KAZU is a Creative Multi FX Processor with smart modulation. It is not one effect — it is a rack of up to twelve effects whose parameters are driven by a matrix of generative modulation sources. You design a signal chain, wire sequencers, curves, LFOs and followers into any knob, and the machine plays the effects for you — in time with your host, or free-running.

Everything is deterministic where it matters: sequences are seeded, patterns survive preset saves, and the whole state — rack, wiring, curves — travels inside a single preset file.

THE CORE LOOP

• 1. STACK — add effects to the rack from eight categories.
• 2. WIRE OR ANIMATE — attach modulation sources to any effect parameter by hand, or hit ANIMATE to let the machine generate coordinated motion across the whole rack.
• 3. GENERATE — roll seeds, draw curves, pick vibes; let GEN and MAP write the motion.
• 4. PERFORM — flip to PLAY and drive it all from the X/Y pad and four macros.

The window has two pages, toggled from the header:

PAGE	PURPOSE
DESIGN	Build the machine — FX rack on the left, modulation column and sequencer workspace on the right, the master rack via the button.
PLAY	Perform the machine — X/Y pad, glyph mosaic, and MACRO 1–4 with their assigned parameter targets.

UNIVERSAL GESTURES

GESTURE	EFFECT
DRAG ↑↓	Edit any knob or value — vertical drag, full range over ~150 px of motion.
SHIFT + DRAG	Make FX values and effect output gain adjustments at one tenth of the usual speed. Press or release Shift during a drag to change sensitivity without a value jump. Percentage values show tenths while fine-adjusting; synced timing and named choices retain their valid steps.
DOUBLE-CLICK	Restore a parameter to its default value. On a macro name: rename. In the graph editor: remove a node.
HOVER	With on, every control explains itself in place.
RETURN / ESC	Confirm / cancel any text entry (SAVE AS, macro rename).

Left to right, exactly as the tooltips name them:

CONTROL	FUNCTION
	The main menu — see the version, check for updates, manage your LICENSE, switch effect ROUTING (serial / parallel), change the THEME, set the WINDOW SIZE (50–200 %), toggle ANIMATIONS, OPEN USER PRESET FOLDER in your file browser, open this MANUAL, or visit the Phase Fiasco site.
	The preset name field. Click to open the browser; a trailing * marks unsaved changes.
	Step through the preset list in place.
	Opens the SAVE / SAVE AS mini-menu.
	Full-state history — every gesture is one undo step, including rack edits and mod wiring.
	SEQ — the sequencer workspace.
	Root note + scale for every pitch-snapped control. See 09.
	Flips between the build page and the performance page.
	The m button — shows the master rack panel. See 10.
	Toggle in-place help text for every control.
	The wand. Generates coordinated motion for every effect in the rack — parameters get sequenced, not just randomized.
	The dice. Rerolls every effect's parameters to new static values.

With an empty rack, the wand and die refuse politely: ADD EFFECTS TO ANIMATE.

Twelve slots, processed top to bottom — in series by default, or switch the whole rack to parallel from the main menu's ROUTING. Click on an empty slot — OPEN THE EFFECT BROWSER FOR THIS SLOT — and pick from the eight categories. Effects land instantly with sensible defaults; DSP for a slot only runs once an effect occupies it.

SLOT HEADER ACTIONS

Hover an effect's title row to expand its action strip:

ACTION	UI STRING
	ADJUST THIS EFFECT'S OUTPUT GAIN
	ANIMATE THIS EFFECT'S PARAMETERS — runs the generative wand on this effect only.
	RANDOMIZE THIS EFFECT'S PARAMETERS
	MOVE THIS EFFECT UP / MOVE THIS EFFECT DOWN — reorder the chain.
	MUTE THIS EFFECT / UNMUTE THIS EFFECT — bypass in place.
	CLONE THIS EFFECT — duplicate into the next free slot, settings and all.
	REMOVE THIS EFFECT

THE AMBER AUX TOGGLE

Many effects carry an amber dot on one knob. It flips that knob between two personalities — most commonly USE CLOCKED RATE (tempo divisions synced to the host) vs USE FREE RATE (continuous Hz/percent). On pitch effects the same slot becomes scale snap (ENABLE SCALE SNAP), and stereo effects use it to link channels — LINK LEFT AND RIGHT RATES, LINK LEFT AND RIGHT SHIFTS.

In the FX REFERENCE, effects with an AUX toggle are marked AUX.

The SEQ button in the header opens the effect sequencer — a rhythmic on/off gate for the whole rack. It is separate from the modulation sources: GEN, MAP, GRAPH, LFO and the followers move parameter values, while this simply switches each effect's wet output in and out of the mix, in time with the host.

• One row per effect slot, each with a 32-step grid (in four groups of eight), plus a LENGTH and a RATE. Empty slots are greyed out.
• Click a step to toggle it; drag across a row to paint. An OFF step drops that effect's wet output for that step; ON lets it through. Every step starts ON, so the sequencer stays silent until you switch some off.
• LENGTH sets the loop length (1–32 steps); RATE its clock division, tempo-synced to the host. A playhead rides each row while the transport plays — stopped, the gate is ignored and every effect passes normally.
• The dice on a row rerolls that row's steps; the dice in the header rerolls all rows. Patterns are saved with the preset.

The right-hand column holds your modulation sources. Click the tile — ADD A NEW MODULATION SOURCE — and choose from the picker. Each source occupies one lane with a live preview of its output and its own compact control strip.

WIRING A SOURCE TO A KNOB

• Or wire straight from the knob: hover any parameter and click its — ADD MODULATION TO it — to open the source picker and attach a source directly. A wired knob then carries an invert (INVERT MODULATION FOR) and a remove (REMOVE MODULATION FROM) control beside its depth bar.
• Hovering a parameter also reveals an ANIMATE button — the wand scoped to a single knob; click it to generate motion for just that parameter.
• Every source lane has — ASSIGN ANOTHER PARAMETER TO THIS SOURCE — and a list of its current targets; each row carries a release-✕ to detach.
• Depth is bipolar and per-target: the blue strength bar sits beside the wired knob. Drag vertically; it reads – when unassigned.
• One source can drive many knobs; one knob can be driven by many sources. Depths sum.

SOURCE LANE ACTIONS

ACTION	UI STRING
	ASSIGN ANOTHER PARAMETER TO THIS SOURCE
	RANDOMIZE THIS MODULATION SOURCE
	BYPASS THIS MODULATION SOURCE
	CLONE THIS MODULATION SOURCE
	REMOVE THIS MODULATION SOURCE

Invariant worth knowing: a source's pattern lives with the parameter wiring, and patterns are never reset by adding or removing targets. Your sequence survives every rack edit.

Hit PLAY / DESIGN and the build surface folds away into a performance instrument: one X/Y pad, a reactive glyph mosaic, and four macros.

Parameters owned by the PLAY page are tagged PERF back on the DESIGN page, so you always know who is holding a knob.

The CURVE action opens the ADJUST CURVE editor below — reshape how a macro sweeps its target, with MIN, SKEW, MAX, START and END controls.

The header button sets a root note and scale for the whole instrument. Any pitch-bearing knob whose amber AUX is set to ENABLE SCALE SNAP — PITCH SHIFT's semitones, TOOTH's comb offsets, CRYSTAL's chords — quantizes to that scale instead of free values.

Sequenced pitch becomes melody: wire GEN into a scale-snapped PITCH knob and the generator plays in key.

After slot 12 the signal hits the always-available master rack — a set of master-bus tools at the end of the chain. Its first tool is a multiband "filth" compressor with its own gain-reduction meter, with room for more master tools alongside it. Toggle the panel with the header's , and switch the compressor on with ENABLE MELT MASTER COMPRESSION.

The compressor's mode switch chooses CLASSIC or three-band MELT processing. Want the same voicing inside the rack, pre-master? Load the MANGLE effect.

A preset is the entire machine: rack order, every knob, every mod source, its pattern seeds, drawn graphs, macro wiring, master settings. Files are .cfp on disk; factory and user banks share one browser.

• BROWSE — click the name field, or step with ◁ ▷. Loading replaces the current state.
• SAVE / SAVE AS — the floppy opens a mini-menu; SAVE AS takes a name inline (RETURN confirms, ESC cancels).
• DIRTY MARK — a * after the name means unsaved edits.
• UNDO / REDO — full-state snapshots per gesture; browsing away and back is one undo step too.

Host automation and DAW session recall capture the same state — the preset system and your DAW project never disagree.

Fifty effects across eight categories. AUX marks an amber toggle (clock sync, scale snap, or channel link — see 04). Knob order matches the on-screen layout.

BITROTAUX

Bit-depth and sample-rate destruction with seven crush models, dither, jitter and pitch tracking.

BITS

Bit depth, 16 down to 1.

RATE

Sample-rate reduction — the converter's hold clock. Free: 0–100 %. Clocked (AUX): hold period as beat divisions, 1/1 down to 1/512.

DRIVE

Saturation into the crusher; 0 = clean.

DITHER

TPDF noise ahead of the quantizer — masks the stairsteps.

JITTER

Randomises the hold-clock timing (analog-converter warmth).

MIX

Dry / crushed blend.

MODEL

Quantizer character: CLAS (classic sample-and-hold), TRED (mid-tread — quiet signals gate to true silence), ULAW (µ-law companding, telephone / 12-bit sampler), FLOT (block-float — grit independent of level), DELT (1-bit delta modulation), GLID (slews between captures — rubbery portamento), ERR (noise-shaped — grit pushed up into hiss).

ROT

Acquisition failures per hold cycle — the converter re-holds a stale value (warble) or drops to zero (sputter).

SMOOTH

Post-hold reconstruction filter: 0 = raw stairsteps, up = rounded “vintage DAC”.

WIDTH

Decouples the left and right hold clocks (right runs up to 25 % slower with its own jitter and rot). 0 = mono-locked crush.

RANGE

Randomises the bit depth every hold cycle, within ±(RANGE × 10) bits of BITS.

TRACK

Locks the hold rate to the detected input pitch (OFF, ¼× … 8×) so aliasing partials land harmonically. Overrides RATE and the clock.

AUX

On RATE — USE CLOCKED RATE (beat divisions) vs USE FREE RATE. Starts free.

CIRCUIT BENDAUX

Clocked stutter and bend — corrupted, stalling repeats like a shorted toy keyboard.

SHIFT

Data-bus damage — bit rotation plus XOR on every sample, 0 to 8 steps.

CORRUPT

Address-line shorts — pairs of address bits stuck together, so reads land in the wrong place. More = more swaps.

STALL

Probability of a stuck-CD buffer repeat at each chunk boundary.

RATE

Chunk length for stall events. Free: 2 s down to 10 ms. Clocked (AUX): 1 beat down to 1/64.

HQ

2× oversampled bit mangling — cleaner top end, less aliasing.

MIX

Dry / bent blend.

AUX

On RATE — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

DISTORT

Twelve-model waveshaper with bias and low/high-cut frequency focus.

MODEL

Twelve shapers: SOFT (tanh), HARD (clip), TUBE (exponential knee), FOLD (triangle wavefold), SINE, ASYM (different curves per half), RECT (full-wave rectify), FUZZ (cascade), MUFF (two clip stages with lowpass between), PL+2 / PL+3 (add the 2nd / 3rd harmonic), MULT (three-band split, per-band soft clip).

DRIVE

Gain into the shaper, with automatic makeup.

BIAS

DC offset ahead of the shaper — asymmetry and even harmonics. Centre = none.

MIX

Dry / distorted blend.

LOW CUT

Bottom of the focus band, 20 Hz–20 kHz. Only what lies between LOW CUT and HIGH CUT is driven; everything outside passes clean.

HIGH CUT

Top of the focus band, 20 Hz–20 kHz.

EXCITE

Aural exciter — synthesized top-end sheen with program-dependent attack and release.

DRIVE

Pre-gain into the harmonic generator, 1× to 4×.

TONE

Highpass corner of the generator, 500 Hz–10 kHz — where the sheen starts.

ATTACK

Envelope-follower attack, 0.1–50 ms — how fast the harmonics react to transients.

RELEASE

Envelope-follower release, 10–500 ms.

MIX

Level of the added harmonics. Parallel topology — the dry signal always passes at unity.

MIRROR

Wavefolder with a spectral trail — folds the wave and lets its spectrum smear behind it.

FOLD

Fold drive, 1× to 13×.

SHAPE

Morphs the fold curve from sine (soft) to triangle (hard).

FB

Feeds the folded spectrum into a decaying spectral trail that shimmers behind the fold. 0 = fold only, trail engine asleep.

MIX

Dry / folded blend.

PLASMAAUX

Hysteretic arc-discharge distortion — an unstable arc that strikes and collapses around the signal.

DRIVE

Gain into the arc gap — the “voltage” pushed at the breakdown threshold.

ARC

Strike threshold. Low: only the loudest peaks ionise, then the arc dies — chopped, sputtery, gated. High: the arc sustains — continuous heavy distortion.

FREQ

Arc-oscillator pitch, 200 Hz–12 kHz — the buzz that modulates the arc. Snaps to the global scale with AUX.

MIX

Dry / arc blend.

AUX

On FREQ — ENABLE SCALE SNAP: the arc pitch locks to the global key and scale.

TAPE

Tape saturation with wow, flutter, hiss, dullness and hysteresis.

MODE

Saturation curve: GENTLY, NORMAL, HARD, HARSH.

BASS

Low-frequency tilt into the saturator (pre-emphasis with an exact inverse after) — drives the lows harder without changing the tone.

TAPE

Saturation amount.

NOISE

Tape hiss, modulated by the programme envelope.

DULL

High-frequency loss — a lowpass from 20 kHz down to ~3 kHz.

WOW

Slow speed drift (pitch wobble).

FLUTTER

Fast speed variation.

GAIN

Output trim, −12 to +12 dB.

HYST

Magnetic hysteresis — memory feedback inside the nonlinearity.

MIX

Dry / tape blend.

WIDTH

Decorrelates left and right wow / flutter. 0 = both channels share the same motion.

WAVEFOLDER

West-coast folder — drive the wave back into itself, from sine folds to hard wraps.

DRIVE

Gain into the folder, 1× to 40×.

FOLDS

Cascaded fold stages, 1 to 6 (fractional positions crossfade).

SHAPE

Fold curve: SINE → TRI → HARD (flat-shouldered).

BIAS

DC symmetry offset, ±50 % — the Buchla “symmetry” control.

MIX

Dry / folded blend.

PITCH SHIFTAUX

Pitch / frequency / ring multi-shifter with granular density, spray and a feedback delay line.

PITCH / FREQ

In PITCH mode: transposition, ±24 semitones. In FREQ / RING modes the same knob is the carrier, ±5 kHz on a cubic curve (fine near the centre, metallic at the extremes).

DENSITY

Grain window length — higher is shorter (350 ms down to 10 ms) and therefore denser.

SPRAY

Random jitter on the granular read position.

MIX

Dry / shifted blend.

MODE

PITCH (granular shifter), FREQ (single-sideband frequency shift), RING (ring modulation). All three engines stay warm and crossfade on switch.

FINE

±100 cents in PITCH mode; ±50 Hz in FREQ / RING.

SPREAD

PITCH: detunes left up and right down by up to 1 semitone. FREQ / RING: opposite carrier phase offsets — width without the image wandering.

TIME / RATE

Delay time for the built-in delay line. Free: 1 ms–2 s. Clocked (AUX2): 1/32 to 1/2 note.

DELAY

Delay feedback, 0–100 %. The minimum switches the delay OFF entirely.

TONE

Lowpass inside the delay feedback loop, 400 Hz–20 kHz.

AUX

On PITCH — ENABLE SCALE SNAP: the shift lands on the global key and scale.

AUX2

On TIME — USE CLOCKED RATE vs free milliseconds. Starts clocked.

FREQ SHIFTAUX

Bode-style single-sideband shifter — inharmonic detunes and barberpole feedback per channel.

SHIFT L

Left-channel frequency shift, ±500 Hz. Positive moves every partial up by the same number of hertz — inharmonic.

SHIFT R

Right-channel shift, ±500 Hz.

TIME

Feedback-delay time, 0–500 ms — repeats re-shift each pass (barberpole).

FDBK

How much shifted signal feeds back through the delay.

SPREAD

Stereo separation of the wet signal: 0 = mono, 100 = full left / right.

MIX

Dry / shifted blend.

AUX

On SHIFT L — ENABLE SCALE SNAP for both shifts: the shift amount snaps to note frequencies of the global scale.

LINK

Inline toggle between the two shifts — LINK LEFT AND RIGHT SHIFTS moves them together.

MOUTHAUX

Phase-reset robotizer / whisperizer — locks speech to a pitch or strips it to breath.

PITCH

The robot's monotone pitch, 80–500 Hz — literally the STFT frame rate.

VOICE

ROBOT (every bin at zero phase) morphs to WHISPER (every bin at random phase).

SIZE

Requested analysis window: 256 / 512 / 1K. Low pitches automatically borrow a larger transform.

GATE

Spectral gate relative to the frame peak, OFF to −20 dB — silences quiet bins.

TONE

Tilt of the resynthesised spectrum, ±12 dB/oct about 1 kHz.

MIX

Dry / robotised blend.

AUX

On PITCH — ENABLE SCALE SNAP: the robot pitch locks to the global key and scale.

VOCODERAUX

Channel vocoder with internal carrier — pitch-tracked, with hiss, sub and formant control.

PITCH

Internal carrier pitch, 40–800 Hz.

CARRIER

Carrier waveform: SAW, PULSE, MIX, NOISE.

BANDS

Analysis / synthesis bands: 8, 12, 16 or 24.

TRACK

Blends the fixed PITCH toward the detected input pitch (0 = fixed, 100 = fully tracked).

FORMANT

Shifts the synthesis bank against the analysis bank, ±1 octave — moves the formants.

SPEED

Envelope attack 1–100 ms, with a linked release.

HISS

Keyed passthrough above 5 kHz — restores consonants and sibilance the bank can't represent.

SUB

Adds a sub-octave saw to the carrier.

TONE

Tilt of the vocoded spectrum, ±9 dB/oct.

MIX

Dry / vocoded blend.

RESONANCE

Narrows the carrier filters — sharper, ringier bands.

SOURCE

INT analyses the audio in this slot; SC analyses the plugin's sidechain input. Either way the internal synth is the carrier.

AUX

On PITCH — ENABLE SCALE SNAP: the carrier pitch locks to the global key and scale.

SPEAKAUX

Order-10 LPC speech-chip codec — your audio re-spoken by a 1980s voice chip.

PITCH

Excitation pitch of the speech chip, 60–400 Hz.

TRACK

Blends the fixed PITCH toward the detected input pitch.

RATE

How often a new LPC frame is accepted, 5–50 frames per second. Slow = held, slurred syllables.

QUANT

Coefficient precision, 8 bits down to 2 — coarser is crunchier, more “chip”.

VOICE

Biases the voiced / unvoiced decision: WHISPER (all noise) to VOICED (all buzz).

CHIRP

Voiced excitation: IMPULSE train morphing to the stored TI-style glottal CHIRP.

ERROR

Probability of corrupted frames — held (stuck-word repeats), swapped or bit-flipped coefficients.

MIX

Dry / coded blend.

AUX

On PITCH — ENABLE SCALE SNAP: the excitation pitch locks to the global key and scale.

FORMANT SHIFT

Spectral-envelope shifter — moves the formants, leaves the pitch alone.

SHIFT

Moves the spectral envelope (the formants), ±24 semitones, leaving pitch untouched.

DETAIL

Width of the peak regions used to estimate the envelope, 0–16 — how much fine spectral detail counts as “formant” versus residual.

MIX

Dry / shifted blend, phase-aligned inside the DSP.

SIZE

STFT window: 512 / 1K / 2K / 4K. Fixed 4096-sample latency regardless.

CRYSTAL

Global-key resonator bank — blooming chordal resonances that wander and slip in scale.

CHORD

Prominence blend: 0 = the whole scale shimmers evenly; up = the PITCH note is foregrounded over the wash.

PITCH

The motif note, ±12 semitones from the root, snapped to the global key and scale.

OCTAVES

How many octaves of resonators, 1 to 7.

FOCUS

Resonator sharpness (Q 20–250).

BLOOM

How long the bank rings, 10 ms to 2 s.

MIX

Dry / resonant blend.

SKEW

Stretches or compresses the tuning between resonators, ±60 % — inharmonic when off-centre.

WANDER

Detunes alternate bands left / right by up to 30 cents.

SLIP

Frequency-shifts the input before the bank, ±500 Hz.

FLUTTER

Amplitude motion — every band pulses on its own slow cycle; rate and depth rise together.

COLOR

Tilt of the resonant output, ±8 dB.

STRETCHAUX

Granular time-stretch and tape-scrub with freeze, scatter and tempo capture.

SPEED

Read-head rate through the capture buffer: 0 = frozen drone, 50 = half speed, 100 = real time.

TEMPO

Capture window. Free: 2000 ms down to 250 ms. Clocked (AUX2): 4/4, 3/4, 2/4, 3/8, 1/4, 1/8.

PITCH

Independent grain pitch, ±24 semitones, without changing the stretch rate.

GRAIN

Grain rate. Free: 5–100 Hz. Clocked: 1/16 down to 1/512.

SPREAD

Pans the two grains apart: 0 = mono, 100 = hard left / right.

MODE

FWD / REV — grain and read direction.

FREEZE

Stops capturing and advancing — holds the buffer as it is.

MIX

Dry / stretched blend.

SCATTER

Random per-grain read offset — whole clock divisions when clocked, continuous when free.

MODE

LINKED keeps tape-scrub pitch tied to SPEED; INDEP gives each grain its own cursor so SPEED only moves the scan.

AUX

On PITCH — ENABLE SCALE SNAP.

AUX2

On TEMPO — USE CLOCKED RATE for TEMPO and GRAIN vs free. Starts clocked.

SCANNER

Clean granular time freeze — scan a held moment while transients pass through.

SCAN

Playback position inside the held capture: 0 = newest, 100 = oldest.

SIZE

Grain size, 10 ms to 1 s (logarithmic).

SHAPE

Where the grain envelope peaks: 0 = start, 50 = symmetric, 100 = end.

TRANSIENT

OFF = exact manual hold. Above that, input attacks re-capture automatically and land at a random point of the held history, snapped to older onsets.

MIX

Dry / frozen blend.

DUCK/FOLLOW

Bipolar: left of centre the live input ducks the frozen layer (D), right of centre the frozen layer follows the live envelope (F). Centre = OFF.

VARISPEED

Tape-speed capture — slews playback speed like a reel changing velocity.

SPEED

Tape speed and direction: 0 = 1× reverse, 50 = stopped, 100 = 1× forward. Passes continuously through stop.

TEMPO

Capture length, 4/4 down to 1/64 — each boundary jumps the read head back behind the write head with an equal-power crossfade.

WOW/FLUTTER

Tape-motion pitch modulation (0.5 Hz and 6 Hz rotators), from subtle to extreme.

TONE

Bipolar DJ filter on the wet: LP below centre, transparent at centre, HP above.

MIX

Dry / tape blend.

CHOPPERAUX

Clocked capture and probabilistic slicing — shuffled, repeating dice-roll edits.

SIZE

Length of each captured bank. Free: 125 ms to 2 s. Clocked (AUX): 1/4, 1/2, 1, 2 or 4 beats.

DIVISION

Slices per bank: 2 to 32 (2, 4, 8, 16, 32 when clocked).

SHUFFLE

Probability that a slice is swapped for a phrase from elsewhere in the bank.

REPEAT

Probability that a slice latches and repeats until the next decision point.

MIX

Dry / chopped blend.

AUX

On SIZE (mirrored on DIVISION) — USE CLOCKED RATE vs free. Starts clocked. Silent while the transport is stopped.

SLOW DOWN

Quantized forward capture player — grabs a bar and replays it slower, in time.

SPEED

Forward playback rate: 1X, 1/2, 1/4, 1/8, 1/16 or STOP.

CAPTURE

Capture length in musical time, 4/4 down to 1/64, locked to the host beat grid.

TONE

Bipolar DJ filter on the wet: LP below centre, HP above.

MIX

Dry / slowed blend.

STUTTERAUX

Continuously refreshed beat repeater — instant retriggers of the last moment.

FREQ

Repeat length. Free: 5 Hz to 1.6 kHz. Clocked (AUX): 1/4 down to 1/256 note.

CAPTURE

How often the repeated window is refreshed from the newest input, 4/4 down to 1/64 — always musical.

WINDOW

Continuity correction at loop and capture seams — longer is smoother, shorter is buzzier.

MIX

Dry / repeated blend.

AUX

On FREQ — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

WENDYAUX

Free or clocked stereo memory disorder — playback that seeks, slips and wears out.

SPAN

Memory period. Free: 12.5 ms to 16 s. Clocked (AUX): 4 bars down to 1/32.

CUTS

How many subdivisions each span is cut into, 1 to 32 — one of them repeats until the next boundary.

SLIP

One varispeed event per span — pitch displacement, direction change, slew and tape wear, all coherent for that span.

SEEK

A companion cut event: subsection relocation, nested repeats and rhythmic gaps.

WEAR

Frequency and length of smoothed signal dropouts after the buffer.

EDGE

Shapes repeat boundaries and read-head hand-offs — from crossfaded to abrupt.

WIDTH

Lets left and right make related but different choices (with a gentle low-end curve to keep the bass centred).

MIX

Dry / disordered blend.

AUX

On SPAN — USE CLOCKED RATE vs free. Starts clocked.

CHROMAAUX

Spectral filter — tilt, sweep and gate a resizable STFT window over the spectrum.

TILT

Spectral tilt, ±6 dB/oct about 1 kHz.

CENTER

Centre of the gaussian band, 20 Hz–20 kHz.

WIDTH

Width of the band in log-frequency.

GAIN

Bell (boost) or notch (cut) at the band, ±24 dB.

WARP

Spectral shift, ±12 semitones — moves energy between bins without resampling.

GATE

Peak-relative spectral gate — suppresses quiet bins.

MIX

Dry / filtered blend, latency-aligned inside the DSP.

SIZE

STFT window: 512 / 1K / 2K / 4K.

AUX

On WARP — ENABLE SCALE SNAP: the shift lands on the global key and scale.

EQ

Single-band parametric EQ — surgical or broad, one honest band.

MODE

HP, LSHELF, PEAK, HSHELF, LP.

FREQ

Band frequency, 20 Hz–20 kHz.

Q

Bandwidth, 0.1 to 10.

GAIN / DRIVE

In shelf and peak modes: ±18 dB. In HP / LP modes the same knob becomes DRIVE, 0–100 % of pre-filter saturation.

MIX

Dry / equalised blend.

FORMANT

Vowel formant filter — A/E/I/O/U through five voice characters.

VOWEL

A, E, I, O, U — five formants each.

Q

Formant sharpness.

SHIFT

Moves all formants, ±12 semitones.

SPREAD

Detunes the formants left / right, up to ±3 semitones.

VOICE

Formant set: MALE, FEMALE, CHILD, ROBOT, ALIEN.

MIX

Dry / vowel blend.

PRISM

Multi-mode filter — four models, four modes, four slopes, with drive.

TYPE

Topology: SVF (clean state-variable), LDR (Moog-style ladder), DIO (303-style diode ladder), DRTY (broken transistor with saturating rear stages).

MODE

LP, BP, HP, NCH (notch).

CUTOFF

Filter frequency.

RES

Resonance, with a feedback curve specific to each topology.

DRIVE

Input drive, 0 to +24 dB, with makeup.

MIX

Dry / filtered blend.

SLOPE

6, 12, 18 or 24 dB/oct — crossfades between pole orders without resetting.

SIEVEAUX

Spectral sieve — patterned bins pass or vanish on a clocked cycle.

PATTERN

Which bands survive: PRIME, FIB (Fibonacci), POW2, ODD, EVEN.

RATIO

How far the rejected bands are pulled down; 100 = removed.

SMOOTH

Temporal smoothing of the mask — softens the switching between frames.

RATE

Walks the pattern across the spectrum. Clocked (AUX): OFF, 1/1 down to 1/32 per step. Free: a continuous 0.05–20 Hz drift.

STEP

Manual band offset — nudges the pattern by hand.

MIX

Dry / sieved blend, latency-aligned inside the DSP.

SIZE

STFT window: 512 / 1K / 2K / 4K.

AUX

On RATE — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

SMEAR

Spectral blur — magnitude diffusion that melts attacks into washes.

SMEAR

Blur radius across neighbouring bins; 0 = clean.

TILT

Bends the blur toward the lows or the highs; centre = even.

FB

Per-bin temporal feedback — the spectrum lingers; 100 = a true freeze.

DIFF

Randomises phase into an ethereal cloud.

MIX

Dry / blurred blend, latency-aligned inside the DSP.

SIZE

STFT window: 512 / 1K / 2K / 4K.

TOOTHAUX

Three-comb resonator — a tuned root and two offsets, feedback into tone.

FREQ

Root comb frequency, 20 Hz–2 kHz.

OFF1

Second comb, offset from the root by ±24 semitones.

OFF2

Third comb, offset from the root by ±24 semitones.

FB

Bipolar feedback — negative inverts the loop so peaks and notches swap.

TONE

Bipolar: lowpass below centre, highpass above.

MIX

Dry / resonated blend.

AUX

On FREQ — ENABLE SCALE SNAP for FREQ, OFF1 and OFF2.

COMBAUX

Single feedback comb — one delay, one resonance, invertible.

FREQUENCY

Comb period as a pitch, 20 Hz–2 kHz.

RESONANCE

Feedback amount — level-normalised so the tone stays steady as it rises.

INVERT

Flips the feedback polarity: peaks and notches exchange places.

MIX

Dry / combed blend.

AUX

On FREQUENCY — ENABLE SCALE SNAP.

TRANSMIT

Helmet radio / damaged comms — bandpass, squelch, bursts and dropouts.

FREQ

Bandpass centre of the radio channel, 600 Hz–2.2 kHz.

WIDTH

Bandwidth, 0.8 to 3.5 octaves.

DRIVE

Overdrive inside the band-limited channel.

COMP

Envelope compression toward a constant level.

NOISE

Static keyed to the programme envelope.

BURST

Density and force of random static packets.

SQUELCH

Squelch threshold, OFF to −15 dB — with transition crackle and a noise tail when it closes.

DROPOUT

How often and how long the transmission randomly cuts to silence. Self-animating; no modulation needed.

MIX

Dry / transmitted blend.

SPECTERAUX

Wavetable convolution filter — scans a drawn table or spectrum across the signal.

TABLE

Which of the 22 wavetables shapes the filter.

SCAN

Sweeps through the table's frames like a cutoff; neighbouring frames blend.

PITCH

Transposes the whole response, ±24 semitones.

REGEN

Feeds the filtered output back into the convolution — resonant ringing at the response's peaks, never runaway.

WARP

Bends the response inharmonic about 400 Hz: up stretches (metallic), down compresses (gong-like). Inert while RAW is on.

TONE

Post tilt, dark to bright.

LOW

Folds the dry low band (below ~70 Hz, which the kernels remove) back under the filter.

WIDTH

Offsets the left and right SCAN positions.

RAW

MIN keeps attacks tight (minimum-phase kernels); RAW keeps each frame's own phase and smears attacks through it.

MIX

Dry / filtered blend.

AUX

On TABLE — READ THE FRAME'S SPECTRUM (SPEC) vs READ THE WAVEFORM AS A DRAWN CURVE (DRAW).

AUX2

On PITCH — ENABLE SCALE SNAP.

CHORUSAUX

Four-voice modulated-delay chorus with feedback and width.

RATE

LFO speed. Free: 0.01–10 Hz. Clocked (AUX): 1/1 down to 1/64.

DEPTH

Modulation depth.

TIME

Base delay of the four voices, 1–40 ms.

FDBK

Delay feedback, 0–85 %.

WIDTH

Equal-power spread of the four voices across the stereo field.

MIX

Dry / chorused blend.

AUX

On RATE — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

CLANGAUX

Ring-mod / grind — tracked or free carrier with its own LFO, sub and mask.

MODE

RING (input × carrier, 2× oversampled) or GRIND (the carrier modulates a 0.05–3 ms delay at audio rate — FM-like sidebands and digital grind).

FREQ

Carrier frequency (RING) or modulation rate and noise-band centre (GRIND), 20 Hz–5 kHz.

WAVE

Carrier: SINE, TRI, SQR, PULSE, NOISE.

RATIO

Snaps the carrier to harmonic ratios of the tracked pitch: FREE, then LK 0–100.

AMNT

Ring drive / sideband intensity (RING) or delay-modulation depth (GRIND).

MASK

Lowpass on the input feeding the engine, 20 Hz (dark) to 18 kHz (bright).

TRACK

Carrier glides toward the detected input pitch; OFF to 100.

LFO

Carrier-frequency LFO rate. Free Hz, or clocked divisions with AUX; 0 = off.

DEPTH

LFO depth, in octaves.

SHAPE

LFO waveform, 19 shapes from SINE to STAIR.

SUB

Blends a sub-octave sine into the carrier.

WIDTH

Left / right carrier phase offset, or independent L / R noise when wide.

MIX

Dry / clanged blend.

AUX

On LFO — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

AUX2

On FREQ — ENABLE SCALE SNAP for the carrier.

ENSEMBLE

Four-mode BBD ensemble — dual-rate bucket-brigade shimmer.

TYPE

JUNO (single triangle LFO, anti-phase stereo), SOLINA (three voices at 120°), DIMENS (trapezoid LFO with cross-feed), LUSH (three voices, two independent LFO layers).

RATE1

Main LFO speed.

DEPTH1

Main LFO depth.

RATE2

Second-layer speed — slow drift (JUNO), fast shimmer (SOLINA), cross-feed (DIMENS), second voice layer (LUSH).

DEPTH2

Second-layer depth.

FDBK

Feedback through the bucket-brigade line.

WIDTH

Stereo width of the wet signal.

MIX

Dry / ensemble blend.

FLANGERAUX

Through-zero flanger — jet swooshes that cross the zero point.

RATE

LFO speed. Free: 0.01–10 Hz. Clocked (AUX): 1/1 down to 1/64.

DEPTH

Sweep depth.

FDBK

Feedback, up to 95 %.

DELAY

Base delay, 0.1–12 ms.

TZF

Through-zero flanging — the wet delay crosses the dry, for the tape-flanger null.

MIX

Dry / flanged blend.

AUX

On RATE — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

PHASERAUX

All-pass cascade phaser with variable stages and center.

RATE

LFO speed. Free: 0.01–8 Hz. Clocked (AUX): 1/1 down to 1/64.

DEPTH

Sweep depth around CENTER.

FDBK

Feedback, up to 97 %.

CENTER

Sweep centre, 80 Hz–5 kHz.

STAGES

All-pass stages, 2 to 12 — more stages, more notches.

MIX

Dry / phased blend.

AUX

On RATE — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

FISSIONAUX

Extended shaped frequency modulation — FM as an insert, with bite and skew.

FREQUENCY

Modulator frequency: 20 Hz–20 kHz in AUDIO range, 0.05–20 Hz in SLOW range.

BITE

Modulator shape — a sine at centre, sharpened toward the edges.

MIX

Dry / modulated blend.

DEPTH

Delay excursion, 0–15 ms — the FM depth.

SPREAD

Modulator phase offset between left and right, 0–180°.

FEEDBACK

Bipolar feedback of the modulated signal into itself.

SKEW

Bipolar duty-cycle skew of the modulator.

RANGE

SLOW (vibrato / chorus rates) or AUDIO (FM sidebands).

RETRIG

Phase-locks the modulator to transport start without clearing the feedback tail.

AUX

On FREQUENCY — ENABLE SCALE SNAP (AUDIO range only).

ORBITAUX

Stereo auto-pan — shaped, smoothed, clockable orbits.

AMOUNT

Bipolar pan depth: centre = none, up = the shape as drawn, down = inverted.

RATE

Free: 0.1–6 Hz. Clocked (AUX): 1/1 down to 1/64.

SHAPE

SINE, TRI, UP, SQUARE, EXP UP, PEAK, SKEW UP.

SMOOTH

Morphs the shape toward a clean full-width sine — softens edges without shrinking the sweep.

MIX

Dry / panned blend.

AUX

On RATE — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

TREMORAUX

Linked stereo tremolo — amplitude pulses in lockstep.

AMOUNT

Bipolar depth: centre = none, up = dips as drawn, down = inverted. Gain only ever dips from unity.

RATE

Free: 0.1–20 Hz. Clocked (AUX): 1/1 down to 1/64.

SHAPE

SINE, TRI, UP, SQUARE, EXP UP, PEAK, SKEW UP.

SMOOTH

Morphs the shape toward a clean full-depth sine swell.

MIX

Dry / tremolo blend.

AUX

On RATE — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

SPIRALAUX

Stereo tape delay — L/R rates, wobble, noise, ducking, ping-pong, reverse and multi-taps.

L RATE

Left delay time. Clocked (AUX): 1/32 up to 2/1, 15 divisions. Free: 0–100 %.

R RATE

Right delay time.

FDBK

Saturated feedback, up to 99 %.

TONE

Bipolar: LP (500 Hz–15 kHz) below centre, HP (20 Hz–2 kHz) above.

WOBBLE

Dual-sine modulation of the delay time, up to ±3 %.

NOISE

Tape hiss riding the wet signal.

DUCK

Ducks the repeats while input is present — threshold from −60 dB up to OFF.

DUCKREL

Duck release, 10 ms–1 s.

PONG

Cross-channel feedback; at maximum the input is fed mono into the left side.

REV

Plays chunks of the delay buffer backwards.

TAPS

Tap pattern: OFF, DOT (dotted), TRIP (triplet), MULTI, SCAT (scattered).

MIX

Dry / delayed blend.

AUX

On L RATE — USE CLOCKED RATE for both channels vs free. Starts clocked.

LINK

Inline toggle between the rates — LINK LEFT AND RIGHT RATES moves them together.

SPECTRAL DELAYAUX

FFT per-bin delay — every frequency gets its own delay time, blur and feedback.

TIME

Base delay per bin, slow-to-fast (higher = shorter). Clocked (AUX): 1 down to 1/16 note. Free: 0–100 %.

TILT

Spreads the delay across the spectrum — lows longer or highs longer; centre = even.

FB

Recirculation.

BLUR

Temporal smear between frames.

SHIFT

Spectral pitch on the tail, ±12 semitones.

SPREAD

Left / right delay offset.

GATE

What is allowed into the delay line — suppresses quiet bins on the way in.

DAMP

High-frequency loss in the feedback.

LOCUT

Removes lows from the feedback tail.

SHAPE

Delay curve across the bins: LIN, EXP, LOG, STP (stepped), RND.

SIZE

FFT window: 512 / 1K / 2K / 4K. Fixed 4096-sample latency regardless.

FREEZE

Holds the current spectral content.

MIX

Dry / delayed blend, latency-aligned inside the DSP.

AUX

On TIME — USE CLOCKED RATE vs USE FREE RATE. Starts clocked.

AUX2

On SHIFT — ENABLE SCALE SNAP.

GRANULAR DELAYAUX

Grain-cloud delay — swung, sprayed, pitched repeats on musical intervals.

TIME

How far back each grain reads. Free: 1–128 ms. Clocked (AUX): 1/16 up to 16/16.

SWING

Offsets alternate grain timing, ±33 %.

SPRAY

Random delay-position offset per grain, up to 500 ms.

FREQ

Grain trigger rate. Free: 1–150 Hz. Clocked: straight and triplet divisions per beat.

PITCH

Grain transposition, −36 to +12 semitones.

RANDOM

Random per-grain pitch, up to ±12 semitones.

INTERVALS

Quantises the random pitches: OFF, OCT, OCT+5, MAJOR, MINOR, PENTA.

TONE

Bipolar DJ filter on the grains: LP below centre, HP above.

FEEDBACK

Grain output fed back, up to 95 %.

MIX

Dry / delayed blend.

AUX

On TIME — USE CLOCKED RATE for TIME and FREQ vs free. Starts clocked.

AUX2

On PITCH — ENABLE SCALE SNAP.

DATTORRO

Plate reverb — the classic figure-eight tank, sized 0.0025× to 4×.

SIZE

Tank size, 0.0025× to 4× the reference plate.

DECAY

Tank feedback — tail length.

DIFF

Input diffusion — how quickly the early energy smears into the tank.

PDLY

Pre-delay, 0–500 ms.

TONE

Lowpass on the wet, 500 Hz up to fully open (centre ≈ 8 kHz).

MIX

Dry / plate blend.

DRIP

Spring reverb — boingy tension with adjustable ripple.

SIZE

Spring geometry — delay 20–150 ms, dispersion base 0.2–2.2 ms.

DECAY

Tail length, T60 0.5–5 s.

DAMP

Lowpass damping, 18 kHz down to 2 kHz.

RIPPLE

Dispersion strength plus modulation depth — the spring's chirp and wobble.

MIX

Dry / spring blend.

GRANUVERB

Granular reverb — a cloud of scattered, pitchable grains as a tail.

SIZE

Reverb line lengths — the room.

GRAIN

Grain length, 20 ms to 500 ms; the spawn interval scales with it.

SCATTER

How far across the captured history grains may start, and how much of the cloud feeds back into the reverb.

DECAY

Reverb feedback; the top 1 % latches FREEZE.

PITCH

Grain pitch zones: OFF, +5TH, +OCT, −OCT, ±OCT, RND.

MIX

Three-way morph: dry → reverb → granular cloud (not a plain dry / wet).

GATE

Noise gate / downward expander with hold.

THRESH

Open threshold, −60 to 0 dB.

ATTACK

0.1–50 ms.

HOLD

Minimum time the gate stays open after the last crossing, 0–250 ms.

RELEASE

2 ms–1 s.

MIX

Dry / gated blend.

MANGLE

The master compressor's MELT voicing in the rack — drive, tilt, lift, pump and grit.

DRIVE

The intensity axis: input gain, gain reduction and saturation rise together, with compensating output trim. 0 = transparent.

TONE

Tilt between the low and high bands: L (darker) to H (brighter), FLAT at centre.

SPLIT

Low / high crossover, 60 Hz–2 kHz.

COMP

Blend of compressed against uncompressed signal.

MODE

CLASSIC (the master compressor's two-band voicing) or MELT (three-band, phase-aligned, with the controls below).

LIFT

MELT: lifts quiet tails, noise floor and band detail toward the ceiling.

PUMP

MELT: stretches low / mid recovery far more than the highs — a breathing spectral tilt.

GRIT

MELT: nonlinear intensity, rougher from low to high band.

SQUASH

OTT-style multiband compressor — upward and downward squash with a time multiplier.

DEPTH

Amount of upward and downward compression across the three bands.

TIME

Attack / release multiplier, 0.1× to 10×.

IN

Input gain, ±12 dB.

OUT

Output gain, ±12 dB.

MIX

Dry / squashed blend.

UTILITY

Gain, pan, auto-level, limiter and stereo width — the housekeeping slot.

GAIN

Attenuation only: −∞, then −36 dB up to 0 dB.

PAN

Equal-power balance, L100 to R100.

AUTO

Upward-only leveler toward −14 dBFS, capped at +18 dB of gain. OFF at zero.

LIMITER

Sample-peak ceiling, −24 to 0 dB — instant attack, 75 ms release.

STEREO

−100 = mono, 0 = unchanged, +100 = double the side signal. No MIX — this is a serial stage.

• LICENSE — paste your key into the boot terminal. Reopen it any time from the theme menu to check activation status or manage machines (KAZU // LICENSE CONTROL).
• UPDATES — KAZU checks for new builds and raises the update terminal when one is available; download or dismiss from there.

0.1.587 — FINE READOUT FIT

• Fine-adjustment readouts use the surrounding column space so decimal percentages fit on all platforms at the original font size.

0.1.586 — FINER FX ADJUSTMENTS

• Hold Shift while dragging FX values or effect output gain for 10× finer adjustment. Press or release Shift during a drag to change sensitivity without a value jump.
• Percentage readouts show tenths during fine adjustment. Synced timing and named choices retain their valid steps, and double-click still restores the default value.
• Added hover hints and manual instructions for Shift-drag. Active parameter readouts show the value being edited even when modulation data is stale.

0.1.580 — MODULATION HOVER SCROLLING

• Turning ANIMATIONS off now also stops modulation-source hover autoscrolling, keeping the source list in place while browsing parameters.

0.1.579 — RELIABLE UNDO / REDO

• Randomize and Animate actions now undo and redo in one step, for individual effects and the entire rack.
• Preserved edits made immediately before these actions and kept already-applied Animate changes undoable when the editor closes before the full batch finishes.

0.1.578 — NUENDO SCANNING FIX

• Fixed a Windows VST3 shutdown hang that could make Nuendo's plugin scanner time out or blocklist KAZU. Successful scanning was verified in Nuendo 15.0.30.
• SPECTER now prepares its tables on first use. Its first insertion or a preset load using it may pause for a few seconds while preparation completes.

DOWNLOADS & UPDATES

• Added a standalone Windows VST3 ZIP alongside the EXE installer, initially as a supplement to 0.1.577.
• Published 0.1.578 through public GitHub downloads and changed the updater's platform links to point directly to the macOS, Windows, and Linux packages.