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

[BONES]

Ace Studio 2.0 is easily affordable. Even the Pro version is only $22/month (purchase plan). Lots of people spend more than that on DAW and Plugin purchases.

That should read "have spent". We've already invested a lot of money in DAWs and plugins. That, combined with the hoarder mentality many people around here exhibit, I think will create considerable resistance to making all that obsolete. I think it's also clear how resistant people are to subscription models, which may also hamper the uptake of AI tools.

We're probably leveraging AI more than anyone else around here at the moment but we have no intention of abandoning our DAW and plugins. Yes, AI right now is a great tool for generating ideas, in the form of (sort of) complete songs, but it's a long way from being able to finish anything to the precise level we're after and I can't imagine it getting to a point where it ever could, simply because I don't really see a big market for that amongst musicians and producers, not enough for anyone to take it seriously. They'll go after different markets, maybe ordinary people who just want to make their own music for their own listening pleasure and/or to share socially with their friends. The Muzak and Stock Music industries would be other targets for AI music, I reckon, where there is a clear path to profit and there'll be far less resistance to the change. Maybe music for games, too.

The other thing, of course, is that even if the plugin market shrank by 50%, we'd still be spoiled for choice. Even at someone's predicted 80% shrinkage, I think we'd still be well catered to for plugins because, let's be honest, right now we are living in an era of chronic over-supply.

[bermudagold]

Ace Studio 2.0 is easily affordable. Even the Pro version is only $22/month (purchase plan). Lots of people spend more than that on DAW and Plugin purchases.

That should read "have spent". We've already invested a lot of money in DAWs and plugins. That, combined with the hoarder mentality many people around here exhibit, I think will create considerable resistance to making all that obsolete. I think it's also clear how resistant people are to subscription models, which may also hamper the uptake of AI tools.

ur probably right in that the pivot will take longer than people think...music press was pushing this back in 2023...look how long it took to convince people
viewtopic.php?p=8776347&hilit=gaw#p8776347
but the pivot will probably be real...wasn't much interest on kvr then...things and people change...yet things and people stay the same

[Zeisner]

Ace Studio 2.0 is easily affordable. Even the Pro version is only $22/month (purchase plan). Lots of people spend more than that on DAW and Plugin purchases.

But those costs come on top of everything else. You still need a proper DAW and plugins and instruments etc to completely redo the AI output just because of all the artifacts. So ultimately you're spending even more money than ever - on music you can't even own legally.

[pdxindy]

Ace Studio 2.0 is easily affordable. Even the Pro version is only $22/month (purchase plan). Lots of people spend more than that on DAW and Plugin purchases.

But those costs come on top of everything else. You still need a proper DAW and plugins and instruments etc to completely redo the AI output just because of all the artifacts. So ultimately you're spending even more money than ever - on music you can't even own legally.

My speculation was that AI based DAW's like Ace Studio will replace traditional DAW's. I didn't claim it already happened.

But even today, many of those artifacts can be mitigated right in AI DAW's like Ace Studio. For example, all the midi in Ace Studio is editable, and doing so fixes most vocal artifacts. With some human editing, the results are solid and suitable for many uses. If someone wants to make an intro for their podcast, it's plenty good. There's lots of low to mid level work that an AI DAW is already suitable for.

And the cost is not that much. Just ignore the annual Komplete upgrade and you have just about paid for it. A user can still keep their regular DAW and plugins. (Ace Studio has a bridge plugin to integrate with another DAW) And then the user can transition over to an AI workflow over the next few years. I'm sure it's already happening.

[Zeisner]

But even today, many of those artifacts can be mitigated right in AI DAW's like Ace Studio. For example, all the midi in Ace Studio is editable, and doing so fixes most vocal artifacts.

By doing so you only replace AI-based automation artifacts with editing-by-mouse automation artifacts. You would still have to record a real human performance. The audio artifacts also remain no matter how much you edit automation, thanks to bad training material and model limitations.

And the cost is not that much.

It is in the long run, especially because it's getting harder each day to generate any income with music, a trend that started long before the great AI hype. Look at how many people here are forced to sell their licenses.

And then the user can transition over to an AI workflow over the next few years.

I doubt that skilled and talented musicians will accept such a downgrade. They rather go back to hardware if they can afford it.

[dionenoid]

How if almost nobody is willing (or able) to spend the aditional costs? So far, the only people who have subscriptions to AI services are researchers, everybody else is using free accounts. I doubt that including ads will result in the revenue AI companies are looking for.

Suno announced a few days ago that they reached $300 million in annual revenue, and that they currently have 2 million active paying subscribers :

https://www.forbes.com/sites/conormurra ... ash-grows/
https://www.musicbusinessworldwide.com/ ... l-revenue/

To put this in perspective : They have 10 times as much paying subscribers as ProTools. And 10 times as much revenue as Imageline.

[Zeisner]

To put this in perspective : They have 10 times as much paying subscribers as ProTools. And 10 times as much revenue as Imageline.

But you have to compare that with the sum of all other (classic) commercial DAWs and plugins and hardware/instruments and hiring session players/audio engineers (because Suno is meant to replace all of them). And you need to consider that Suno lost against GEMA, forcing them to introduce a new model this year. Chances are high it will perform worse without all the illegaly scraped data from sources like Anna's Archive. And if Suno keep parts of their old model they will successfully get hit with more lawsuits from collecting societies.

Keep in mind that none of those AI companies have a meaningful long-term business plan, it's all about generating as much money as fast as possible, selling it right before everything blows up. Just the usual tech startup scheme.

[pdxindy]

By doing so you only replace AI-based automation artifacts with editing-by-mouse automation artifacts. You would still have to record a real human performance. The audio artifacts also remain no matter how much you edit automation, thanks to bad training material and model limitations.

If you want to say it isn't good enough for you, fine. However, you're trying to speak for everyone and not everyone agrees with you.

For various use cases, there's no need to record a real human performance. The AI generation is plenty good.

Regarding artifacts, people release music all the time that has artifacts, some of it quite successful. Every standard DAW has features that create artifacts. Sometimes the artifacts are even the point.

AI DAW's are already good enough to satisfy many users needs and in 3 years, it will be even better.

[Tiles]

To put this in perspective : They have 10 times as much paying subscribers as ProTools. And 10 times as much revenue as Imageline.

But you have to compare that with the sum of all other (classic) commercial DAWs and plugins and hardware/instruments and hiring session players/audio engineers (because Suno is meant to replace all of them). And you need to consider that Suno lost against GEMA, forcing them to introduce a new model this year. Chances are high it will perform worse without all the illegaly scraped data from sources like Anna's Archive. And if Suno keep parts of their old model they will successfully get hit with more lawsuits from collecting societies.

Keep in mind that none of those AI companies have a meaningful long-term business plan, it's all about generating as much money as fast as possible, selling it right before everything blows up. Just the usual tech startup scheme.

You’re bundling several different claims together.

Suno is not realistically replacing every DAW, plugin, hardware instrument, session musician and engineer. That’s an exaggeration. Tools like Ableton Live or Logic Pro didn’t eliminate musicians either. They lowered barriers and changed workflows. AI tools are another layer in that evolution, not a total substitute.

Comparing Suno’s subscription to “the sum of all commercial music production” is economically flawed. Most musicians don’t own the entire industry’s toolchain. The relevant question is marginal utility for specific users, not theoretical full replacement cost.

Also, Suno does not represent all AI music. There’s a huge spectrum ranging from fully generative systems to assistive tools inside DAWs, stem separation, mastering aids, MIDI generators and so on. Even if one company fails, the broader field doesn’t disappear.

Legal pressure from collecting societies like GEMA is part of how new media technologies get integrated. Litigation happened with Napster, YouTube and others. Whether training data practices change says nothing about the long term viability of AI as a category.

As for “AI music is bad”: that assumes a fixed gatekeeper definition of what counts as music. Historically, every production shift was dismissed at first. Drum machines, sampling and loop based software like MAGIX Music Maker all lowered the skill barrier. The result wasn’t the death of music, but more participation at different quality levels.

Lowering effort thresholds does not eliminate high skill music. It just expands who can participate. Taste is not an ontological filter.

History suggests adaptation, not annihilation.

[Zeisner]

However, you're trying to speak for everyone

Absolutely not! I'm aware of the fact that I belong to a small and (unfortunately) shrinking minority. I'm just tired of those "AI is already on par with top quality production" comments. I also don't want to be the old man yelling at clouds here (Which would be weird, considering that I'm much younger than the average KVR user if today who falls in the age bracket between 60 and 70). It's nice though to talk to somebody for once who's (still) able to hear the difference.

Regarding artifacts, people release music all the time that has artifacts, some of it quite successful.

I just don't know a single example of an AI artist who has been successful beyond the usual fraud or ad revenue trick. No human seems to actually buy songs or albums from The Velvet Sundown or Let Babylon Burn, they just let (free) Spotify or YouTube play in the background and sometimes AI music like that pops up. The vast majority of listeners are just bots. Dead internet.

Every standard DAW has features that create artifacts. Sometimes the artifacts are even the point.

But they sound completely different. AI artifacts don't follow (necessary) psychoacoustic patterns which make them even more annoying than hiss or aliasing, they're too erratic. At least if you're (still) able to hear all of that of course.

AI DAW's are already good enough to satisfy many users needs and in 3 years, it will be even better.

Barely because of technical limitations (which I explained in a lengthy comment before). But I wouldn't be surprised if AI-generated music would still sound much better to the average human listener thanks to increasing hearing damage, reduced brain activity, lack of training (from not listening to old records and real instruments) and declining attention spans.

[BONES]

You would still have to record a real human performance.

Why? Industrial music, for example, is all about removing the human element as thoroughly as possible. Using a mouse to do your automation is perfection, every value precise. Playing things in is sloppy, it adds too much useless data into the sequencer and you have to go in and tidy it all up afterwards. It's all too messy.

It is in the long run, especially because it's getting harder each day to generate any income with music, a trend that started long before the great AI hype.

It's not been easy for 40+ years unless you were willing to sell out to do something commercially popular. Even that's hard now, which just feels like justice to me.

I doubt that skilled and talented musicians will accept such a downgrade. They rather go back to hardware if they can afford it.

Then they will inevitably be left behind, just as in other industries. That's just how evolution works.

Suno announced a few days ago that they reached $300 million in annual revenue, and that they currently have 2 million active paying subscribers...

To put this in perspective : They have 10 times as much paying subscribers as ProTools. And 10 times as much revenue as Imageline.

That's what you get when you cater to the masses, rather than a bunch of irrelevant, whiny toads like us.

But you have to compare that with the sum of all other (classic) commercial DAWs and plugins and hardware/instruments and hiring session players/audio engineers (because Suno is meant to replace all of them).

And for 2,000,000 people that seems to have happened already, although it's likely almost none of those people have ever used a DAW. Most of them probably can't play an instrument, either. But just as ProTools is just one DAW, Suno is just one music generating AI so it doesn't have to do 100% of the heavy lifting, it's single piece of a larger puzzle.

I'm just tired of those "AI is already on par with top quality production" comments.

Poor baby, does it hurt that you've been replaced? It was inevitable, most jobs like yours were a consequence of the bloated music industry of late 20th Century, they were never going to survive anyway. They are skills any idiot can learn by doing.

I just don't know a single example of an AI artist who has been successful beyond the usual fraud or ad revenue trick.

How many artists did you know who were using a DAW 2 or 3 years after Cubase was released? Dominance takes time but is inevitable.

The vast majority of listeners are just bots. Dead internet.

Pure speculation unsupported by any actual evidence. How, for example, do you generate 12,000 fake comments on a single YouTube video without Google tumbling to your scam and shutting you down?

But they sound completely different. AI artifacts don't follow (necessary) psychoacoustic patterns

Which makes them a refreshing change, something a bit different. In 10 years time, when AI music is pristine and perfect, people will be prompting it to add in those old school '20s artifacts for a dose of authenticity.

Barely because of technical limitations

Which will be overcome, as always happens with new technologies. You are naive in the extreme to think otherwise.

[BertKoor]

I propose we call algorithmically produced music (or muzak) from now on "Industrial", referring to the method.
Yay, paradigm shift !!!

[pdxindy]

I'm just tired of those "AI is already on par with top quality production" comments.

I didn't say that. Perhaps save your rebuttals for someone who does?

[pdxindy]

I just don't know a single example of an AI artist who has been successful beyond the usual fraud or ad revenue trick. No human seems to actually buy songs or albums from The Velvet Sundown or Let Babylon Burn, they just let (free) Spotify or YouTube play in the background and sometimes AI music like that pops up. The vast majority of listeners are just bots. Dead internet.

There are many people making music that has nothing to do with Spotify... music for games, tv commercials, events, etc., etc. I have a DAW and plugins and enjoy crafting music with them. However, if it was a friend's birthday and I wanted to write a quick song with lyrics just for them, I would happily use an AI DAW. With minimal effort, everyone at the party would love it and nobody would care how it was made.

I bet lots of songwriters will be happy with AI DAW's. In Ace Studio, you can sing in your lyrics and turn it into a credible rendition of your intended song in a day or less of working on it.

Likewise, someone who makes documentary videos and wants to create soundtracks for minimal cost would also find it a useful tool.

[jamcat]

I was listening to some T-Pain and I noticed he had pitch artifacts all over his voice. This must be why nobody bought his record.