How much potential remains for developing new non-AI audio plugins, business-wise?

[sergeant]

The industry is so saturated, and AI is getting better so fast. If I try to get into the industry at this point, will my plugin ideas be obsolete by the time I get to finish developing them? Is it too late to start an audio plugins business in 2024?

[kerfuffle]

I'm in the same boat and the answer seems to be that you need to innovate in order to stand out (which has always been true). Not all new ideas need AI, but AI does have the most obvious new applications.

[gambero]

I can't really tell if it's a good time to get into the audio plugin industry, as I have a couple of conflicting thoughts, but I'll still share them because I think they'd make an interesting discussion in this subject. I don't have nearly enough business and work experience to give any advice, but maybe some of you could help me get my ideas together.
First of all, the industry is very saturated, both in terms of plugin manufacturers and plugins themselves. It has been this way for quite a while, so maybe trying to "make it big" is not the best strategy. I've seen that offering really good products for a small, niche, but very loyal userbase can work out quite well. To me, being liked a lot is way better than being liked a little by many, but that's just my opinion.
Regarding AI, I'm honestly tired of it, but I'll try to keep that feeling away for this. It does provide a new and interesting approach to audio processing in general, but I think that a lot of users are quite skeptical, given all the buzzword marketing that's been going on for the past couple of years (plugins claiming to use AI when they don't, tons of low quality services and webapps, everybody trying to cash in). At the end of the day, if it sounds good and it's nice to use, it's a good plugin, be it with AI or not.

[whyterabbyt]

If your plugins dont innovate somehow versus what's available, or you cant market the shit out of them via YT inflatulencers, then (commercially speaking) there's probably no point.

Do something better, or more original, or that hasnt been done, or even emulate a novel bit of hardware that everybody lusts after, then yeah, maybe there is. Not a guarantee though, so keep your primary income.

[JustinJ]

The industry is so saturated, and AI is getting better so fast. If I try to get into the industry at this point, will my plugin ideas be obsolete by the time I get to finish developing them?

The industry being 'saturated' means it's thriving but you cannot expect to make something mediocre and expect it to be a success. It means you must make something exceptionally good - quality wise -AND find routes to market.

You will now find this the case with any technological endeavour. This is nothing new.

However, I'd argue that it doesn't need to have brand new technological innovation. It just needs to do some things better than encumbants, better workflow, novel features, look nicer, address existing issues. Offer market value.

Up your game. Compete by being competitive.

[Urs]

Yeah... I remember the market being "saturated" when there were fewer than, hmmm, 500 plug-ins available in total. Like "What? Who needs another synth plug-in in 2002?"

In my view, the companies who've got a problem are those whose demographic of customers retiring is larger than the demographic of people entering the music scene.

But who knows what the future holds. If AI is useful for anything, it's probably language. One stand out feature for a company or brand could be absolute localisation. Develop your otherwise non-AI plug-ins so that the UI and built-in documentation works in any language that whatever AI translator supports. Tap into markets that no-one has gone into yet.

[jupiter8]

Is the market for "AI" plugins really that big? Will it ever be?

[JustinJ]

Is the market for "AI" plugins really that big? Will it ever be?

It may seem gimmicky at the moment, but personally I can foresee a time when plugins will have AI features as standard. As will most software. Then we'll look back at all the tedious grunt work that music makers had to do with sound design and mixing etc and wonder how we ever managed.

The big mistake to make with this sort of thing is to imagine the future will be like it is now based on current limitations.

[kerfuffle]

Is the market for "AI" plugins really that big? Will it ever be?

This is only one data point but there are currently almost 15000 models on https://tonehunt.org/ for the Neural Amp Modeler plug-in. This is a free plug-in and the quality of the models varies, but it's clear that people are willing to train their own AI models for reproducing amps and pedals.

[VariKusBrainZ]

The industry is so saturated, and AI is getting better so fast. If I try to get into the industry at this point, will my plugin ideas be obsolete by the time I get to finish developing them? Is it too late to start an audio plugins business in 2024?

If AI is so great why bother. Some AI bot will come up with better ideas than you, from your attitude

[jupiter8]

Sure it will have some impact but how much? Some think it'll replace everything but i just don't see it. Maybe it's the luddite in me. Most of the interest still seems to be replicating the past.

And with that said i give you my most moist boomerludditetake on the whole thing: imagine being in a studio in 1984 and thinking about how the future will be in 40 years. And the answer is: like now only slightly worse but way way waaay cheaper.

[jupiter8]

Is the market for "AI" plugins really that big? Will it ever be?

This is only one data point but there are currently almost 15000 models on https://tonehunt.org/ for the Neural Amp Modeler plug-in. This is a free plug-in and the quality of the models varies, but it's clear that people are willing to train their own AI models for reproducing amps and pedals.

Yeah sure, if you count that way. To me it's still one plugin. How many of the 100 latest or most popular plugins use AI?

[Lind0n]

Is the market for "AI" plugins really that big? Will it ever be?

This is only one data point but there are currently almost 15000 models on https://tonehunt.org/ for the Neural Amp Modeler plug-in. This is a free plug-in and the quality of the models varies, but it's clear that people are willing to train their own AI models for reproducing amps and pedals.

The first issue here is this is a "snapshot" based plugin - its an amp - but with defined and fixed knob positions. So its not really modelling an amp that much, its modelling an amp in a particular state, the usefulness of this is probably more marginal than the hype might be suggesting. It seems to be post processing this with std controls for bass, middle, treble etc.- which mean its not dealing with the amp dynamics correctly.

Next I think we need to separate out Machine Learning and Neural Network based AI from "AI", dismissing AI in total because Neural nets and Machine Learning are poor in a domain is not probably what we should be doing - given that, ML and NNs are useful for some things - and Neural DSP seem to have found an approach to model all the controls in an amp - so many things are possible, but right now they are very very domain specific, so a generalised AI overlord seems a bit of a way off...as does "my AI builds my plugin for me...", anyone who's even a reasonable level coder that has looked at ChatGPTs coding efforts will see its filled with errors a lot of the time, and the amount of effort required to get "decent code" out of it is probably about = the amount of effort to "just write the code".

[JustinJ]

...anyone who's even a reasonable level coder that has looked at ChatGPTs coding efforts will see its filled with errors a lot of the time, and the amount of effort required to get "decent code" out of it is probably about = the amount of effort to "just write the code".

The thing is that only a short while ago interfacing with AI to write code wasn't even possible at all. Now it's got utility if you know how to use it. For example, I've given it chunks of code and asked it to review it and make suggestions, with fair results. It's reasonable at reviewing code in general too. It's good at suggesting comments and other documentation.

And it's not going to stay at this level for very long. It's not good at doing task X...until suddenly it is. There's going to be a lot of that coming.

[Kalamata Kid]

In the next five years:
Will producers/musicians use a prompt to compose music?
Will prompts be available to purchase?

Will developers make more specialized AI plugins that
producers/musicians will use to compose music?

Beyond ten years?