A voice is a single instrument track. Each voice has a search query that pulls random samples from the catalog. The voice loads 8 random matches into a sample pool and rotates through them as the pattern plays.

Searching for sounds

Type a query into the search field (e.g. kick, snare, 808, vocal chop). The field has an autocomplete with common instrument types. Press Enter or click away to trigger a search — the pool will reload with 8 random matches.

The re-roll button (↹) refills the pool with the same query — you get different random samples without changing instrument type. The dice button (🎲) on the voice header picks a random instrument type (kick, snare, hi-hat, etc.).

Sample picker

The dropdown above the search field shows all 8 samples in the pool by filename. Selecting one pins it — the voice will always play that sample regardless of rotation settings.

Pin, mute, and solo

Pin (📌) locks the current sample so it never changes. M silences the voice. S solos it — all other voices are muted. Multiple voices can be soloed simultaneously.

Volume & pitch

VOL controls the voice’s output level (0–100%). PIT shifts the sample pitch ±12 semitones (one octave up or down). Pitch changes affect both tuning and duration — higher = faster, lower = slower. This is the same mechanism used in classic drum machines.

The step grid is a row of cells — one per beat. Gold cells fire the voice on that beat; dark cells are silent. Click any cell to toggle it.

The step count dropdown (default 16) sets how many steps are in the loop. Every voice can have a different step count. Different lengths across voices create polyrhythms — a 16-step kick against a 12-step hi-hat will phase in and out.

Euclidean rhythms

Euclidean rhythms are a mathematical way to distribute pulses evenly across a pattern. Three sliders above the step grid let you dial in patterns without clicking individual steps:

• P (pulses) — how many hits to place. Must be between 1 and L.
• L (length) — the total number of steps, 1–32. Changing this also updates the step count.
• R (rotation/offset) — shifts the pattern left by this many steps.

Drag a slider to generate a new pattern. The step grid updates instantly. You can still click individual step cells to tweak the pattern manually — your edits stay until you touch a euclidean slider again. It’s a pattern generator, not a mode — both tools work together.

Common Euclidean patterns:

P,L	Name / feel
3,8	Tresillo — the backbone of Latin and Middle Eastern music
4,16	Four-on-the-floor — kick drum in house/techno
5,16	Bossa nova / clave
7,16	Complex Afro-Cuban bell pattern
5,8	Cinquillo
2,3	Hemiola — two against three. Try R=1 or R=2 for variation

Each step can have its own overrides — pitch, volume, probability, and trigger conditions. The meta-row has three toggles for per-step work: SEL, PROB, and AUTO.

SEL — Step select (p-locks)

Click SEL to enter select mode. Click a step to select it — VOL and PIT now edit that step’s values. Labels turn blue. Steps with overrides get a blue corner dot. The override bar shows × CLR (clear), COND (trigger condition), and MACRO (preset bundles).

PROB — Step probability

Click PROB for probability mode. Each cell shows a fill bar. Click to cycle: 100% → 75% → 50% → 25% → 0% → clear.

AUTO — Automation lanes

Click AUTO to see a heat map. Toggle PITCH (color: blue low, red high) or VOL (brightness) with the second button.

Trigger conditions

Cond	Fires on
1:2	Every other loop
1:4	Every 4th loop
1:8	Every 8th loop
2:2	Skip first, play second
3:4	3rd of every 4 loops
4:4	4th of every 4 loops
PRE	Only if previous step fired
NOT PRE	Only if previous step didn’t fire

Macros

Macro	Effect
accent	Full volume, pitch +1
ghost	30% vol, 30% chance — ghost note
soft	50% volume
hi	Pitch +4, 90% vol
lo	Pitch -5, 90% vol
x	Pitch +7, 50% chance

Each voice has a play style that controls how samples are triggered and stopped. Open the ENV panel and use the STYLE dropdown to change it.

Style	What it does	Use for
one-shot	Sample plays to completion each time a step fires. Multiple overlapping triggers are allowed (polyphonic).	Kicks, snares, one-shot FX
cut	Each new step immediately stops the previous one. Only one trigger plays at a time (monophonic). No release ramp on choke.	Hi-hats (open hat cut by closed hat), any monophonic line
gate	Sample plays while a step is active and stops with the release envelope when the step becomes inactive. Each step re-triggers the sample.	Percussive sounds that shouldn’t ring out, rhythmic gating
legato	Like gate, but if consecutive steps are active the sample continues without re-triggering. Only re-triggers when coming from an empty step. Releases when the last active step ends.	Sustained sounds, synth lines, vocal phrases

The ENV panel controls how each trigger fades in and out. It has an interactive visual envelope display.

Drag the handles on the envelope curve:

• Left handle (attack) — how long the sample takes to fade in from silence. 0s = instant. Max = 2 seconds.
• Right handle (release) — how long the sample takes to fade out. 0s = instant cut. Max = 3 seconds.

The area between the handles is the sustain portion — the sample plays at full volume here. For one-shot samples, the envelope is applied as a fade-in/fade-out over the full sample duration. For gate and legato, the release happens when a step becomes inactive.

Curve shaping

Each phase has a lin/exp dropdown:

• lin (linear) — a straight ramp from 0 to 1 (attack) or 1 to 0 (release).
• exp (exponential) — an eased curve. On attack it starts slow then accelerates. On release it drops faster at first then tails off.

Preview

The ▶ preview button plays the current pinned/active sample with the envelope applied so you can hear your settings. Enable auto to trigger a preview automatically when you adjust parameters (with a short debounce so you don’t get rapid-fire playback while dragging).

Each voice has a per-voice effects chain in this order: delay → reverb → filter. Open the FX panel to adjust. All effects are applied to the voice only — the master section has global controls.

Delay

• TIME — delay interval, 10ms to 1 second
• FB (feedback) — how much of the delayed signal feeds back in, 0–95%. Higher = more echoes
• WET — dry/wet balance. 0 = no delay, 1 = full wet

Reverb

• WET — dry/wet balance. Uses a synthetic impulse response for a small-room sound. 0 = dry, 1 = full reverb

Filter

• TYPE — lowpass (LP), highpass (HP), bandpass (BP), or notch
• FREQ — cutoff/center frequency, 80Hz to 20kHz
• Q — resonance. Higher = sharper peak around the cutoff frequency, 0.1–20

Each voice pre-loads 8 random samples in its pool. The rotation dropdown controls how often the pool advances to the next sample.

Setting	Advances	Best for
/HIT	Every time a step fires	Maximum variation — different sample each hit
/BAR	At the start of each bar (step 0)	Changes each loop — controlled variety
/4BR	Every 4 bars	Long-form consistency, then a new sample for variation
PIN	Never — always plays the pinned sample	You picked a specific sample you want to keep

The sample picker dropdown shows all 8 pool entries. Picking one is equivalent to pinning it. Use the re-roll button to get a fresh set of 8 random samples if you don’t like the current batch.

The MASTER bar at the bottom has global controls that affect the entire mix.

• VOL — master output level
• COMP — compressor with adjustable THR (threshold, -60 to 0dB) and RATIO (1:1 to 20:1). Collapses dynamics to make the groove punchier.
• 🔗 SHARE — copies a URL to your clipboard. The URL encodes the entire app state (voices, patterns, FX, BPM, envelope settings, euclidean parameters) so the recipient opens your exact groove.

Toolbar buttons

The toolbar has randomize buttons for steps, queries, BPM, and voices. The 🎲 ALL button randomizes BPM + all voices at once. The 💥 CHAOS button (red explosion) randomizes every parameter — patterns, FX, envelopes, pitch, play styles, rotation, euclidean settings, and BPM. Use it for instant creative destruction.

Key	Action
P / Space	Play / stop
⌘Z	Undo (last 50 actions)
⌘⇧Z	Redo
↑ / ↓	BPM ±1
→ / ←	BPM ±5