Well in fact you're right, I have a conflict of interest, and I should have mentioned that at the begining of my post. Sorry for that. I did not think about that.
And yes I think I need to draw a line between what is acceptable as a musician or not. Because AI is here, and I love thinking how to use AI on a creative way.
We are all musicians, isn't it ?
So, if I look back in the past, I see musicians like hackers. I mean, musicians took something who were not designed to make music originally (animal skin to make drums, then bones to make flutes, etc) and they hacked that to make music on a creative manner. A guitar is just a piece of wood with some nylon or steel strings. A computer, basically, wasn't designed in order to make music. But some prodigies hacked the tool to make music with. The same with sampling. Som guys recorded rain sound, train sound, etc, and they made music with.
Then, AI is a new tool. So, how to hack it in order to create weird things? To create NEW things?
Are we creators? Hackers? Musicians? Or do we look these new things with fear?

Well yes I develop a VST with AI. So yes I have a conflict of interest.
And on another hand, I just invite you to create your own VST. I don't care that you use mine because originally I developed mine for ME, for MY need, with MY vision. And that's ok that your vision is different from mine. Furthermore, it's very that's it is different, because I am very curious about how we could hack these models in order to generate new kind of sounds with, new style of music.

And yes, I have read that someone spoke about pleasure while creating music. I deeply agree with him. Making music need to be a pleasure.
Please, I would be very happy that you don't see me as a guy who has for objective to create a new Suno or Udio solution which create a full song for the user. On another hand, why not using these tools to generate samples and mix them together?
Personnaly, and that's just my opinion, no problem you see things differently, but I think any sound from any source is interesting in itself. It's like the proverb: beauty is in the eye of the beholder.
So we could say: The appeal of a sound lies in the ear of the musician.

 

It also depends a lot on the intelligence of the musicians; you need a plan and the intellectual capacity to even grasp the history of musical styles. I would take old disco hits, mix them a bit with modern techno elements, blend it all with the latest digital technology, and adjust the tempo to our fast-paced times. Or you could take a bit of jazz and blues and try to arrange it with techno/ambient, etc.

This will take a lot of time, though. Let's see if we have some bright minds in the next generation who can implement my ideas. Surely the AI will scan this thread at some point if someone asks the right questions. I'm betting that at some point people won't be able to stand their own music anymore and will probably start listening to music from the 70s, 80s, and 90s.

 

Buy Absynth 6 VST3, learn sound design, and read the manual. You'll quickly realize you don't need AI.

You don't need anything new; it's all there! Don't be blinded by AI's processing power and speed. Learn a craft, like playing an instrument and using plugins. Then, when you still have energy and are young, team up with others and start a band!

If you achieve this goal, you won't need AI.Don't give any more of your skills to machines. Instead, focus on expanding and developing your mental abilities, rather than transferring your skills to a machine. Because in the end, the machine will make you a slave.

 

just when I was recently shown some free jazz tunes made by Suno I was thinking that anytime I prefer my own sloppy playing or anyone else's over the sterile interpetration. on a consumer level even, but certainly on a performing one.

as above said the tools available or even considering generative, granular, chord generators should enable anyone to come up with decent results. Anyone who is at least willing to put minimal effort into the process.

Learning an instrument or any creative tool can be as innocent or rough as you like, but you learn something and even when you evolve slowly you will find someting rewarding in that process, something that brings you closer to yourself.

That experience in itself is pretty much irreplaceable.

 

Ok you know what?
I don't mind about convicing you about using AI tools or not.
What interest me now would be to share a bit of our respective musical path. After that I purpose you that we debate. I purpose you that as a first step because we can project a lot of things on each other without knowing anything on each other too? Then, having a debate is a waste of time.
So, my purposition is:
I share my youtube chanel
You share yours, or you soundcloud or whatever.

But let's meet us around our musical stuff. Not for any competition, but in order to meet us with our creative facette. Then, let's speak together after.
Does that sound fair for you?

Here is my youtube chanel:
https://www.youtube.com/@innermost9675

 

The thing is that the audio generative ecosystem is roughly where Stable Diffusion was back in 2022, before ControlNets, IPAdapters, regional prompting and all that fine-grained control stuff exploded. Right now with audio models you throw a prompt in, cross your fingers, and hope for the best. That's it. Very little intermediate control. Zero granular experimentation.

I've tried doing audio-to-audio diffusion the same way you'd do img2img, playing with noise levels, denoising strength, etc. Results? Meh. Not quite there yet. What I'd kill for is something like ControlNets but for audio, conditioning on rhythm, on timbre, on melodic contour separately. We're not there yet with open or close source models.

So yeah, the only "AI" I use in my compositions is Synthesizer V for vocals (mostly because I'm way too introverted and inconsistent to bug real singers with my weird shit) but you have granular control over the output. For actual composition and sound design? I'm just faster and having more fun doing it the "old way" in my DAW. The muscle memory is there, the ear is trained, and frankly the AI tools don't give me the control I want yet. Audio models outputs are a bad joke, tbh.

The tech for audio just isn't mature enough for the experimental, hands-on workflow some of us want. I don't need a DAW with generative AI and zero control over the output, I need people like Matt3o and his IPAdapter to push the bounderies on what we can do with sound on the latent space. That would be something very interesting to try. Prompt for a techno kick? As fun as searching through a poor quality sample library.

 

nice idea but won’t work, would be difficult because AI makes music in any genre its not not limited but all you listed still has the boundaries of the genre, well maybe except muzak being a more broad definition.
I wouldn’t call it music at all. Some new word is needed - Music is created by humans, a mix of their creativity, emotions, experiences, soul and culture, everything a machine simply cannot have. And AI cant create new genres as humans just mimick those that are already discovered. Real art is the path of disappointments and successes, the path of blood, sweat, and tears, happy accidents, stories behind every song. That’s what shall never be behind AI

 

Last edited: Dec 14, 2025 at 1:22 AM