How will AI music services change workflows, DAWs, especially Cubase?

Steinberg Lounge
1

Hi all! What do you think about the music industry in the near future from the side of AI, creativity and technical tools? My thoughts about it:

Generative music powered by artificial intelligence, represented by services like Suno, Udio, Soundraw, AIVA, Boomy is no longer just a toy. They create complete arrangements with vocals in mere seconds. However, the real tectonic shift in the industry won’t happen when AI simply learns to generate audio, but when it allows for the flawless conversion of that audio back into MIDI data. And that moment is very close.

As soon as reliable tools emerge for extracting clean MIDI stems (individual tracks) from AI compositions, the musician’s workflow will change forever.

The New Workflow: “Musical Re-amping”

Imagine this: A composer generates a dozen arrangement variations in a specific style using an AI service. Instead of taking the raw and sonically imperfect audio file, they receive a full package of MIDI data: drum parts, basslines, harmonies, and even the vocal melody.

Even now, we already have tools:

  • Steinberg SpectraLayers 12 for drum separation from 1 audio: kick, snare, hh, etc.
  • NeuralNote VST for manual extraction MIDI from raw audio
  • Drum Replacer vst’s for replacing drums
  • AI services for audio spectral restoration and enhancement

From here, a process begins that can be compared to “re-amping” in the guitar world. This MIDI material is imported into Cubase or another familiar DAW. In this scenario, the AI acts not as the final producer, but as an incredibly fast co-writer or session musician who has sketched out a “demo.”

Music creation becomes unimaginably accessible. The barrier to entry—once defined by virtuosic skill on an instrument or deep knowledge of music theory—is lowered. What comes to the forefront is taste, production vision, and the skill of “finishing” a track.

The Challenge for DAWs: Synergy or Death

This is where an existential threat arises for classic DAWs. Giants like Steinberg Cubase, Ableton Live, or Pro Tools were built as “blank canvases”—professional environments for creating music from scratch.

But if the primary task shifts from “creation” to “editing and refinement” of AI-generated material, the old approach could lose. New, “AI-native” DAWs will appear, built from the ground up around synergy with generative models.

To survive and stay relevant, traditional DAWs will have to urgently change their development vector.

In future updates, we must see:

  • Deep Integration: Not just “Import MIDI,” but seamless connectivity with AI services directly from the interface.

  • “Cleanup” Tools: Functions to automatically correct “dirty” MIDI from AI.

  • AI Assistants: Built-in AI arrangement aids that suggest ways to develop an existing MIDI idea.

In the coming years, the winner won’t be the DAW with the most plugins and usability, but the one that best combines human talent with the computational power of artificial intelligence. For Cubase and its competitors, the race has already begun.

2

I would consider moving this to The Lounge. That seems like a far better category for this thread in my opinion. Just a thought.

1 Like
3

Agreed, done.

1 Like
4

iow a machine generates options, not a composer.

I think this is just describing a race to the bottom and an erosion of the term “creation”. I machines take care of 90% of the creation of a music track what did you really “create”, and is that something to cheer on broadly speaking?

Of course it doesn’t mean other people can’t gain skills on instruments, but it’s a bit depressing reading posts like these after having studied for years and years. Music skills are already devalued enough as it is and further machine productions will just bring that value even lower.

There are probably outliers, like people with disabilities that can now create, but outside of that I really think essentially none of this is good for society as a whole. Just bad.

That could be true.

An alternate take on that though is that a minority of people still want to create the “old fashioned way” and play live and produce using their brains and that this niche will be serviced by software like the current DAWs, and then the people who just want to click the button that says “bossa nova rhythm” can do that online or use some of the other tools. So maybe there is still a place for a less AI-heavy DAW. Maybe.

Either way the bigger threat is to society as a whole. By the time Steinberg’s Cubase and Nuendo are facing greater threats from competing “machines” our society as a whole will be in a world of trouble anyway, because of AI.

3 Likes
5

The whole discussion about whether so-called “AI” will obviate the need for human creativity is based on the incorrect assumption that humans create e.g. music as a commodity, as if it were something to be traded, or as a store of value. This incorrect assumption is understandable, given the prevalence of what is known as a “music industry”, which should be more correctly a recording industry.

We create music because we must; music is expression.

So, will so-called “AI” music services change workflows? I would say, if you’re in e.g. the library music business, yes, plan to pivot, but I also believe there will still be customers in years to come who, for whatever reason, won’t have yet realized there are cheaper ways to get the same result, just as I still see small businesses still pay large amounts of money to companies to provide them with a static web page and sell it as being “online”.

What features would be useful in Cubase? I’d first think of backing tracks, constructed based on a prompt such as “give me 16 bars of a jazz-influenced bass line that would not be too complex underneath a basic blues theme, and generate that as a new MIDI track using the HALion patch xyz”. Whether I choose to use it or not is then up to me, but I do imagine (not being a native bassist) it would give me the initiative to learn the bits I like, and eventually play it in live myself.

As far as integration of so-called “AI” into Cubase, well, it’s already there to a degree in SpectraLayers, as it enables stem separation. I think similar machine-learning techniques could be applied to drumming in Groove Agent, where it could learn (i.e. actively “listen”) to what’s being played on other tracks, rather than just reacting to “intensity” as it currently does.

To summarise, if so-called “AI” can currently mimic what you do, then, well, it’s over. If, on the other hand, you enjoy the process more than the end result (assuming it doesn’t matter to you financially) then you are at the dawn of a golden age!

6

I agree with most of your statements, especially in the first paragraph.

However, I don’t think there will be a golden age for anyone if the fundamental problem of AI and copyrights is not adressed. We’d be more or less stuck in a circle of self-reference and suffocated by a tsunami of perpetuations and variations based on the same training material that AI is based on.
AI and its possibilities might sound like fun for a particular group of hobbyists (e.g. prompt producers) and it sure will be for a limited time. Nonetheless, there will be barely anymore professional musicians left at one point if it’s impossible to spend the hours and ressources needed to become one. This will result in a massive decline of skills over the years. Skills that have been necessary to produce the material that AI is based on in the first place.

The devaluation of culture in general and music in particular is a very dangerous path.
I know, you can’t turn back time and of course, AI is not all bad. But if politics fail to come up with new copyright terms then
there’s a lot more at stake here than most people might think.

2 Likes
7

Copyright legislation will not protect from so-called AI any more than tin foil hats protect from mind control.

Probably, but that does not mean that there will be no musicians – musicians will just earn even less than they currently do, which is – let’s face it – in western society, a pittance anyway.

Not sure why it would be impossible, but when I learned to play, I had little or no resources (“money”) and spent the hours because I was compelled to … I could not envisage not playing music.

Full disclosure though: I don’t think I ever generated any real wealth from it.

8

I didn’t want to reply but here it is:

Basically, the same argument was brought forward with regard to sampling back then. Copyright infringements will always happen to a certain degree, of course. Selling out culture is not an option with the way AI is taking over. Tin foil hat, really..?

Becoming a professional musician takes a lot of time and money, no matter what.
If professional musicians cease to exist then music culture will suffer a great deal. Learning the art of tracking, mixing, mastering, playing, composing doesn’t come overnight. If musicians can’t pursuit their vocation because they can’t make ends meet then a lot of talent might never reach their full potential. You see where I am going with this, right?

Nothing wrong with autodidacts! That’s not the point and I respect your opinion as an enthusiast. Nonetheless, we wouldn’t have this rich musical heritage without professional musicians. Giving up on them just like that - I am not sure if you have really thought this through.