some other business sectors)?

If not for that, to me it seems there are a lot of hobbyists like me who just like to play with these things.

As for the working professionals, I imagine and hope for a rebound, where it would be a point of honor and a selling point (for those who can afford the luxury of going to the movies) for a movie (for instance) to brag that they have a human composer for the score and that the orchestral pieces are played by human musicians -- admittedly, that seems a little idealistic even to me.

[Zeisner]

A mischievous trick; a prank.

Which proves my point. All you can get out of LLMs is a joke. And not even good ones. Totally useless, just like their users.

[BONES]

I didn't comment on using AI tools, but AI generated music.

I understand that but I was commenting about AI generated music, because that's where/how our process starts. We get Tunee to create entire songs for us, songs that we would be 100% confident in releasing as is. As I've said before, if I didn't know where it came from, it would be my new favourite band and I'd be hoovering up everything they put out like a teenage fanboy.

A lot of us aren't so fixated with negativity and our attention is more attuned to the wonder of life.

I reckon I am probably far more attuned to that stuff than you are, I just live in the real world, the world that would trample the wonder to death for it's own ends and not care in the slightest.

Come on in, the water is warm, there's no need to be scared.

Sharks love the warm water. We had four bull shark attacks in Sydney in a week recently. I'm staying above the water for the time being, thanks.

just sayin'...

Which is precisely why I said "if they were starting out today". McCartney is an old, old man now, like most of you he is too set in his ways.

I'm thinking such "musicians" should give credit to the AI they used to make the music.

Absolutely. We've always given our DAW credit in our sleeve notes and Tunee will absolutely be getting co-writing credits for the next one. Anything less would be dishonest.

Who will still have a job and spare money to support plugin devs a year from now?

Yeah and by 1980 the paperless office was going to be a reality for every worker. 45 years later, it still hasn't happened. Jobs go from one place and new ones pop up in others. It's how it always goes.

[Zeisner]

At the moment I'm more worried about the whole economy being disrupted by AI (including employment, with no obvious public-policy scaffolding in place), not just the music-plugin sector.

It looks like we're heading this way. Total crash.

[ghettosynth]

Ambient, Techno, Bubblegum Kawaii 8-bit nintendopop are all heavily reliant on digital pathways with minimally complex patterns (in general). I'm pretty sure even today an AI model could generate hundreds of believable tracks in these type of genres and no one would bat an eye. They're all largely derivative anyways so that is easy work for AI which only does derivative work.

Lyria3 produced this with the input "Bubblegum Kawaii 8-bit nintendopop song mp3"

You're all welcome.

Yes, but, and this is coming from an appreciation of minimal techno/house. It has a very limited palette in those domains. Bones' assertion that it will be more successful in traditional song structures is correct. That actually is something of an averaging effect, the model tries to steer towards expected outputs. You have to constrain it to stay out of those lanes. Getting the model to consistently produce reliable facsimiles of standard genres is child's play; getting it to reliably produce variety in domains where rules are broken and variance is subtle, is far more work.

Also Google's model is really quite shit IMSHO, but then, who am I to judge.

I think that this might be the intended use case, quick corporate advertisements like this video that celebrates Botox.



I can't take credit here. This campaign was the suggestion of one of the artists on my current roster, one Miss Chablis Supreme, who is way ahead of the game in supporting her man's choices.

[guitarzan]

Well, I have three minutes to spare, so I fed a couple half-baked prompts to the new Gemini Lyria beta thing about chiptunes 8-bit type stuff either morphing into or combined with analog synth tones and styles. I have no appreciation for 8-bit synth tones so I have no opinion on these two results, but in my earlier tests I really liked the instrumental output of this new Beta.

https://g.co/gemini/share/e3739aa5f733

https://g.co/gemini/share/729cc15d8674

[ghettosynth]

A mischievous trick; a prank.

Which proves my point. All you can get out of LLMs is a joke. And not even good ones. Totally useless, just like their users.

Ohh, no, not quite. There is bit of a point in there, I'll give you a bit longer to see if you can catch it. Read the joke post carefully, as an expert, it should jump out at you.

[ghettosynth]

So, these posts kind of got buried and forgotten, I thought that I would bump them so that we can continue our very serious and on-topic conversation about "plugin ideas." I combined both into a single post to make it easier to quote them, if one wishes. I unquoted the image though, so that it's easier to read.

I think that is the solution that was described in 2)

No, it's not. My approach does not change anything about dynamics nor is it M/S based.

I took a look. So, when you say it’s not M/S based, do you mean you didn’t explicitly use an M/S encode/decode block, or are you saying it’s structurally different in the linear-systems sense? In other words, since your construction looks like “a common signal in both channels plus an anti-symmetric add-on that is added to one side and subtracted from the other (so it cancels when you sum to mono),” what property of the construction prevents it from being rewritten as a sum/difference (change-of-basis) representation?

You know, you can just admit that you are not very experienced at prompting.

Well, if you're so experienced at prompting, then why didn't you try to solve the provided test with the help of your precious LLM?

Oh, there's a very good reason. I'm gonna let you think about what that might be.

Here's one, what do you think of this?

[Zeisner]

The level of stupidity is astonishing. You aren't even able to take a look at the Wikipedia page of the Haas effect to find out what it actually is. No, it's NOT ITD. So once again you and your precious LLM are wrong. As wrong as anybody who writes crap like "Flip the phase" instead of "Invert the polarity". Those are two different things. You understand just as much as a LLM - nothing. Aside from the fact that you needed this long to come up with a wrong text - so much about "power prompting".

By the way, "your" technique is not stereo compatible and therefore useless for stereophony (!) while mine is both mono and stereo compatible. Something a thinking person would have realized after replicating and listening to it (with and without the non-delayed copy). Which you obviously didn't. A thinking person would also have realized that the delay times in the text are WAY too high (because, well - it's about stereophony featuring ITD, not creating the Haas effect). You're just another AI zombie who takes everything a LLM says at face value no matter what, unable to think for yourself.

So what you're going to do now? Learning can be ruled out, that's for sure. Maybe pretending it was just another joke?

[VitaminD]

Ambient, Techno, Bubblegum Kawaii 8-bit nintendopop are all heavily reliant on digital pathways with minimally complex patterns (in general). I'm pretty sure even today an AI model could generate hundreds of believable tracks in these type of genres and no one would bat an eye. They're all largely derivative anyways so that is easy work for AI which only does derivative work.

Lyria3 produced this with the input "Bubblegum Kawaii 8-bit nintendopop song mp3"

You're all welcome.

Yes, but, and this is coming from an appreciation of minimal techno/house. It has a very limited palette in those domains. Bones' assertion that it will be more successful in traditional song structures is correct. That actually is something of an averaging effect, the model tries to steer towards expected outputs. You have to constrain it to stay out of those lanes. Getting the model to consistently produce reliable facsimiles of standard genres is child's play; getting it to reliably produce variety in domains where rules are broken and variance is subtle, is far more work.

My original point was, at the start of this thread, and still is that the more formulaic the genre, the easier it is going to be to copy. That shouldn't be a controversial statement, I think it is rather evident.

I think what you said fits in with what I think about how the AI models are going to allow youtubers and short film types get soundtracks/soundbeds for their video content with a bit more control on the style.

But I think for pure music, all the current AI models don't seem to have any design towards creating original works through fresh ideas and techniques. It can only mash together what it has in it's dataset.

If a new sound technique comes about in reality, say (as example) from speeding up a drum machine and applying micropitch shifts, and that technique wasn't recorded already in the dataset, how is the AI model going to invent that on it's own? It can't. It's not a creative tool in itself to create brand new things. It's a mashup and averaging tool, to create semi-new things.

I think we'll need some kind of AI tool that is designed and trained on a simulation of physical models of real instruments in a simulation of a physical room (size, materials, air pressure) that simulates a player to some degree. Then somehow let it 'explore' all possibilities to get new sonic elements. Then we'll need it to be able to give the AI tool a specific human personality in order for it to understand what is musically appealing and what is not.

Otherwise its just a dumb mashup machine.

[havran]

Who will still have a job and spare money to support plugin devs a year from now?

Yeah and by 1980 the paperless office was going to be a reality for every worker. 45 years later, it still hasn't happened. Jobs go from one place and new ones pop up in others. It's how it always goes.

I have the impression that AU (and to a greater extent NZ) exist in some kind of isochronic pocket that lags behind the rest of the world by a decade or two -- maybe a different kind of "On the Beach" experience is headed your way.