The Filter

Phaseburn One’s four sound sources — the Phase-Locked Loop (PLL) voice, the VECTOR phase-shaping oscillator, MORPH and the sub — are cabled into FILTER 1, and a second, identical filter, FILTER 2, waits behind it. Because the instrument runs a single voice there are no per-voice filter copies to manage: two filters with the whole CPU budget behind them, sitting after the oscillators and ahead of every effect. By default all four sources go through FILTER 1, sub included, and FILTER 1 runs on into FILTER 2; the patchbay lets you re-cable which sources reach each one and in what order. Both filters live on the FX tab; the filter envelope they share is edited in the envelope panel at the bottom of the Oscillators tab. The coloration stage — the analog-mess controls that keep the voice from sitting perfectly still — belongs to the PLL and is covered in the oscillators chapter (Chapter 2 — Oscillators).

The filter

Pick the machine before you sculpt. TOPOLOGY chooses between two complete filter designs that share one set of controls. The transistor-ladder is four saturating filter stages in series — the thick, vintage-flavored lowpass lineage — built after D’Angelo and Välimäki’s published research on nonlinear ladder filtering. It colors everything it touches: push level into it and the stages bend, sweeps sound expensive, resonance turns vocal. The state-variable filter (SVF) is the surgical alternative, built on Zavalishin’s zero-delay-feedback (ZDF) designs: its resonance peak stays exactly where you put it and the response is clean until you deliberately dirty it. When a patch needs weight and grime, start with the ladder; when it needs a precise, repeatable sweep, start with the SVF.

The response shape is one knob: the pole morph. It blends the filter smoothly from one landmark shape to the next, and the in-between positions are real filter responses, not crossfades between two signals; the knob’s tick marks name the landmarks. On the ladder the journey runs from a steep lowpass through a gentler lowpass and a bandpass, past a notch, to a steep highpass. On the SVF it is longer — lowpass, bandpass, highpass, notch, peak and allpass. Steep slopes cut decisively — bass patches keep their focus; the gentler lowpass leaves more of the upper harmonics breathing. Park the knob on a tick when you want a textbook response; sweep it slowly from an LFO and you get response morphing that no fixed shape selector can play.

The filter is bypassed until you engage it, and it opens fully wide when you do — so switching it on never dulls a patch by surprise. By default all four sources feed it together, each through its own cable into FILTER 1 — sub included, so engaging the filter shapes the low end along with everything else. Each of those cables has its own level, so the balance into the filter is set right there. The patchbay on the Mixer tab re-cables this. Pull a source’s cable and it leaves the filtered path — and the output — until you patch it somewhere else: into FILTER 2, straight into an effect, or around the filters with a direct-to-master cable that arrives unfiltered. To keep a sub solid under a closing filter, route the raw sub around the whole chain with the sub’s FX BYPASS switch (see the oscillators chapter (Chapter 2 — Oscillators)).

MIX, the last knob in the filter row, sets a dry/wet blend between the signal that entered the filter and the filtered output — fully wet by default. Back it off to fold parallel dry signal back in: a way to keep the low end and transient attack intact while the ladder does its coloring above, or to dial the whole effect in gently. It is the filter’s own internal blend, and there is no output-gain knob to manage beside it: the engaged filter is automatically loudness-matched to bypass, so bringing it in colors the sound without a level jump. The blend is click-free and only acts while the filter is engaged; with the filter off, its input passes through untouched.

STEREO SEP splits the left and right cutoff frequencies apart — left shifts down, right up, by up to an octave each — which widens a mono patch without touching a single effect. The first half of the dial covers the subtle range; past it the two sides start to read as two different filters. Each side’s resonance is level-matched at its own cutoff, so opening SEP widens the image without leaning it to one side, however high the resonance.

Two filters

FILTER 2 is a complete second filter, identical to the first: both machines, the same pole morph, drive, TYPE, CHARACTER, bass lock, resonance, key tracking, STEREO SEP and MIX, each with its own settings. It ships switched off, and a filter that is off passes its signal through untouched, so a new patch — which already runs FILTER 1 into FILTER 2 — sounds exactly like one filter until you switch the second one on. Presets and projects from before FILTER 2 existed keep their own routing, with FILTER 2 arriving unconnected.

There is still one filter panel on the FX tab. The FILTER 1 | FILTER 2 tabs to the right of its ON switch choose which filter the panel shows: the knobs, the ON switch and the LADDER/SVF toggle all belong to the selected filter, and the editor remembers which one you were looking at.

Series is the obvious move — a ladder lowpass into an SVF notch, or a driven ladder into a clean one that tames what the drive added. The patchbay also takes the pair in parallel: give each filter its own sources, or the same sources at different levels, and send both onward — a bandpass on the PLL beside a lowpass on the sub. Either filter can sit anywhere in the chain, after an effect as easily as before one; the patchbay chapter (Chapter 5 — Mixer & Patchbay) covers the wiring. Either filter’s output can also feed one of the three feedback returns, so a filter can sit inside a delay’s, chorus’s or reverb’s loop.

When what you want is not character but a clean cut — trimming the low end off a reverb return, or narrowing a source to a band — the patchbay carries two more filters for exactly that: the HP/LP slots, with six slopes and one frequency knob each and no envelope, drive or colour of any kind. They live only in the bay; see HP/LP (Chapter 5 — Mixer & Patchbay).

Resonance, drive and self-oscillation

Turn RES up until the filter starts to sing. Resonance emphasizes the region around the cutoff, and on the ladder topology the feedback is deliberately allowed past the point where the filter becomes its own oscillator: pushed to the top of its range it self-oscillates at any cutoff frequency, with the per-stage saturators bounding the level instead of a damping fudge that would thin the sound. The SVF crosses into self-oscillation only in the last stretch of its resonance range, and does it with a cleaner, steadier whistle. A self-oscillating filter plus full key tracking (next section) is a playable voice in its own right — tune it, sequence it, treat it as a fourth oscillator.

Resonance costs bass. As the peak grows, the region below it thins — the classic complaint about ladder designs. BASS LOCK is the fix: it tracks the low-frequency energy entering the filter and restores what resonance steals. It stays out of the way at low resonance and engages progressively as resonance rises, so it never bloats a non-resonant patch. BL FREQ sets the corner of that low-frequency tracker — low values guard the sub range, higher values reach into the low mids.

DRIVE pushes the input into the filter’s saturating stages, from clean through thick to overdriven — one knob now spanning the full 0 to +24 dB, with no separate boost step. The TYPE knob beside it selects the nonlinearity every stage uses — transistor, diode or tube, three saturation curves with progressively different edges. Because TYPE feeds both the drive input and the filter’s own per-stage saturation, it colours the sound whenever the filter is working, not only when DRIVE is up; sweep it and keep the curve that flatters the patch. CHARACTER sharpens the saturation curve itself across all stages, and it is loudness-matched end to end: the cutoff frequency stays put and so does the level, so the knob can be swept as a pure texture control from smooth to snarling with nothing else moving under it. Hard settings still read as more squash — the peaks come down and the harmonic content thickens — they simply no longer read as less volume. It is audible at any resonance setting, not only extreme ones. The match is calibrated for normal program level; a very quiet filter input or a fully-driven one can still drift a couple of decibels across the knob’s travel.

The filter envelope

This is where the filter starts playing notes with you. The filter has its own envelope — attack, hold, decay, sustain, release, fully independent from the volume envelope — so the brightness of a note can move on a completely different schedule from its loudness: a slow volume swell with an instant filter bite, or a percussive amplitude hit whose harmonics bloom afterwards. The stages are edited in the envelope panel at the bottom of the OSC tab — its tab strip reads VOLUME ENV / FILTER ENV / TRANSIENT ENV — switched to FILTER ENV. Both filters follow this one envelope, each by its own amount: the last two sliders in the envelope row, Filter 1 and Filter 2, are those amounts, and each filter’s panel on the FX tab carries the same control as its ENVELOPE knob. An amount dims while its filter is off, but stays editable.

The amount sets how far the envelope reaches. It is bipolar: positive values sweep the cutoff up from where you parked it, negative values flip the whole movement downward — the same envelope becomes a closing gesture instead of an opening one — and at full amount the sweep spans eight octaves. KEY TRACK completes the picture by making the cutoff follow the played note, semitone for semitone at the full setting: without it, high notes duck under a fixed cutoff and low notes sail through; with it, the filter keeps the same voicing across the keyboard (each filter has its own KEY TRACK) — mandatory for basslines that walk, and for playing a self-oscillating filter in tune.

The stages themselves are shapeable. Attack, decay and release each have a shape control that bends the curve from logarithmic through linear to exponential, plus an S-curve switch that splits the curve into an ease-in/ease-out pair — the shape slider then sets how pronounced the S is. HOLD pins the envelope at its peak between attack and decay, useful when a sweep should sit fully open for a beat before falling. DIP pulls the envelope down briefly when the sequencer retriggers a note that is still sounding, giving re-articulated steps a percussive filter pluck; like the volume envelope’s Dip it acts on sequenced lines only — live MIDI takes the smooth path — and it runs the same law, where depth buys length as well as depth — see The volume envelope (Chapter 4 — Modulation). LOOP switches the envelope from one-shot to a repeating attack–hold–decay cycle for as long as the note is held — a rhythmic wobble with no LFO involved. The four time stages — attack, hold, decay and release — can also leave milliseconds behind: each has its own sync toggle and division selector, locking that stage’s time to the host tempo.

The filter envelope also answers your playing the way the volume envelope does. A key-track slider shortens its stages as you play higher and lengthens them lower down — this is envelope timing, distinct from the KEY TRACK knob above, which moves the cutoff itself — and four bipolar velocity routes shift the attack, decay and release times — up to a second either way at full depth — and offset the sustain level with how hard you hit, using the same formulas as the volume envelope’s routes (see Velocity and aftertouch (Chapter 4 — Modulation)). All five sit at zero by default and only act when raised, so a patch that never touches them keeps its filter movement identical on every note. A copy button (COPY VOL) pulls the volume envelope’s settings across — key track and velocity routes included — when you want both movements to start in lockstep. One housekeeping note: RANGE, in the table below, rescales the envelope panel’s editing time axis so long stages stay visible — in the current v0.3.0 early-access builds it does not itself change the sound.

Cutoff, resonance, drive, envelope amount, pole morph, bass lock, CHARACTER, STEREO SEP and MIX are live modulation targets on both filters, beyond this envelope — see the modulation chapter (Chapter 4 — Modulation) for LFOs and step modulators pointed at the filter.

Formant

A formant is a resonant peak that stays put while the pitch moves underneath it — it is why a singer holding two different notes still sounds like the same vowel. Phaseburn One has no single formant knob; it has three honest ways to build one, and they stack.

The first is the filter you just met. A bandpass response with moderate resonance and KEY TRACK left at zero is a formant by definition: the emphasized band stays fixed while the sequencer walks notes under it, and the patch takes on a vowel-like identity. Automate the cutoff gently between two positions and the vowel mouths open and closed; on the SVF, the peak region of the pole-morph path is made for exactly this. Keep the resonance short of self-oscillation — a formant is an emphasis, not a whistle.

The second lives in the VECTOR oscillator. Its FORMANT switch retunes the Y control into a formant-position control, planting resonant peaks directly on chosen harmonics of the played note — formant synthesis as described in the original Vector Phase Shaping research (Kleimola et al. 2011). This produces vocal, resonant character at the source, before any filtering happens. The third is MORPH’s formant wavetable mode, which sweeps through vowel shapes directly with the morph slider. Both are covered in the oscillators chapter (Chapter 2 — Oscillators).

The combinations are the point. Generate a vowel at the source with the VECTOR FORMANT switch, then focus it with a fixed bandpass from this chapter — the oscillator supplies the character, the filter supplies the mouth. Or run MORPH’s vowel sweep into a slowly morphing SVF and let two formant systems drift against each other. Because everything upstream is one voice by design, the result stays a single voice with a single throat — which is precisely what makes it read as vocal.