Are we reaching the limits of “conventional” composition? 

At a time when most AI-driven tools are designed to replicate human logic — optimizing for coherence, familiarity, and predictability — I find myself drawn toward the opposite direction: the unpredictable, the unstable, the emergent.

With the release of ECLO.RE/GEN, I want to open a broader question:

Why do we remain so attached to traditional song structures, when they represent only a tiny fraction of what sonic storytelling could be?

ECLO.RE/GEN is not just a plugin. It’s a procedural stems smasher — an environment where composition is no longer about arranging fixed elements in time, but about destabilizing structure itself.

Instead of refining timelines, you fracture them. Instead of sequencing notes, you navigate states. Instead of control, you engage with emergence.

Built in Reaktor 6, the instrument sits somewhere between a mini-DAW and a performance engine — but its intent is neither efficiency nor automation. It is about expanding aesthetic possibility.

One of the most striking things I experienced while working with it came from an almost paradoxical situation. I fed ECLO.RE/GEN with a track that, on the surface, was extremely linear — highly progressive, with no abrupt dynamic shifts, no dramatic structural turns. The kind of piece that unfolds smoothly, almost transparently.

Precisely because of that, it became one of the most fertile materials for the system.

By deconstructing and recomposing it, ECLO.RE/GEN began to reveal micro-sequences — tiny, previously unnoticed fragments embedded within the flow. Details that would normally dissolve into continuity suddenly surfaced, gained autonomy, and started to suggest their own internal logic.

It wasn’t just a transformation of the track.It was a transformation of listening itself.

What once felt linear became granular, almost topographical. Hidden articulations emerged. Accidental rhythms appeared. And from these fragments, a completely new sense of narrative began to take shape — not imposed, but discovered.

At its core, ECLO.RE/GEN treats stems not as finished objects, but as raw material for transformation:

Here, mechanical failure becomes groove. Clock drift becomes phrasing. The result is closer to audio impressionism than generative automation.

There’s a paradox in today’s landscape:

This is not about improving composition. It’s about questioning its foundations.

The project also exists as an interactive album:

I’m fully aware this speaks to a niche — a small circle of creators interested in radical reconfiguration rather than refinement.

But maybe that’s exactly where new languages emerge.

It’s important to clarify that I’m not opposed to AI at all. ECLO.RE/GEN itself is, in a way, a kind of machine — and there’s no reason why AI couldn’t eventually evolve into systems capable of generating genuine unpredictability, provided we choose to train them for that purpose.

What I’m questioning instead is the current tendency: most AI models today are built primarily around imitation, designed to reproduce, to emulate, to replicate existing patterns.

But imitation is already one of humanity’s core abilities — we observe, absorb, reinterpret, and reproduce. We are masters of mimicry by nature.

Where things get interesting, though, is in the boundary between imitation and invention. Using AI as a tool to gain efficiency, to streamline the production of an idea inspired by an already existing artistic piece, can indeed be very pleasant — recreational, stimulating, even downright fun. It can feel cool, empowering, and creatively liberating.

Yet to make that process a long-term ambition or creative philosophy — unless one works in advertising or purely utilitarian content creation — raises a real question about the kind of musical ecosystem we’re building for the future. Are we heading towards an environment that endlessly refines the same shapes, colors, and emotional grammar?

Having influences is perfectly fine; it’s even essential. That’s how we create our own recipes, our mixtures, our sonic identity. But if every new creation comes down to recombining preexisting patterns under the guidance of algorithms, won’t we face, sooner or later, a shortage of evolution — a decline in artistic skill and vision?

If AI were only used to duplicate sonic characteristics — timbres, textures, frequencies — that might still be acceptable. But when it begins to shape narrative structures, rhythmic signatures, or compositional frameworks, we risk losing the subtle chaos, the deliberate contradictions, and the spontaneous decisions that define human art.

Because there are things that only the human hand, or rather the human impulse, can bring forth: those unintentional details, those “happy accidents” that emerge from intuition, from mistake, from hesitation — all those cracks in logic where something unexpectedly alive slips through.

The actual way AI algorithms are designed, by their very nature, rarely allows for such accidents. They produce within parameters; they normalize — not out of malice, but because they’re built that way. They synthesize to predict, not to surprise.

Perhaps one day soon we’ll see systems specifically trained to generate chaotic and unexplored audio structures, to remix material in unpredictable ways. That could open fascinating possibilities.

Still, the element of free will — the decision, the moderation, the choice to cut, to silence, to distort — will remain, at least for a while, uniquely human. Provided, of course, that humans stop behaving exclusively like the very machines they have built to imitate.

So I’d like to leave it here.

ECLO.RE/GEN is simply an open field — a space for those who feel curious about stepping outside familiar structures, and exploring what might emerge beyond them.

No fixed method, no expected outcome. Just a tool to listen differently, to reshape, to wander.

If it resonates with you, take it somewhere unexpected.