Fear and Loathing in Stereo

At one point the bench had a rack of hardware oscillators warm enough to heat the room, something close to two kilograms of patch cable knotted into a shape between a switchboard and a crime scene, and a mains hum at the bottom of everything whether we asked for it or not. Real analog voices. Big, fat, glorious — and mono, every one of them, right down to the metal. So to make any of it sound wide, we did what everyone does: ran the whole beautiful mess into a pseudo-stereo box at the very end and let a widener fake the rest.

Then somebody closed the lid on the hardware and opened the laptop — the favorite software synth, the expensive one everyone swears by — and there it was again. Same gorgeous tone, same trick at the tail: one voice, cloned into two, smeared apart by an effect that had no idea what it was widening. All that gear, all that cable, all that money, and everybody was solving stereo in the same place. Last. Long after the sound itself was done.

That was the moment we decided to take the longer road.

The usual trick

The standard recipe is the one we kept falling back into: generate the voice once, in mono, then split it to two channels and make the copies differ after the fact. A chorus detunes a delayed copy against the original. A doubler nudges one side by a few milliseconds or a few cents. A widener plays phase games — filters and mid/side gain arranged so the left and right ears receive slightly different signals. Different tools, same idea: the width is applied to the sound, in an effect stage, after the sound is already finished.

It works, and there is nothing wrong with it. But the differences between the channels are whatever the effect imposes: a fixed delay, a slow wobble, a static phase curve. Sweep the filter, slam the pitch, hold a note for eight bars — the widening layer keeps doing the same thing, because it knows nothing about the sound it is widening.

What we do instead

The longer road looks like this: Phaseburn One generates left and right independently, at the oscillators. There is no mono voice that gets widened afterwards — there are two channels from the first sample. The voice is monophonic by design — one voice, two channels — and every sound source contributes its own kind of width.

The Phase-Locked Loop (PLL) — the control circuit from radio engineering that we turned into our main voice: an oscillator listens to a reference, measures how far its phase has drifted, and constantly corrects itself — runs per side. Each channel’s loop chases, overshoots and settles on its own. Keep the loop tight and the sides agree; loosen it and they diverge, correct, and converge again. That motion between the channels is the width.

The VECTOR oscillator — vector phase shaping — bends the phase of a waveform around a movable inflection point: two coordinates, X and Y, sweep one cycle from smooth and flute-like to bright and biting without ever swapping waveforms. Its X∆ and Y∆ offsets run the two channels at the same pitch but diverging shapes. The pitch stays anchored in the middle; the timbre splits between the ears. Width from the waveform itself, not from a chorus.

The MORPH oscillator carries a STEREO ∆ control across its twelve waveform families, and its detuned unison saw stack runs seven voices per channel — not one stack panned apart, but two full stacks, one per side.

Even the filter joins in: a stereo separation control (STEREO SEP on the panel) offsets the two channels, so a filter sweep is a slightly different sweep on each side.

Why it sounds different

Widened mono and source stereo can measure similarly on a correlation meter, but they behave differently over time. In widened mono, the inter-channel difference is a constant layered on top of the sound. In source stereo, the difference is produced by the same mechanisms that produce the sound — so it evolves with the sound. Hit the loop with a pitch jump and each side rings its own overshoot. Sweep X and Y and the two channels pull apart and drift back together. Nothing about the image is static, because the image is not a layer. It is the patch.

And in mono?

Fair question, because fold-down is the classic weakness of wideners: the phase differences that made the width can cancel when left and right are summed to one channel — the club system, the phone speaker. Two independently generated channels sum more like two takes of the same instrument than like a signal and its phase-shifted copy. And we give you explicit control either way: the master section has a mono-bass anchor that sums the low end to the center below a set point, and a stereo width control to rein the whole image in — or collapse it entirely, so you can hear exactly what a mono system gets.

Chorus as a choice, not a crutch

There is a chorus in Phaseburn One — four modes, up to eight voices with independent stereo offsets, cross-feedback and filters inside the feedback path, from subtle width to seasick pitch soup. When you want that specific flavor, nothing else does it, and we like it too. The point is not that chorus is bad. The point is that it should be a color you reach for, not a structural repair for a mono engine. Here it sits on top of a voice that is already stereo, which is a different starting position.

Width is a tool, not a virtue

We want to say this plainly: wider is not automatically better. A bass that wobbles across the stereo field falls apart on a big system. That is why the sub oscillator has its own escape route — SUB DIRECT routing lets it bypass the master tone stage entirely, so the low end can stay dead-center and untouched while everything above it moves. The loops diverge, the shapes split, the unison shimmers, and the foundation does not move an inch. Stereo where it helps, mono where it matters.

Hear it on your own monitors

Width is the one thing a spec sheet cannot show you. Get Phaseburn One — free through the beta, a demo after October 15, 2026 until you register — and put the stereo image on your own speakers.