8 Min Read

Every synth plugin's marketing copy leads with the same number: how many presets are inside. Five hundred patches. A thousand. "Endless sonic possibilities." It's an easy number to put on a product page, and it maps onto something real — more starting points, more chances one of them fits the track you're working on. But it's also a number that measures the wrong thing, and most producers who've spent real time with a large preset bank eventually notice it: a library of five hundred patches that all modulate the same shallow way is a smaller instrument than a library of fifty that don't.

This isn't a contrarian take about quality over quantity in the abstract. It's a specific, audible property of synthesis that preset count doesn't capture at all: how much a sound is actually doing to itself over time, independent of what you do to it. Call it modulation depth. It's arguably the single biggest predictor of whether a patch still sounds interesting on the fortieth listen, and it has almost nothing to do with how many patches shipped in the same folder.

A static patch isn't necessarily a boring waveform — plenty of static patches use rich, complex oscillators, layered unison, elaborate filter curves. Static means the relationship between those elements doesn't change once the note is held. Play the same note for four seconds and a static patch delivers, in effect, the same instant repeated four hundred times. The ear is remarkably good at detecting this, even when it can't articulate why a sound feels flat. Sustained tones that don't evolve read as synthetic in the pejorative sense — technically correct, perceptually inert.

This is why a preset bank can have enormous surface variety — different waveforms, different filter types, different effects chains — and still feel exhausted after twenty minutes. Variety between patches isn't the same as movement within a patch. A library can offer five hundred different flavors of static and never once offer genuine depth.

It helps to break "modulation depth" into three separate, addable properties, because a patch can be strong on one and weak on the others, and knowing which is missing tells you exactly what to fix.

Rate diversity. A single LFO modulating a single parameter at a single speed is modulation, technically, but it reads quickly as a predictable wobble rather than organic movement — the ear locks onto the period within a couple of cycles and then effectively tunes it out as noise. Real depth usually comes from multiple modulation sources running at different, ideally non-aligned rates simultaneously: a slow filter sweep, a faster subtle pitch drift, an even faster amplitude flutter, none of them locked to the same clock. The complexity isn't in any single modulation lane — it's in the interference pattern between lanes that never quite repeats.

Modulation amount. This is the more obvious axis and the one most patches actually get right: how far a parameter travels, not just whether it moves. A filter cutoff that drifts two semitones' worth of brightness is doing something meaningfully different than one drifting a tenth of that, even at an identical rate. Undershooting here is the most common reason a patch that technically has modulation still reads as static — the movement exists but is too subtle to register as intentional rather than incidental.

Cross-parameter coupling. The deepest layer, and the hardest to program well: modulation sources that affect more than one parameter at once, ideally with slightly different scaling on each, so the relationships between timbre, pitch, and amplitude keep shifting relative to each other rather than moving in lockstep. A patch where filter and amplitude modulate together in a fixed ratio still eventually reads as one movement. A patch where they're each responding to a different combination of sources, at different depths, produces something closer to the internal complexity of an acoustic instrument, where breath, embouchure, and bow pressure are never perfectly correlated either.

"A shallow patch runs out of new information in the first second and a half. A deep one is still revealing new detail at second seven."

There's a persistent instinct in synth design and synth marketing to add depth by adding sound sources — a third oscillator, a sixth unison voice, an extra layer stacked on top. More sources do add richness, but they add it in a way that's audible immediately and then plateaus fast; ears adapt to a thicker static chord about as quickly as they adapt to a thin one. Modulation sources don't plateau the same way, because their effect isn't a fixed addition to the timbre — it's a continuously changing one. An instrument with three oscillators and one shallow LFO will sound rich for the first second and static for the following three. An instrument with one oscillator and four well-tuned, non-aligned modulation sources will often sound more alive across an entire held note, because there's simply more happening that the ear hasn't already fully parsed.

This is part of why granular and continuous-modulation engines in particular reward deep patch programming disproportionately: they're architected around movement as the core mechanism rather than an add-on, so the ceiling on how interesting a patch can sound is set by how thoughtfully the modulation is layered, not by how many additional sound sources got stacked in.

None of this means preset count is meaningless — more starting points genuinely does mean more chances of finding something close to what a track needs, especially under deadline. But it explains why the relationship between preset count and perceived value flattens out fast, and sometimes even inverts. A five-hundred-patch library assembled quickly, optimized for surface variety across broad categories, will often contain far fewer patches with genuine multi-axis modulation than a fifty-patch library where every single one was programmed by someone specifically chasing that quality. The larger number looks better in a product listing. The smaller number is very often the more useful instrument to actually own.

This is a real, testable claim, not just an aesthetic preference: play any two patches from a library, hold the same note for eight seconds on each, and count how many independent things change in the timbre before the note ends. A shallow patch runs out of new information in the first second and a half. A deep one is still revealing new detail at second seven. That gap is the entire argument, and it's audible in about fifteen seconds of listening, no technical vocabulary required.

The practical version of all this is a listening test, not a spec sheet. Before committing to a preset bank, hold a single note from three or four patches for at least six to eight seconds each and listen specifically for whether the sound is still changing at the end of that window, and whether the changes at second seven feel related to but distinct from the changes at second two. A patch that's identical at second one and second seven, regardless of how rich it sounded in the first half-second, is a static patch wearing a complex-sounding costume. A patch that's audibly, specifically different at the end than the beginning — without having done anything as obvious as a filter sweep from closed to open — is doing real modulation work, and that quality tends to be consistent across an entire bank once a sound designer is actually chasing it rather than defaulting to a stock LFO on cutoff.

For producers programming their own patches rather than buying banks, the same three axes translate directly into a checklist. Before finalizing a pad or texture, ask whether more than one modulation source is active, whether their rates are deliberately non-aligned rather than left at convenient round numbers, whether the modulation amount is large enough to register as intentional rather than incidental, and whether at least one modulation source is touching more than a single parameter. A patch that fails all four of those questions can still sound fine in isolation — plenty of static patches sound fine in isolation, which is exactly the trap. The test that actually matters is whether it still sounds fine, or better, at second seven of a held note, sitting in a full mix, with nothing else drawing attention to it. Preset count will never answer that question. Only the patch itself can.

More on synthesis, sound design, and studio technique in the Frequencies archive.