Type in a few words, select a music style, and shortly afterward, a complete song plays—complete with drums, bass, melody, and vocals. What would have sounded like science fiction just a few years ago is now a reality. Artificial intelligence is changing the way music can be created. But is that really “making music”? And what does this development mean for people who compose and produce music themselves?
From the Recording Studio to the Algorithm
I still remember well the time when electronic music production mainly meant grappling intensively with the technology.
Synthesizers had to be programmed, drum machines tuned rhythmically, and individual tracks carefully arranged. If you wanted a specific sound, you had to figure out how to create it.
A good arrangement didn’t just happen by pressing a button. You experimented with melodies, altered chord progressions, shifted individual notes, and sometimes spent hours trying to give a track exactly the atmosphere you had in mind.
Of course, even back then, not everything was done entirely by hand. Sequencers, samplers, presets, and—later—increasingly powerful music software simplified production considerably.
So technology has always been a part of electronic music.
The key difference compared to today lies in how much a machine can generate on its own.
With generative AI systems, it’s now possible to create a nearly complete music production in a short amount of time based on a textual description.
No instrument needs to be played for this. Nor is knowledge of musical notation strictly necessary for the initial generation.
That’s technically impressive.
And at the same time, it raises questions that go far beyond sound quality.
A good song is more than just good sound
What actually makes a song good?
Is it a catchy melody? A powerful rhythm? A voice that conveys emotion? Or the interplay of all these elements?
In electronic music, sound is undoubtedly an essential component. A precisely tuned bassline, the right synthesizer pad, or an effective transition can turn an average idea into something special.
But technical perfection alone is far from a guarantee of interesting music.
A song can be superbly produced and still fail to resonate.
Conversely, there are recordings whose technical quality is hardly impressive by today’s standards, yet they still move us decades later.
Perhaps because they’re linked to a personal experience. Perhaps because a certain melody evokes memories. Or because someone found the right words at exactly the right moment.
This is where, for me, the difference lies between simply creating music and consciously shaping it.
An algorithm can generate compelling musical structures. Whether that results in a work with personal significance is another question.
What Artificial Intelligence Can Actually Do
It would be wrong to downplay the capabilities of today’s generative music systems.
The technology is already capable of producing astonishingly complex musical results.
It can generate instrumentations, simulate vocals, develop arrangements, and combine a wide variety of musical styles.
The technical execution can also be impressive. An AI-generated track no longer automatically has to sound like a primitive computer composition.
This opens up interesting possibilities, especially for people who want to develop musical ideas.
A rough concept can be brought to life immediately. Different tempos, keys, or instrumentations can be experimented with without having to build a complete production for each variation.
This can be enormously helpful for creative work.
However, such results do not arise because AI understands music in the same way a human does.
Generative models learn statistical relationships from extensive training data. From this, they can derive musical patterns, structures, and sonic characteristics and generate new audio content.
This does not mean that they have experienced a concert, associated personal memories with a song, or developed a musical message based on their own experiences.
The distinction is important, even if the audible result is convincing.
When AI Becomes a Tool in the Studio
For me, artificial intelligence is particularly interesting when it doesn’t replace the entire creative process, but rather enhances it.
A musical idea often begins with a vision that’s hard to pin down at first.
Maybe I have a certain atmosphere in mind—an electronic track with the character of classic dance productions but with modern dynamics. Or a combination of instruments that don’t necessarily seem to go together at first glance.
AI can help bring such ideas to life.
It can provide variations, highlight new approaches, and sometimes make musical possibilities audible that you wouldn’t have thought of on your own at first.
But for me, the crucial work doesn’t necessarily begin with the first generated version.
It begins with the selection process.
Which melody really works? Is the chorus strong enough? Do the lyrics make sense? Does the voice fit the musical expression? Is the arrangement structured in a way that makes sense?
And above all: Does the song have a recognizable identity?
Simply adopting a generated result without scrutiny is different from developing it into a consciously crafted production through your own musical decisions.
The quality of a tool does not automatically replace the quality of the decisions made with it.
The Difference Between Generating and Producing
In the discussion about AI music, terms are often confused.
However, generating, composing, arranging, producing, and mastering describe different activities.
When a system generates a finished song from a text prompt, it delivers a musical result.
In contrast, when a person develops a melody themselves, determines chord progressions, designs musical structures, and then uses AI for the sonic realization, this represents a different form of creative involvement.
There are numerous hybrid forms between these two examples.
Some use AI solely for brainstorming. Others use it to create individual instrumental tracks. Still others have complete productions generated and then edit them further.
It would therefore make little sense to categorize all AI-assisted music productions as equivalent or identical across the board.
What matters is what actually originates from humans and which tasks the system has taken over.
This applies to both creative evaluation and legal classification.
Can an artificial voice convey emotions?
I find the development of artificially generated vocal tracks particularly exciting.
Today, systems can generate voices that are remarkably convincing in terms of phrasing, timbre, and expression.
A voice can sound fragile, powerful, melancholic, or euphoric.
The audible result can evoke genuine emotions in the audience.
But does the artificial voice itself feel anything?
This question isn’t necessarily decisive for the musical impact. People can also be emotionally moved by fictional stories, animated characters, or instrumental music.
What often matters is the meaning we ourselves ascribe to what we hear.
Nevertheless, one should distinguish between an artificially generated interpretation and a human vocal performance.
A singer brings physical experience, individual technique, personal decisions, and often their own story to the recording.
This dimension cannot be assessed solely on the basis of a perfectly sounding audio file.
I therefore do not believe that artificial voices will fundamentally render human singers obsolete.
However, they undoubtedly change the ways in which singing can be incorporated into music productions.
The Big Question of Copyright
At the latest when AI-generated music is released, a creative discussion turns into a legal one.
Who owns a song if its melody, lyrics, and instrumentation were generated entirely by a generative system?
And what about cases where a person has created only individual components themselves?
In Switzerland, this distinction is particularly important.
According to SUISA’s guidelines, musical works generated entirely by AI without any human creative input are not protected by copyright and cannot be registered with SUISA as independent, protected works.
The situation may be different if a human has created independent, copyrightable musical or lyrical contributions and uses artificial intelligence merely as a tool.
A prompt alone is not automatically equivalent to a copyright-protected composition.
Furthermore, the rights to a composition must be distinguished from the rights to a specific sound recording.
The terms of use of the AI services employed, distribution agreements, and potential third-party rights also play a role.
A commercial license granted by an AI provider does not automatically mean that a fully generated song would be protected by copyright as a personal human composition.
For anyone who professionally releases AI-assisted music, transparent documentation of the creative process is therefore particularly important.
Which parts were composed by the creator? Which lyrics were written independently? Which elements came from the AI? And what edits were made afterward?
Such questions may seem bureaucratic at first. In the long run, however, they can be decisive.
What will become of the musicians?
If a machine can generate music in a matter of moments—music that used to require several people and many hours of studio work to produce—this will inevitably transform the music industry.
Economic pressure could increase, particularly in the realm of standardized background music or simple commissioned productions.
Anyone who simply needs a functional soundscape might wonder why they should still commission a traditional production for it.
This is not an unrealistic concern.
At the same time, new tools also open up new creative possibilities.
People without access to expensive studios can experiment with musical ideas. Small projects can develop more elaborate arrangements. And experienced producers gain additional tools for their work.
For me, therefore, the crucial question is not whether AI is fundamentally good or bad for musicians.
What’s far more important is how it’s used, under what economic conditions it operates, and whether the contributions of human music creators continue to be respected and fairly compensated.
Technological progress is not automatically fair.
But it’s also not necessarily the end of human creativity.
Should AI-generated music be recognizable?
Another point of discussion concerns transparency.
Should listeners know whether a song was created entirely or partially by artificial intelligence?
I believe transparency is generally a good thing.
Not because AI-assisted music is automatically inferior. But because the story behind its creation can be part of a musical work.
People who listen to a particular artist may be interested not only in the sound but also in the people behind the music.
What decisions were made? Who composed it? Who sang it? Who produced it?
Such information builds trust.
However, transparency should not be confused with a blanket dismissal.
A synthesizer doesn’t make a musician any less creative than an acoustic instrument. A digital audio workstation doesn’t automatically make a production impersonal.
And even the use of artificial intelligence, in and of itself, doesn’t tell the whole story about a person’s creative contribution.
What matters is how the tool was used.
The future doesn’t belong solely to algorithms
I am convinced that artificial intelligence will remain a permanent part of music production.
The technology will continue to evolve. The quality of generated music is likely to improve, and the line between traditional production and AI-assisted workflows will become increasingly blurred in some areas.
But I don’t believe this will automatically lead to a future without human musicians.
Music is more than just the technical creation of an audio file.
It accompanies personal memories, social changes, and individual life stories.
A concert isn’t meaningful simply because every note was played perfectly. A song isn’t automatically unforgettable just because its frequency spectrum is technically flawless.
And a melody doesn’t necessarily need elaborate production to move people.
A machine’s ability to produce musical results changes the tools we use.
But it does not definitively answer the question of what music means to us.
So, can artificial intelligence make music?
My answer is: Yes, artificial intelligence can create music.
It can produce entire songs, provide musical ideas, and create results that inspire people.
We should neither ignore nor downplay this.
But for me, the more interesting question isn’t whether a machine can create a compelling song.
Rather, it is: What do we humans do with these new possibilities?
Do we use them to produce more and more interchangeable content as quickly as possible? Or do we use them to develop our own ideas, explore musical boundaries, and give voice to the things that truly matter to us?
For me, the answer remains a matter of conscious creative choice.
AI can generate music. Whether something original, personal, and meaningful emerges from it doesn’t depend solely on the technology.
Perhaps that is the most important insight from this development.
We shouldn’t deny machines what they’re already capable of.
But we shouldn’t forget what fascinates us about making music ourselves.
Because in the end, I’m not just interested in how a song came to be.
I’m interested in whether it has a story to tell.