SynthWorld synthesizers: user manual

Introduction #

• What this manual is: the parts every SynthWorld synth has in common. A synth's own processors and sounds are in its own manual.
• Where each panel is (tabs, pages, sizes) is up to each synth: this manual describes the panels, not their place. A synth may also leave some out (e.g. the preset browser or the graph).
• The panels: processors, modulator selector, mod matrix, graph, presets, settings, info box.
• The fastest way to learn a control: hover it and read the info box.
• Formats: CLAP, VST3, AU. Same synth, same controls in all of them.

a whole synth UI with each panel labelled

How a SynthWorld synth works #

The patch

• A patch is everything that makes the sound: every value, every connection, every modulation, plus the synth's settings and the graph's view.
• A preset is a patch saved in a file.
• Changes never click: values glide, connections and modulations fade in and out.

Processors

• A synth is a fixed set of processors (oscillators, filters, envelopes, LFOs…). You can't add or remove them; you turn them on, connect them and modulate them.
• Chained processors pass sound along signal cables to the output. Others only produce a signal used to modulate.
• Any processor's output can be a modulator, sound-producing ones included (that's how FM is made).
• What isn't used costs no CPU: a processor that reaches neither the output nor a modulation in use isn't computed.
• Several processors can share one place: processor containers.
• Some processors come with others built in after them (e.g. an oscillator with its own gain): extended processors.

Voices and the global side

• Per-voice processors: every note played has its own copy of them (its own envelope, its own filter…).
• Global processors: one of each, after the voices, working on all the notes added together. Not every synth has them. See The global side.
• How many notes at once, and what happens beyond that: Voices, Voice stealing.
• Note processors: before the voices, working on the notes played rather than on sound. Not every synth has them. See Notes before the voices.

Modulation

• A modulation makes a modulator move a parameter, by an amount.
• It happens at audio rate: a modulator can be as fast as sound.
• Almost anything can be a target: any processor parameter, and other modulations' amounts.
• Loops are allowed (A modulates B, B modulates A).
• Up to 32 modulations (and 32 more on the global side). Details: Modulation.

diagram: voices (per-voice processors) → sum → global processors → output, with modulation arrows

Basic controls #

Knobs and sliders

• Drag to change the value.
• Hold Shift or Ctrl while dragging for fine changes.
• Mouse wheel: small steps.
• Double-click: back to the default value.
• Right-click: Enter value, Reset to default, copy/paste value, copy/cut/paste modulations, Show info.
• The value's text: beside the control or inside it; left out inside a control too small to read it.
• On a modulated control: the coloured range and the moving mark. See Amount, range and live value.
• Small marks at a control's top left corner are the DAW's: Automation and mapping marks.

a knob and a linear slider, plain and modulated

Value bars

• A box of a few values side by side, each under its name, where a row of knobs would take too much room (e.g. an oscillator's tuning, or every value of a simple oscillator).
• Each value is a control like a knob: drag it up or down (anywhere in its section, name included), Shift or Ctrl for fine changes, mouse wheel, double-click for the default, the same right-click menu. Lit while the mouse is over it.
• Typing a value in: right-click → Enter value (a double-click resets, it doesn't open the value popup).
• Modulated like a knob: with a modulator selected, the +, then the amount handle and the modulation's buttons, are at the right of its section, in two rows; a modulator can be dropped on it. In a narrow section the name and the value make room for them, in smaller text, and have the whole section when there's none.
• The coloured range and the moving mark are on a line along the bottom of its section: its left end is the parameter's lowest value, its right end the highest. See Amount, range and live value.

a value bar of three values, the middle one modulated

Typing a value in

• Open the value popup: right-click → Enter value, or double-click the value's text when it's beside the control.
• Enter or a click outside keeps the value, Escape cancels.
• The same popup shows the value under any knob, slider or value box while it's dragged, the settings' and the note processors' included.
• The popup also shows the range the parameter's modulation sweeps.
• No typing for the view size, nor on the internal tempo's slider (its field beside it is where a tempo is typed).
• Values are typed in the parameter's own unit.

Copying and pasting values

• Right-click a processor's control: Copy value, Paste value.
• A value can be pasted into any other control: it lands at the same position of its range (the value itself when both controls have the same range).
• A modulation's amount handle has them too, for the amount.
• For all of a processor's values at once: Copying a processor's values.

Buttons and boxes

• Buttons: click. Toggle buttons stay on or off (power button, modulator button, a modulation's direction and on/off).
• Boxes (drop-down lists): click and choose.
• Tabs: click one to show its page. A row with more tabs than fit has an arrow at each end that hides some, over the tabs, which fade away under it: click it to bring the next one into view. An end with nothing hidden has no arrow.
• Lists (preset browser): click selects, double-click or Enter acts, Up/Down move.

Right-click menus

• Nearly everything has one; it holds whatever applies to what was clicked and to what it's inside (a knob's menu also has its processor's items).
• The last item is always Show info: see The info box.
• In a DAW that offers them, a parameter's menu also has the DAW's own entries for it.

Mouse cursors

• The cursor shows what the mouse can do there: click, double-click, drag up and down, drag sideways, move, connect a cable, type.
• List of the cursors with what each means.

the cursors, each with its meaning

Text fields

• Where text is typed: value popup, preset forms.
• Ctrl+C / X / V / A (Command on macOS), shared with the system clipboard.
• Tab next field, Enter confirms, Escape cancels. In a several-line field, Enter starts a line and Ctrl+Enter confirms.

Undo and redo

• The two arrow buttons (where the synth places them): undo takes back your last change to the patch, redo does it again. Each is dimmed while there's nothing to undo or redo.
• Ctrl+Z undoes; Ctrl+Y or Ctrl+Shift+Z redoes (Command on macOS), unless you changed these shortcuts. In a text field these keys are the field's.
• One step per action: a whole drag of a knob, a paste of a processor's values, a modulation added, removed or moved (with the modulations of its amount), a cable connected, moved or removed, a container's choice, a wavetable chosen, a setting, a loaded preset.
• Undone changes fade in and out like any edit; nothing clicks. Undoing a loaded preset brings the previous patch back whole.
• Only your own changes: what the DAW automates isn't undone.
• Not undone: moving around the graph (node positions, zoom), the view size, the skin, the keyboard shortcuts, the preset files themselves (saving, deleting...).
• The history stays while the synth's window is closed: close it, reopen it, and undo. It goes when the DAW loads a project or one of its own presets into the synth, and isn't saved with the project.
• A new change after undoing clears what could be redone. The last 200 steps are kept.

Comparing two versions: A and B

The A and B buttons, A lit, with their right-click menu.

• The A and B buttons (where the synth places them): two versions of the patch. The lit one is the one you hear and change; click the other to go to it.
• B starts as a copy of A: go to B, change it, then click A and B to compare the two.
• Right-click: Copy A to B (on B: Copy B to A) makes the other version the same as the one you hear. Not a step to undo: the patch you hear doesn't change.
• Everything you do is done to the version you hear, loading a preset included; the other stays as you left it.
• Switching is one step to undo: undoing it goes back to the other version, with nothing lost in either.
• Switching loads the other version whole, like a preset: the notes that sound are cut.
• Not compared: the graph's view (the same in both) and what isn't the patch (MIDI mappings, view size...).
• Only the version you hear is saved with the DAW project: reopened, it's A, and B is a copy again. Save the one you want to keep as a preset. A project or a DAW preset loaded into the synth starts again on A too.
• No key until you give it one: Switch A / B in the keyboard shortcuts.

Randomising the patch

The randomise button (a die) with its right-click menu.

• The die button (where the synth places it): moves the knobs and sliders of every processor that's on, at random, a quarter of the way to anywhere on their travel: a variation of the patch.
• Right-click to choose how far: Randomise slightly (a tenth of the way), Randomise (the click's quarter), Randomise a lot (half), Randomise everything (anywhere: another sound).
• What stays as it is: which processors are on and which a container has chosen, switches, choices and whole-number values, the modulations and their amounts, the cables, the synth-wide settings, bypassed processors.
• It doesn't know what a knob is: tunings and levels move like the rest, so a big amount can detune or silence a patch.
• One step to undo, done in place like any edit (the notes keep sounding). Undo then randomise again for another variation of the same patch; with A and B, keep the original in one and try variations in the other.
• No key until you give it one: Randomise in the keyboard shortcuts.

Panic: cutting all sound

The panic button (an exclamation mark).

• The exclamation mark button (where the synth places it): cuts everything that sounds, at once. For a note that hangs or a feedback that runs away.
• Ctrl+P does the same (Command on macOS), unless you changed the shortcut.
• What's cut: every note, with no release; the echoes and tails of the effects; the notes the note processors hold (an arpeggiator stops until a key is played again); the keys held on the computer's keyboard.
• Notes sent on to other instruments are ended there too.
• Nothing of the patch changes, so it's no step to undo, and the next note plays as usual. Keys still held down stay silent until played again.
• A sound that comes back at once is the patch's own (a feedback, a loud audio input): turn that down.

The info box #

• Shows what the control under the mouse is and how to use it, after a short hover.
• Parameters show their own description; controls without one show that of what they're in.
• Long texts scroll by themselves.
• Expand it: double-click it, or right-click any control → Show info (expands it on that control).
• Shrink it back: click outside it or press Escape.

the info box small, and expanded

Processors #

A processor's panel

• Its name, its power button, its controls (see Knobs and sliders).
• A modulator button, on processors meant as modulators.
• The same processor also appears as a node in the graph, where expanding the node shows the same controls. Both stay in step.
• What each processor does: hover its panel (info box), and the synth's own manual. The stock ones: The stock processors.

a processor panel with its parts labelled

Turning a processor on and off

• The power button: the thin bar at the left of the header. Off: a chained processor lets the sound through unchanged; a modulator stops modulating.
• Its light: purple while on, red while off (bypassed). In a stack, a processor turned off has no light: it's silent, not bypassed.
• Off, its panel is dimmed and its display goes dark: only a grey trace of what it draws is left, with no mark of where the processor is. Its controls still work, and the trace still follows them.
• An off processor uses no CPU, and neither do the modulations of its parameters.
• Its settings and its modulations are kept and come back when it's turned on again.
• On/off is part of the patch but can't be automated.
• Also in the right-click menu, anywhere in the panel or on its graph node: Bypass <name> while it's on, Activate <name> while it's off. The same as the power button.

Copying a processor's values

• Right-click anywhere in the panel or on its graph node: Copy <name> values, Paste values into <name>.
• Processor presets: <name> presets... in the same menu opens that processor's preset browser over the synth (the only way to it).
  • A processor preset is the values of one processor's controls, as a file. It's for a kind of processor: every processor of that kind lists it, in every synth of the maker.
  • The browser works like the preset browser: packages, folders, search, favourites, right-click menus, export. SAVE saves the processor as it is now; a double-click or LOAD sets its controls.
  • Loading changes the controls only: whether the processor is on, its connections, its modulations, the file it plays and its curve stay. One step to undo.
  • The browser stays open after a load, to try another. Escape, CLOSE or a click outside closes it.
  • Init, its first row: the processor as the synth's default patch has it.
  • A processor made of several (an oscillator with its pan and gain): one preset for all of it.
  • Folders are shared by every kind of processor, so one can be empty in this browser.
• Lock <name> on preset load, in the same menu: keeps the processor as it is while presets are loaded.
• Paste only works into a processor of the same kind (e.g. one oscillator to another of the same type).
• The copy survives loading another preset: a way to carry settings from one patch to another.
• Modulations aren't part of it: Copying modulations.

Processor containers

• Several processors sharing one place in the patch (e.g. a choice of filter types): only the one chosen in the box at the top is in use.
• Choosing another one swaps it in with the same connections, without a gap in the sound.
• Each processor keeps its own settings and its own modulations, which come back when it's chosen again.
• Lock <name> on preset load in its right-click menu keeps the whole container while presets are loaded.
• The power button turns the container on or off, whichever processor is chosen. Also in the right-click menu, anywhere in its panel or on its node: Bypass <container> / Activate <container>.
• A choice of the box can be a stack: several processors in use together, side by side (e.g. two oscillators), in the container's place and with its connections. Which containers have one depends on the synth.
• A stack's processors are shown together, one under the other or side by side (depending on the synth), each with its own power button and, if it's a modulator, its own modulator button: turn one off and the others go on. One turned off is silent (its input doesn't pass through, unlike a processor bypassed on its own place); the container turned off passes its input through, as any bypassed processor.
• Each keeps its power button's state while the container is off or another choice is in use.
• Right-click in a stack's processor: Turn off <name> / Turn on <name> for that processor (its own power button), and Bypass / Activate for the container.
• Some containers have two arrows right of the box: the previous and the next processor, without opening the box, going round at the ends. Each is a step to undo.
• Some panels have routing buttons (a strip of short names under the header, e.g. a filter's sources and whether it goes into the next filter or beside it: serial or parallel): each is a connection of the patch, lit while it's there. A click makes or removes it, with no click in the sound, exactly as a cable in the graph does, where it shows as one; it's a step to undo, and it's saved with the patch and its presets.
• Where a source has a button per filter, turning the last of them off doesn't silence it: it then goes round the filters, straight to what comes after them, until one is turned on again.
• A routing button follows the patch whoever changes it: a cable drawn or removed in the graph, an undo, another choice in a container's box (the connections go to the processor chosen). One whose connection would make a loop with the cables already there does nothing, and says so.
• In the graph: one node, with the box in place of the name; the modulation cables shown are the chosen processor's.
• A stack in the graph: still one node and one set of signal cables, with a modulator button for each of its modulators; expanded, each processor's controls under its own name and power button, one under the other. A modulation cable dropped on the node lists every processor's parameters.
• If the chosen processor was the selected modulator, the newly chosen one becomes it (a stack's first modulator).

a container panel with its box open

a container panel showing a stack of three oscillators

Extended processors

• A processor with others built in after it (e.g. an oscillator followed by its own gain): one panel, one node, one name, the controls of all its parts together (e.g. the gain as one more knob after the oscillator's).
• The sound always goes through its parts in the same order; there's no cable between them to change.
• One power button for the whole of it: a part can't be bypassed alone (the right-click menu's Bypass / Activate is the whole processor's on any part).
• Used as a modulator, it gives the signal of its first part, before the others (an oscillator with a gain modulates at full level whatever its gain).
• Every control of every part can be modulated and automated like any other.
• Right-click a part's control: copy/paste that part's values, to or from any processor of the same kind. On the header: the whole processor's.

an extended processor's panel: an oscillator with its gain knob

Modulation #

• The idea: Modulation in the overview.
• Two ways to work: on the sliders, with a selected modulator (this section), or with cables in the graph. Both edit the same modulations, listed in the mod matrix.

Selecting a modulator

• With a processor's modulator button, or in the MODULATOR box.
• While one is selected, every slider it can modulate shows its modulations from it, and a + button to add one.
• Deselect: click the button again, choose NONE in the box, or press Escape.
• The modulator button in the MODULATOR box is lit while one is selected. Click it, or press M, to deselect; again to go back to the last modulator selected (the first one of the patch, if none was yet).

a modulator selected: its button lit, + buttons on the sliders

Adding a modulation

• Select the modulator, click + on the target slider, then drag the amount handle.
• Or drag a modulator button onto a slider: a faint symbol follows the mouse, and a slider that would take the modulation lights up. Dropping it there adds the modulation and selects that modulator. Dropped anywhere else, nothing changes.
• Dragging the modulator button of the MODULATOR box does it for the selected modulator (with none selected, for the one a click on it would select).
• Dropped on a slider it already modulates: only selects the modulator.
• Or in the graph: drag the modulator button onto a node (Modulation cables).
• A matrix holds up to 32 modulations. The one that fills it gives a notification; from then on no slider offers a +, and a modulator dropped on a slider gives the notification again, until a modulation is removed.

Amount, range and live value

• The amount handle: how far the selected modulator moves the parameter. Lights up under the mouse. Drag, double-click to reset, right-click for Reset amount to default and copy/paste.
• The handle is at the knob's right, the modulation's other buttons in a column after it; beside a small knob, too short for that column, they are in two rows of two, the handle still next to the knob.
• The coloured range: the values the modulation sweeps through.
• The small mark: where the parameter's modulations take it right now, in the note played last. The control itself doesn't move. Shown for every modulated parameter, whichever modulator is selected.
• The mark needs audio running; a modulated value never goes past the parameter's range.
• A control that only takes modulations (e.g. an oscillator's PM): drawn as a cable's input socket, not a knob, with nothing to set; it lights up with a cable in it, once something modulates it. Its modulations are added, edited and listed like any other's; without them it does nothing.

close-up of a modulated knob: handle, range, live mark

Direction, on/off, removing

• Direction: one way from the value up to the amount, or both ways around the value.
• On/off: mutes the modulation, keeping its settings.
• Remove: also removes the modulations of its amount.
• Where these buttons are: on the slider while its modulator is selected, and on the modulation's row in the mod matrix.

The curve

• A modulation needn't follow its modulator evenly: its curve bends how the parameter follows it.
• Where: the small display at the right of the modulation's row in the mod matrix, and only there (not on the sliders, not in the graph). Across: what the modulator gives; up: how far the parameter moves.
• Drag it up or down to bend it; Shift for fine steps, the mouse wheel too. Double-click, or right-click and Reset to default: a straight line again.
• Bent upwards: the parameter moves fast at first and eases into the end of its way. Bent downwards: slowly at first, fast at the end. The ends stay where they are: the amount is still how far it goes.
• A two-way modulation (direction) is bent alike on both sides of the parameter's value: the display shows both.
• It bends the modulator's value, before the amount: a modulated amount scales the bent result.
• A straight line costs the least CPU; a bent one a little more.
• Not a host parameter: saved with the patch, undone like any edit, not automated.

a mod matrix row's curve display: straight, bent up, a two-way one bent

Modulating an amount

• A modulation's amount is itself a target: e.g. an envelope fading an LFO's vibrato in.
• How to add one: select the second modulator, + on the amount. Or drag the second modulator's button onto the amount (Adding a modulation).
• Can go several levels deep.

Copying modulations

• Right-click a slider: Copy modulations, Cut modulations, Paste modulations. On an amount handle: the same, for the modulations of that amount.
• What's copied: every modulation of the parameter (modulator, amount, on/off, direction, curve), with the modulations of their amounts.
• Paste adds to what the target already has; a modulator already modulating it takes the copied settings.
• Modulators that can't modulate the target are skipped (What can modulate what); a full matrix gives a notification.

The mod matrix

• Every modulation of the patch, one row each (up to 32). Scroll to see them all.
• A row: modulator on the left, parameter on the right, the amount slider, its curve, and the remove, direction and on/off buttons.
• The GLOBAL MOD MATRIX: the modulations of global parameters (up to 32 more). Only in synths with global processors.

the mod matrix with a few rows, one row labelled

What can modulate what

• A per-voice modulator: per-voice parameters only.
• A global modulator: global parameters, and per-voice parameters too (see Global modulators on the voices).
• A note processor: its controls aren't modulated, and it modulates nothing.
• With a modulator selected, only the sliders it can modulate offer a +.
• A modulation is only worked out if both its modulator and its target are on and in use.
• A macro: any parameter, per-voice or global. Its own knob isn't modulated.

Macros

• What a macro is: a knob that is a modulator. Route it to several parameters and one turn moves them all, each by its modulation's amount. How many a synth has, and where their panels are, depends on the synth.
• Its panel: a header with its modulator button, over its one knob (0 to 100 %). Routed like any modulator: select it and use the + on a slider, drag its button onto a slider, or draw a cable in the graph.
• Or the macro bar (if the synth has one: a strip shown on every page): under the title MACROS, a cell per macro, each with the macro's number, its modulator button and its knob (same right-click menu as in its panel). Hover a cell and the info box names the macro.
• A bar with more macros than it shows (depends on the synth) has an arrow at its right end for the next ones; the arrow is then at the left end, to go back.
• A macro in the bar has no on/off there: it's turned off from its node in the graph, and its cell is then dimmed.
• It reaches both sides of the synth: per-voice parameters and global ones (the effects), with the same knob.
• What it's for: one host automation lane, or one MIDI controller (MIDI learn on the macro's knob), for a change of several things at once. The macro's knob is a host parameter; the parameters it moves keep their own values, saved as they are.
• On a hardware controller: a CLAP host with remote controls (Bitwig Studio's remote control pages) gets the macros' knobs as pages named Macros, eight a page, in the macros' order, with nothing to set up. In other hosts and formats, map each knob with the host's own assignment or with MIDI learn.
• Each of its modulations has its amount, its direction (one-way from the parameter's value, or two-way around it with the macro's middle doing nothing), its on/off and its curve, as any modulation.
• It turns the knobs it's routed to: a parameter stops at the end of its range. A slider shows where the macro has taken it (Amount, range and live value).
• A macro's knob has no +: nothing modulates a macro. A macro can modulate the amount of a modulation, another macro's included.
• It costs no processing, however many parameters it's routed to. Turned off, what it added is taken away.
• Limitation: a move is smoothed over one audio buffer, like a move of the parameter's own knob, not over a fixed time.

a macro's panel, and a knob it's routed to showing where the macro has taken it

The graph #

• The whole patch at a glance: one node per processor, signal cables and modulation cables.
• It edits the same patch as the panels: use whichever is handier.

the graph of a simple patch, with node, signal cable, modulation cable and output labelled

Nodes

• Drag a node to move it.
• Double-click its body (anywhere but a control or its output) to expand it and show its controls; again to collapse. Also Expand / Collapse in its right-click menu.
• An expanded node's controls work exactly like the processor's panel.
• A node has a symbol, its name, a modulator button if it's a modulator, and an output on its right if it's chained and gives sound.
• A node whose processor has a side-chain input has a small arrow for it on its left side.
• A container is one node, with its box in place of the name; showing a stack, it has one symbol and one modulator button per processor.
• An extended processor is one node too, with no cable between its parts: expanded, it shows the controls of all of them.
• Right-click: Bypass / Activate the processor (a container's node: the container), copy/paste the processor's values.

Signal cables

• Sound goes from a node's output (right side) to another node's input (left side), or to the synth's output.
• Connect: drag from a node's output to a node or to the output.
• Move: drag a cable's end to another node. Remove: drag it onto the background, or right-click it → Remove cable.
• A node can feed several nodes and take several cables (they're added together).
• Loops of signal cables aren't allowed.
• What isn't connected to the output, directly or through other nodes, isn't heard (and isn't computed, unless it modulates something).

Side-chain inputs

• A side-chain input is a second input: a sound the processor reads without passing it on. The compressor and the limiter have one beside their own input (the sound that turns theirs down); the envelope follower has nothing else (the sound it follows).
• Only the nodes of processors that have one show it: a small arrow on the node's left side, under where its other cables arrive.
• Connect: drag from a node's output and drop the cable on the arrow (or right around it). Dropped anywhere else on the node, the cable goes into the node's own input. While the cable is dragged, the arrow lights up when it would go into the side chain, the node's outline otherwise.
• A cable into a side chain ends at the arrow. Move it, remove it and right-click it like any signal cable; drop its end on the node's body to make it the node's own input instead, and the other way round.
• The same node can be connected to both inputs of a node. Loops through a side chain aren't allowed either.
• A node whose side chain is all it has (the envelope follower) takes a cable dropped anywhere on it, and has no output: it passes no sound on.
• The arrow is dim while the side chain is turned off (the processor's SIDECHAIN button): the cables stay, and nothing is computed for them.
• What only goes into a side chain isn't heard: it's computed for the processor that reads it.
• A bypassed node passes on what goes into its own input, never its side chain.
• From a per-voice node into a global node's side chain: a send, with a small node of its own on the line.
• Selecting another option of a container keeps the cables into its side chain if the new option has one too, and removes them if it hasn't.

a compressor's node with a cable into its input and one into its side-chain arrow

Modulation cables

• Go from a modulator to the node it modulates.
• Add: drag a modulator button onto a node. Dropped on a slider of an expanded node: that parameter. Dropped anywhere else on the node: choose the parameter from a menu.
• Doing so also selects that modulator.
• Move: drag the cable (or a slider's cable handle, on an expanded node) onto another slider or node. Remove: drop it on the background, or right-click → Remove cable.
• Click the cable (without dragging it): a small panel next to it shows its modulation, with what modulates what, a knob for the amount and the on/off, direction and remove buttons. The amount can be modulated from there too, as in the mod matrix: with a modulator selected, the knob offers a +, and a modulator button can be dragged onto it. Click anywhere outside the panel to close it, except on a modulator button: clicking or dragging one keeps the panel open. The cable stays highlighted meanwhile.
• Moving a modulation removes the modulations of its amount (see Known limitations).
• A greyed cable: a modulation that's in the patch but does nothing right now. It's switched off, its modulator or the processor it modulates is bypassed or not connected to anything, the modulation whose amount it modulates is itself greyed, or the voices reach its global modulator (Global modulators on the voices).

Moving around and the view menu

• Mouse wheel: zoom. Drag the background: move the view.
• A new patch opens zoomed out as far as it takes to show the whole graph: all its sides, every node.
• Right-click the background:
  • Set view as preferred: remembers where every node is, which are expanded, the zoom and the part in view. Saved with the patch.
  • Reset view: back to the preferred view, or to the default one if none was set.
  • Reset view to default: collapses every node, puts them where they first were, and zooms back out to show the whole graph. The preferred view is kept.
• The view is part of the patch: it's saved in presets and with the DAW project.

The two sides

• Only in synths with global processors. A dashed line: per-voice nodes on its left, global nodes on its right. Nodes stay on their side.
• The output is on the right for both: a per-voice node connected straight to it skips the global side.
• Connecting a per-voice node to a global node, and the small nodes on the line: Sending sound to a global processor.
• Drag the line left or right (anywhere on it but a cable or a node) to give the per-voice side more or less room. The global nodes move with it, as they were among themselves. It stops at the per-voice nodes.
• Where the line is is part of the view: saved with the patch, kept by Set view as preferred, put back by the resets.

a graph with both sides, the dashed line and a node on the line

The note side

• Only in synths with note processors. Another dashed line, on the left: the note processors' nodes are left of it, one under the other.
• The cables show the way the notes take: from the top node down, the last one into the line, towards the voices. These cables can't be dragged or removed.
• Drag a note node up or down to change the order (Their order). It doesn't move anywhere else, and nothing can be connected to it or modulate it.
• Under the last note node, NOTES OUT: the note output (Sending the notes on). A cable goes into it from the note processor whose notes leave the synth: the last one at first.
• Drag a note node's output (the square on its right) onto NOTES OUT to connect that one instead: only one is connected at a time. A note node's cable goes nowhere else.
• To send nothing: drag the cable's end onto the background, or right-click it and pick Remove cable. The cable runs behind the nodes: grab it where it shows, next to NOTES OUT. Its symbol is lit only while something is connected.
• NOTES OUT stays under the last node: it can't be moved, and no sound or modulation cable goes into it.
• Expanding, collapsing, the power button: as any node. An expanded one grows to the left and the view moves to show it whole.
• Per-voice nodes stay right of the line.
• Drag the line to the right to give the note side more room (for expanded note nodes), or to the left to give it less: everything on its right moves with it, the other line too. It can go some way left of where it first is (two nodes' width), no further.

the note side of the graph: two note nodes, their cables, the line, and NOTES OUT under them with its cable

Notes before the voices #

Note processors

• Only in synths that have them; which ones depends on the synth.
• What they're for: changing what's played before the synth plays it (moving the notes, turning a key into a chord, a chord into a pattern).
• A panel like any processor's: controls, power button, copying values. Its controls can be automated by the DAW but not modulated.
• Off, it's skipped: the notes go on as they came.
• Turning one on or off, or moving it, while keys are held: the notes sounding are released and the held keys are played again through the new setup. Nothing is left hanging.
• Pedals and the pitch wheel go through them unchanged.

Their order

• The notes go through the note processors one after the other; the order changes the result (transposing before or after an arpeggiator isn't the same).
• Changed in the graph, by dragging a node up or down: The note side.
• Saved with the patch; one step to undo.

Sending the notes on

• What the note processors make of what's played (an arpeggiator's pattern, a chord maker's chords) can also leave the synth, to play other instruments with it.
• In a DAW: through the plugin's note output (Notes out). On its own: to a MIDI device (MIDI output).
• Which notes: the ones coming out of the note processor connected to NOTES OUT in the graph (The note side). At first the last one, so exactly what the synth itself plays; connect an earlier one to send the notes as they are at that point.
• The connection is the processor's: moved to another place of the order, it's still the one connected.
• Turned off, the connected note processor passes on what it's given. With every note processor off, the notes played leave as they came.
• With the notes go the pedals, the pitch wheel, the controllers and aftertouch played.
• The synth keeps playing them itself: sending them on doesn't silence it.
• Nothing is left hanging on the other instrument: connecting another note processor, removing the cable, or changing the note processors under held keys ends the notes sent.
• Saved with the patch and its presets; one step to undo.

The arpeggiator

• Plays the held keys one at a time, in a pattern, instead of together. The first key starts it at once (unless SYNC is on).
• MODE: the order of the notes.
  • UP, DOWN: lowest to highest, highest to lowest.
  • UP DOWN, DOWN UP: one way then back, the lowest and highest played once. UP & DOWN, DOWN & UP: the same with both played twice.
  • AS PLAYED: in the order the keys were pressed.
  • CONVERGE: lowest, highest, second lowest, second highest, and so on inwards. DIVERGE: the same backwards, from the middle outwards.
  • LOW PEDAL: the lowest note before each of the others, which go up (lowest, second, lowest, third...). HIGH PEDAL: the highest note after each of the others.
  • RANDOM: any note, by chance (the same one can come twice in a row). WALK: from the lowest, one note up or down from the last, by chance. SHUFFLE: every note once, in a chance order that changes each time round.
  • CHORD: all the held keys together at every step.
• RATE: a stepped knob, the length of a step as a note length at the tempo (the DAW's, or the synth's own), from 4/1 (four whole notes) to 1/64. A dot after one (1/4.): dotted (half as long again). T: triplet (three in the time of two).
• The note beside it (TEMPO, on at first): turned off, the knob is TIME instead, the length of a step in milliseconds or seconds (10 ms to 4 s), whatever the tempo. Each keeps its own value. A change of either, or of the switch, is taken from the next step on.
• FIG (FIGURE in the DAW's parameter list): what's played at each place MODE goes to.
  • NOTE (at first): one note.
  • GROUP 3 to GROUP 6: that note and the ones after it. Groups of 3 going up: 1-2-3, 2-3-4, 3-4-5, 4-5-6. Groups of 4: 1-2-3-4, 2-3-4-5, 3-4-5-6.
  • 3RDS to 6THS: a broken interval, that note and one further on. Thirds going up: 1-3, 2-4, 3-5, 4-6.
  • MODE says where the figures start: up, down, up and down, by chance, and so on. While it goes down, each figure is played downwards (groups of 3 going down: 6-5-4, 5-4-3, 4-3-2, 3-2-1), so in UP DOWN they turn with it.
  • The numbers are the held keys from the lowest (with OCT at PATTERN, their octaves too), not the notes of a scale: thirds over a held scale are the scale's thirds, over a held chord every other note of the chord.
  • With fewer notes than a figure needs, it's as long as they allow. CHORD has no figures.
• OCTAVES: the pattern repeated one or more octaves higher.
• OCT (OCTAVE MODE in the DAW's parameter list): in which order, with more than one octave.
  • PATTERN (at first): the keys of every octave are one long row that MODE goes over (up and down goes up through all the octaves, then back down).
  • The others play MODE over the held keys alone, and move it to another octave each time it has been through them: UP from the keys' own octave upwards, DOWN from the highest downwards, UP DOWN up then back, RANDOM any of them by chance.
  • With CHORD the octave moves at every step (PATTERN: upwards). With the random modes, after as many steps as there are held keys.
• Two held keys that come to the same note in a step (keys an octave apart, in CHORD over several octaves) are one note. A note that would be above the keyboard's top is played octaves lower.
• GATE: how much of its step each note lasts.
• JITTER (0 to 100 %, none at first): moves each note's start off its step's time, by chance, for a looser feel: at 100 %, up to half a step early or late. The steps themselves stay in time (and on the beat with SYNC), so the pattern doesn't drift. The first note of a pattern isn't moved. A note still lasts GATE of a step from where it starts, or until the next one starts.
• CHANCE (0 to 100 %, all at first): each step's chance of playing, drawn every step. A step that doesn't is a rest: the pattern goes on through it as if it had played (up over three keys, the second left out: 1, rest, 3, 1...). The first step of a pattern always plays, so a key is heard at once.
• RATCHET (1 to 8, 1 at first): each step's notes played that many times, quickly one after the other, evenly up to the next step's notes (with a groove or JITTER, a late step's repeats are closer together), each lasting GATE of its share. A key released during the step isn't in its repeats after that.
• R. CHANCE (RATCHET CHANCE in the DAW's parameter list; 0 to 100 %, all at first): each step's chance of being ratcheted, drawn every step: low for an occasional roll. Nothing with RATCHET at 1.
• GROOVE, at the right: the groove the pattern is played with, a swing for one. The button with its name chooses it (A processor's groove); Straight, at first, is none.
• Under the button, the groove drawn as it's played, once through: its steps side by side, each with a short line where the step is and a taller, lit one where its note starts. Right of its step: late; left of it: early; on it: on time. Straight is four steps on time. Nothing is set on it.
• The knob under it (0 to 100 %, all of it at first): how much of the groove is played. At 100 % each step's note starts as early or late as the groove says, at 50 % half as far, at 0 % on its step. The drawing follows it.
• A groove is counted in steps, so it feels the same at any RATE or TIME: a swing swings eighths at 1/8 and sixteenths at 1/16.
• The groove starts with the pattern: its first step is the first note. With SYNC on and the DAW playing, its first step is the song's first instead, so a swing's late notes are the song's offbeats whenever the keys are pressed.
• Like JITTER, it leaves the steps themselves in time. The two together never move a note more than half a step, so the notes keep their order. The first note of a pattern is never early.
• Another groove chosen while the pattern plays is heard from the next step.
• HOLD: the pattern keeps playing after the keys are released; keys added while others are down join it, a key played once all are up starts a new one; turning it off stops it.
• SYNC: puts the steps on the DAW's beat while it plays (counted from the song's start, at the rate's length). The first key then waits for the next step. With the DAW stopped or on the synth's internal tempo, the pattern starts with the first key as without it. On its own, the beat is a MIDI clock's while one is followed and playing; without one, as with the DAW stopped. After a tempo change, a loop or a jump, the steps go back on the beat. Only with TEMPO on: with it off the switch isn't shown, and steps of a time of their own are on no beat.
• Up to 32 keys at once.

The chord maker

• Plays a chord for each key: the one its chord map has for the key's note. One finger plays a scale's chords.
• CHORD MAP, at the right: the map it plays. The button with its name chooses it (A processor's chord map); EDIT opens the chord map editor on it. Which maps come with the synth, and which one it starts with, depend on the synth.
• KEY: the key (the tonality) the map is played in. A map says its chords for the notes of one octave counted from its first note; KEY says which note that is. With C, a major scale's map gives C major's chords on the white keys; with D, D major's, on the keys of D major.
  • A map of twelve notes to the octave: the box lists the twelve notes by name.
  • A map of another number of notes (made for another tuning): the box lists as many, as keys above middle C (+0, +1...).
• A note with several chords in the map: one is drawn each time, by the chances the map gives them.
• OTHER (OTHER KEYS in the DAW's parameter list): what a key plays whose note has no chord in the map (in a scale's map, the notes outside the scale).
  • THRU (at first): itself alone.
  • MUTE: nothing.
  • NEAREST: the chord of the nearest note that has one (the lower of two as near), as if that key were played. Every key then plays a chord of the map.
  • With no map (None), every key is such a key.
• ROOT, INV 1, INV 2, INV 3: how often a chord is played in its root position (as the map has it) and in each inversion. Chances weighed against each other: 100 and 100 is one chord in two each. All in the root position at first.
  • The first inversion: the chord's lowest note an octave up, so its second lowest is at the bottom. The second: its two lowest up. The third: its three lowest, for chords of four notes or more.
  • An inversion a chord has too few notes for is left out of the draw.
• LEAD (LEADING): voice leading. Each chord's inversion is chosen from the chord played before it, so the chords flow into each other instead of jumping.
  • OFF (at first): the four chances above choose.
  • SMOOTH: the inversion whose notes are nearest the last chord's: every voice moves as little as it can. C then F in C major: C E G, then C F A.
  • COMMON: the one keeping the most notes where they were.
  • TOP: the one whose highest note moves least (a steady melody on top). BASS: the one whose lowest note moves least.
  • STRICT: the one with the fewest parallel fifths and octaves with the last chord (two voices a fifth or an octave apart both moving the same way by as much), then the nearest. Playing up a scale by step, where the chords as the map has them all move in parallel, each chord is turned so they don't. Counted between chords of as many notes.
  • The chord stays within an octave of where the map has it at the key played, so chords follow the keys up and down the keyboard.
  • The first chord has nothing to be led from: it's drawn by the four chances. With leading on they do nothing else.
  • Leading chooses how a chord is voiced, never which chord: a note's chords are still drawn by their chances.
• MEMORY (LEAD MEMORY, 1 to 4): how many of the last chords leading counts the voices' movement from. At 1, the last chord alone. With more, the chords before it count too, each half as much as the one after it: a chord is also kept near where the hand was a few chords ago, so the voicing wanders less.
• NOTES (1 to 8, 8 at first): how many notes of each chord are played, the lowest the map gives it first. At 3, a seventh chord is played as its triad. The bass note isn't counted.
• VOICE (VOICING): how a chord's notes are spaced, after its inversion.
  • CLOSE (at first): next to each other.
  • DROP 2: the second note from the top an octave down. DROP 3: the third from the top. DROP 2+4: the second and the fourth.
  • SPREAD: every second note from the bottom an octave up, for a wide chord.
  • A chord with too few notes for one is left close.
• ROLL (0 to 500 ms, 0 at first): the time between a chord's notes. At 0 they start together; above it one after the other, like a strummed or rolled chord. The first note always starts with the key.
• The note beside it (ROLL TEMPO, off at first): turned on, the knob is RATE (ROLL RATE) instead, the time between the notes as a note length at the tempo, from 1/64 to 4/1, or OFF for none. Each keeps its own value. The roll still starts with its key, on no beat.
• ROLL's box (ROLL MODE): the order they start in. UP, DOWN; ALTERNATE: up for one chord, down for the next; CONVERGE: lowest, highest, second lowest, and so on inwards; DIVERGE: the same from the middle outwards; RANDOM: a chance order for each chord.
• R. JITTER (ROLL JITTER): moves each note of a rolled chord off its time by chance, up to half of ROLL early or late at 100 %, so the notes keep their order.
• R. VEL (ROLL VEL, -100 to 100 %): slopes the notes' velocities along ROLL MODE's order, rolled or not. Below zero each note is softer than the one before (the last down to nothing at -100 %); above zero the first is that much softer and each one louder, up to the key's own velocity.
• V. JITTER (VEL JITTER): each note a little louder or softer than that, by chance (up to half its velocity either way at 100 %).
• OCTAVE (-3 to +3): moves every chord by octaves.
• BASS: adds the chord's root one or two octaves below, whatever inversion the rest is in.
• HOLD: keeps the chords sounding after their keys are released, rolls included. Keys played while another is still down add their chords; once all are up, the next key ends them and starts anew. Turn it off to stop.
• The six small boxes of numbers (R. JITTER to MEMORY) are dragged up and down like knobs.
• A key's chord lasts until the key goes up: changing a control or the map changes the chords played from then on. A key released during a roll leaves the rest of its chord unplayed.
• Two held keys whose chords share a note: the note is struck again for the second key, and ends once both keys are up.
• A note past either end of the keyboard is brought back by octaves.
• The display under the map's name: the chord played last. A stretch of the keyboard with its keys lit (while a key is held; fainter once it's let go of), under the name the map gives the chord. The marks under the keys: where the map's octaves start, at KEY.
• Before an arpeggiator in the order: one held key is arpeggiated as its chord.
• Up to 32 keys at once.

the chord maker's panel, a chord held

Transpose

• Moves every note by octaves and semitones.
• A change applies to the notes played after it: a held note keeps its pitch.
• A note moved off the keyboard's range isn't played.

Tuning #

• In every synth. Where its panels are depends on the synth; some don't show them.
• What it is: the pitch each key sounds at. A tuning system (a scale, and optionally which note each key plays), the key the scale starts on, and the key and frequency everything is tuned from.
• Saved with the patch, so a preset has its own tuning. Every patch starts in 12-EDO (twelve equal steps to the octave).
• Applies to the notes played from then on: a note sounding keeps its pitch.
• While the project has an MTS-ESP master, its tuning is heard instead of the patch's (MTS-ESP).

The patch's tuning

• The tuning's name, at the top. An edited tuning says "(edited)" until it's saved under a name.
• Right-click → Lock tuning on preset load: presets then leave the tuning as it is. A small lock shows in the editor's bottom right corner.
• A =: the concert pitch, as the frequency of the A above middle C in twelve equal steps to the octave; 440 Hz by default, 400 to 480. Moves every key.
• REF KEY: the key every other key is tuned from (69, that A, by default). It sounds at the frequency it has in twelve equal steps at the concert pitch, whatever the scale: 69 at 440 Hz, 60 (middle C) at 261.63 Hz. If it plays no note in the keyboard mapping, the nearest key that does takes its place.
• ROOT KEY: the key that plays the scale's first note (60, middle C, by default). The keys above it play the scale's notes in order, one per key, and so do the ones below (unless there's a keyboard mapping, which starts at this key). Makes no difference in equal tunings without a mapping.
• KEY MAP: the keyboard mapping; empty, one note per key.
• The fields: one per note of the scale (24 in a 24-note tuning), numbered. Each shows how far its note is above the scale's first note; the last one is where the scale starts again (the octave, usually: 1200 cents, 2/1).
• Editing a field: click it, type, Enter (Escape puts it back; clicking elsewhere takes it too). One step to undo.
  • A number is cents (701.955); with a slash, a ratio (3/2).
  • What isn't a pitch is refused with a notification, and the field goes back.
  • The last note can't be at or below the first.
• CENTS / RATIOS: switches what the fields show. Changes nothing in the tuning: a pitch typed as a ratio stays one and reads as its cents meanwhile; cents read as a decimal ratio (1.498307). While ratios are shown, a typed number is a ratio (1.5; 2 is 2/1).
• RESET: back to 12-EDO, one note per key, key 69 at 440 Hz, root key 60, as one step to undo.
• Many notes: the fields scroll.

the tuning editor showing a 19-note tuning in ratios

The keyboard mapping

• What it's for: scales that don't have one note per key. A seven-note scale on the white keys, a scale of 19 or 31 notes of which 12 are played, a five-note scale that still repeats every octave of keys.
• KEY MAP empty (the default): the keys play the scale's notes in order from the root key, one per key.
• Otherwise it's a pattern of keys, starting at the root key and repeated up and down the keyboard. For each key: the number of the note it plays (0 is the scale's first note, 1 the first field, and so on; a number past the last field is in the next period), or x for a key that plays nothing.
  • 0 x 1 x 2 3 x 4 x 5 x 6: a seven-note scale on the white keys; the black keys are silent.
  • From one repetition of the pattern to the next, the pitch rises by the scale's period (its last field).
• Editing it: click, type, Enter (Escape puts it back). One step to undo. What isn't a pattern (or has no key that plays) is refused with a notification. Up to 128 keys.
• A key that plays nothing starts no note at all, whatever sends it (the keyboard, the arpeggiator).
• KBM...: imports a Scala keyboard mapping file (.kbm) onto the patch's tuning, as one step to undo. It sets the pattern, ROOT KEY (the file's middle note), REF KEY and A = (the file's reference note and its frequency).
  • Also KEYS (the file's first and last notes) and REPEAT AT (its "formal octave").
  • A mapping file is made for scales of a certain number of notes: use the scale first, then import its mapping.
  • A reference further than 400 to 480 Hz for A is kept within it, with a notification.
• REPEAT AT: the note of the scale the pattern starts again at; the scale's number of notes unless changed (each repetition one period higher). Only matters with a pattern.
• KEYS ... TO: the first and last keys that play (0 to 127: all). The keys outside play nothing.
• The mapping is part of the tuning: saved with the patch, and with the tuning when it's saved in the browser, which keeps the root key, the reference key and the concert pitch too. So a scale and its imported .kbm, saved, are one tuning file.

the tuning editor with a seven-note scale mapped to the white keys

The tuning browser

• Works like the preset browser: packages, folders, info (name, author, description, the package's version, website and image), right-click menus, copy/cut/paste, drag and drop, NEW PACKAGE, ADD FOLDER, REFRESH.
• First row: 12-EDO, always there, in no package.
• A picture of the selected tuning (or 12-EDO) under its info, before it's used: a line at each of its notes along one period, the twelve-tone scale's notes faint behind them, and how many notes and how long a period under.
• Double-click a tuning (or LOAD, or Enter) to use it: it becomes the patch's tuning, at once, as one step to undo. The rest of the patch is untouched.
• A tuning you saved comes back as it sounded: with its keyboard mapping, and with the ROOT KEY, REF KEY and A = it was saved with. A tuning that is only a scale (an imported Scala file, 12-EDO, most stock tunings) leaves those three as they are.
• SAVE: saves the patch's tuning as it is (edited or not) as a tuning file, with its mapping, root key, reference key and concert pitch, where the selection is; the tuning takes the name it's saved under. Same name there: replaced.
• Tunings built into the synth (its stock package): read-only; use them, or copy them to a package of your own.
• Tunings aren't tied to one synth: your packages show up in every synth of the same maker.
• Files: .swtuning, in the tunings folder of the maker (Show in file manager).

the tuning browser with the stock package open and a tuning selected

Importing and exporting Scala files

• Scala scale files (.scl) are the usual way tunings are shared; thousands exist.
• IMPORT: pick a .scl file; it's kept as a tuning where the selection is (or in the User package), named after the file, with the file's description. A name already taken there gets a number.
• Right-click a package or a folder: Import Scala file..., or Import Scala folder... for every .scl file of a folder at once (into a folder of the same name; files in folders inside it aren't taken). Files that aren't scales are counted in a notification, with the first one's reason.
• Importing doesn't change the patch: double-click the imported tuning to use it.
• Limits: 1 to 128 notes per scale. Keyboard mapping files (.kbm) aren't imported here but in the tuning editor (KBM...), onto the tuning in use.
• Exporting: right-click a tuning, a folder or a package → Export..., choose TUNING FILES (.swtuning, as they are) or SCALA FILES, then pick a folder of your computer. Folders and packages go into a folder of their name; nothing there is replaced.
• As Scala files, each tuning is a .scl (its scale, with its description), plus a .kbm of the same name if it has a keyboard mapping or was saved with its ROOT KEY, REF KEY and A =. A tuning that is only a scale gets no .kbm.

Making your own

• NEW: type a number of notes; the tuning becomes that many equal steps to the octave (19 gives 19-EDO). One step to undo.
• Then edit its notes in the fields, and SAVE it in the browser under a name.
• Group yours in a package: NEW PACKAGE, as for presets (name, author, version, website, description, image).
• A scale needn't repeat at the octave: the last field is the period, whatever it is (3/1, or a single note of 78 cents for equal steps with no octave).

MTS-ESP: tuned by the project

• What MTS-ESP is: a way for one plugin of the DAW project, the master, to tune every MTS-ESP instrument in it at once. The synth is such an instrument (a client): nothing to connect, it finds the master by itself.
• It needs a master in the project (ODDSound's MTS-ESP Master or Mini, or any tuning plugin that can be one), which is also what installs MTS-ESP on the computer. With none, nothing changes.
• MTS-ESP, ON by default: with a master, every key sounds at the pitch the master gives it, in place of the patch's tuning. Where the setting is depends on the synth; one built without MTS-ESP doesn't have it.
• Live: the notes that sound follow the master as it changes, within a few milliseconds, without restarting. So a master that moves its tuning in time retunes held chords.
• A key the master gives no note plays nothing.
• MTS-ESP MASTER: the name of the master's tuning, NONE without a master, and (NOT FOLLOWED) after the name while the setting is off.
• The patch's tuning is kept, not replaced: it can still be edited and saved while a master is followed (without being heard), and it's back the moment the master is removed or the setting turned OFF. The notes sounding then keep the pitch they started at.
• OFF: the patch's tuning whatever the project has. For one instrument that must stay in its own tuning.
• It also works with the application on its own, for a master running in another program.
• It isn't part of the patch: it's kept with the plugin instance in the DAW project, loading a preset never changes it, and it's no step to undo.
• What isn't followed: a master's separate tunings per MIDI channel (the one for all channels is used), and tuning sent as MIDI (MTS SysEx messages).

The MTS-ESP setting on, its master line showing a tuning's name.

Wavetables #

• A wavetable: a set of waveforms a processor plays and morphs between (the wavetable oscillator's POS).
• Wavetables are files, kept in packages and folders like presets and tunings. Which processors play one, and which wavetables come with the synth, depend on the synth.
• There's no page of wavetables: a wavetable is always chosen for one processor, from that processor's panel.
• A wavetable is played with only the harmonics that fit at each note's pitch: a bright one doesn't turn harsh (aliasing) on high notes, where it has fewer harmonics instead.

A processor's wavetable

• A processor that plays a wavetable has a button with the wavetable's name in its panel (also in its expanded graph node).
• The arrows at its ends: the previous and the next wavetable, without opening the browser. Same order as the browser's list (the default one first, then every package), going round at the ends. Each is a step to undo.
• While the wavetable browser is searched (words typed, or the favourites alone), the arrows step through what the search leaves only, without the default one: from a wavetable it leaves out, to the first it leaves. A search that finds nothing doesn't count. Empty the search box in the browser to step through everything again.
• Click the name: a menu of every wavetable, names only, whatever is searched. Its first item is Default (the processor's own); then Favourites, if you have any, which opens a menu of them; then each package, which opens a menu of what's in it: its folders first, each opening its own, then its wavetables. The one in use is ticked, and so are the package and the folders it's in. An empty package or folder is dimmed.
• In the menu: hover or click a package or a folder to open it; Up/Down move, Right or Enter opens, Left goes back, Enter on a wavetable picks it, Escape or a click elsewhere closes the menu. Picking one plays it at once, as one step to undo. One that can't be read: a notification, and nothing changes.
• The button with a folder, after the arrows: that processor's wavetable browser opens over the whole view. Each processor has its own, which remembers what was selected and opened in it; what's searched for is the same in all of them. Files added there are in the menu afterwards.
• The name is followed by (missing) when the patch names a wavetable this computer doesn't have (Wavetables and patches).
• Drag a file from the computer (a file manager, the desktop) onto the processor's panel, anywhere on it: the panel is framed while the file is over it, and dropping it imports the file and plays it at once. No browser opens.
• What can be dropped: a PNG or JPEG picture (Pictures as wavetables), a WAV file (Audio files as wavetables), or a wavetable file (.swwavetable), which is copied as it is.
• Where it's kept: the User package, named after the file (a name taken there gets a number). A wavetable file that's already in one of your packages isn't copied: it's played from where it is.
• Several files at once: all are imported, the first is played.
• Playing it is one step to undo; the imported file stays in the User package.
• A file that can't be read as a wavetable: a notification, and nothing changes.
• In a plugin on Linux, whether files can be dropped depends on the host and on the program they're dragged from. IMPORT in the wavetable browser always works.

an oscillator's panel with its wavetable button over the wavetable display

The wavetable browser

• Works like the preset browser: packages, folders, info (name, author, description, the package's version, website and image), right-click menus, copy/cut/paste, drag and drop, NEW PACKAGE, ADD FOLDER, REFRESH, Export....
• Its title says whose wavetable is being chosen. It opens with the wavetable in use selected, unless what's searched for leaves it out: then nothing is.
• First row: Default, the wavetable the synth gives that processor, in no package.
• A picture of the selected wavetable under its info, before it's used: its waveforms one behind another, as the oscillator draws them.
• Double-click a wavetable (or LOAD, or Enter) to use it: the processor fades to it, at once, as one step to undo, and the browser closes. To try one after another while playing: the arrows of the wavetable button.
• A file that can't be read as a wavetable: a notification, and the browser stays open.
• SAVE: saves a copy of the wavetable the processor plays, under a name, where the selection is.
• Closing it: CLOSE, Escape, or a click outside its panel.
• Wavetables built into the synth: read-only; use them, or copy them to a package of your own.
• Wavetables aren't tied to one synth: your packages show up in every synth of the same maker, and in every processor's browser.
• Files: .swwavetable, in the wavetables folder of the maker (Show in file manager).

a processor's wavetable browser open over the synth, the stock package open

Pictures as wavetables

• IMPORT: pick a PNG or JPEG picture (or an audio file: Audio files as wavetables); it's kept as a wavetable where the selection is (or in the User package), named after the file. A name already taken there gets a number.
• How a picture is read: each row of pixels is a waveform, the top row first; along a row, a pixel's brightness is the waveform's level there (black lowest, white highest). So the picture is the wavetable seen from above.
• Colours count by how bright they look (green more than red, red more than blue); see-through pixels count as black.
• Each waveform is centred, and the whole wavetable brought to full level, so a dim or low-contrast picture isn't quiet.
• Sizes: up to 2048 points per waveform and 256 waveforms; a bigger picture is averaged down. Smooth pictures (gradients) sound mellow, sharp edges and noise sound bright and harsh.
• Refused, with a notification: a file that isn't a picture, and a picture whose rows are each a single colour (silence).
• Right-click a package or a folder: Import image or audio file..., or Import image or audio folder... for every picture and audio file of a folder at once.
• Importing doesn't change the patch: double-click the imported wavetable to use it.

Audio files as wavetables

• IMPORT (and the same right-click items as for pictures): pick a WAV file; it's kept as a wavetable where the selection is, named after the file. For wavetables made for other synths, and single-cycle waveforms.
• How it's read: the file is cut into waveforms of the same length, one after another, the first waveform first.
• The length of a waveform: what the file says, if it does (wavetables saved by some synths do); else 2048 samples.
• A short file (up to 4096 samples) that isn't a multiple of 2048 samples: one waveform, the whole file. So single-cycle files of any length work.
• A longer file that isn't a whole number of waveforms: what's left at the end is dropped. More than 256 waveforms: 256 of them are kept, evenly spread from the first to the last.
• Not a recording turned into a wavetable: nothing finds the pitch or the cycles of a sound. A file that isn't already made of equal waveforms gives whatever its slices are.
• Levels: kept as in the file (unlike a picture's); only a file going over full level is brought down to it.
• Formats: WAV, 8 to 32-bit or float, any sample rate (it doesn't matter: a waveform is one cycle). Stereo: the left channel. No FLAC, AIFF or MP3.
• Refused, with a notification: a compressed WAV, a file too short, a silent one.
• Importing doesn't change the patch: double-click the imported wavetable to use it.

Wavetables and patches

• A patch (a preset, a project) remembers which wavetable each processor plays by its package, folder and name. It doesn't hold the wavetable itself.
• So a preset using one of your wavetables needs that wavetable where it's loaded: Export... the preset's package in the preset browser, which puts the wavetables its presets play in with them, and ADD FOLDER it on the other computer. A single preset exported doesn't take them.
• A wavetable that isn't found: the processor plays its default one and its button says (missing); the patch keeps the name, so it's back once the wavetable is there: add its package in the wavetable browser and close the browser, or load the patch again.
• Renaming or moving a wavetable that patches use makes it missing for them. A package renamed is still found.
• However many processors and instances use a wavetable, it's in memory once.

Samples #

• A sample: a recording a processor plays back (the sampler), one or two channels.
• Samples are files, kept in packages and folders like presets, tunings and wavetables, and chosen the same way as a wavetable: for one processor, from that processor's panel. Which processors play one, and which samples come with the synth, depend on the synth.
• A sample is 30 seconds long at most, and always as loud as it can be without distorting (Audio files as samples).

A processor's sample

• A processor that plays a sample has a button with the sample's name in its panel (also in its expanded graph node): None while it has no sample, and then it plays nothing.
• The arrows at its ends: the previous and the next sample, without opening the browser. Same order as the browser's list (None first, then every package), going round at the ends. Each is a step to undo. While the sample browser is searched, only through what the search leaves, like a wavetable's.
• Click the name: a menu of every sample, names only, like a wavetable's: None, then Favourites if there are any, then each package opening its folders and its samples, the one in use ticked. Picking one is one step to undo.
• The button with a folder, after the arrows: that processor's sample browser opens over the whole view.
• The name is followed by (missing) when the patch names a sample this computer doesn't have (Samples and patches).
• Choosing another sample while notes sound: those notes fade out quickly and end; the next ones play the new sample.
• Drag a WAV file, or a sample file (.swsample), from the computer onto the processor's panel, anywhere on it: the panel is framed while the file is over it, and dropping it imports the file and makes it the processor's sample. No browser opens.
• Where it's kept: the User package, named after the file (a name taken there gets a number). A sample file that's already in one of your packages isn't copied.
• Several files at once: all are imported, the first is played. Playing it is one step to undo; the imported file stays.
• A file that can't be read as a sample: a notification, and nothing changes.
• In a plugin on Linux, whether files can be dropped depends on the host and on the program they're dragged from. IMPORT in the sample browser always works.

a sampler's panel with its sample button

The sample browser

• Works like the preset browser and the wavetable browser: packages, folders, info (name, author, description, the package's version, website and image), right-click menus, copy/cut/paste, drag and drop, NEW PACKAGE, ADD FOLDER, REFRESH.
• Its title says whose sample is being chosen. It opens with the sample in use selected.
• First row: None, no sample.
• A picture of the selected sample under its info, before it's used: its waveform, as the sampler draws it.
• Double-click a sample (or LOAD, or Enter) to use it, as one step to undo; the browser closes. To try one after another while playing: the arrows of the sample button.
• SAVE: saves a copy of the sample the processor plays, under a name, where the selection is.
• Export... (right-click a sample, a folder or a package): asks whether as sample files, to use in another SynthWorld synth, or as WAV files (24-bit, as the sample is kept), for anything else.
• Closing it: CLOSE, Escape, or a click outside its panel.
• Samples built into the synth: read-only; use them, or copy them to a package of your own.
• Samples aren't tied to one synth: your packages show up in every synth of the same maker.
• Files: .swsample, in the samples folder of the maker (Show in file manager). They hold the audio: a few megabytes for a long one.

a processor's sample browser open over the synth

Audio files as samples

• IMPORT: pick a WAV file; it's kept as a sample where the selection is (or in the User package), named after the file. A name already taken there gets a number.
• Right-click a package or a folder: Import audio file..., or Import audio folder... for every WAV file of a folder at once.
• Length: the first 30 seconds are kept; what a longer file has after that is left out.
• Level: what's kept is made as loud as it can be, its loudest point at full level (what was left out doesn't count). Both channels by the same amount, so a stereo recording keeps its balance.
• Channels: mono or stereo; of more channels, the first two. A mono sample plays in both sides.
• Sample rate: kept as recorded (up to 192 kHz) and played at the right speed whatever rate the host runs at.
• Bit depth: kept with 24 bits, whatever the file has (a 32-bit or float file is brought down to 24).
• Formats: WAV, 8 to 32-bit or float. No FLAC, AIFF or MP3. Loop points and root notes saved in a WAV file aren't read.
• Refused, with a notification: a compressed WAV, a file too short, a silent one.
• Importing doesn't change the patch: double-click the imported sample to use it (a file dropped on the panel is used at once).

The sampler

• One of the stock processors: a synth has it or not. Each note plays the processor's sample.
• START and END: where in the sample a note starts and stops, from its beginning (0 %) to its end (100 %). END before START: it plays backwards. Reverse (the arrow pointing left) swaps the two.
• MODE, what a note's end does: NORMAL, the sample plays to END, or once the key is released it fades out over RELEASE while playing on; ONE SHOT, it always plays to END, whatever the key does; LOOP, it plays from START into the loop and stays in it while the key is held, and through RELEASE.
• The loop (LOOP mode): from LOOP START to LOOP END, both kept between START and END. The sample enters it when it reaches LOOP END; what comes after the loop is never played. LOOP END before LOOP START: the loop plays backwards. Loop reverse (two arrows pointing left) swaps the two.
• The loop mode: FORWARD, from the loop's start to its end, then from its start again; BACK AND FORTH, to its end and back, with no jump.
• XFADE (going forward): fades the end of the loop into its start, so the jump back isn't heard. The longer, the smoother and the shorter the loop; half the loop at most.
• FADE IN: how much of the sample, from START, the level takes to rise from silence. FADE OUT: how much of it, before END, the level takes to fall to silence, reached right at END (not in LOOP mode, which never gets there).
• The two fades and XFADE are lengths of the sample, in seconds of it as it was recorded: played an octave up they go by twice as fast. RELEASE is a real time.
• Pitch: with key follow on (the keyboard button), the sample plays at its own pitch on the ROOT note (C3 unless changed) and follows the keys from there, glide and pitch bend included: higher is also faster. Off: every key plays it as recorded.
• COARSE (a ratio: 2 is an octave up), ST (semitones) and CT (cents) tune it, as an oscillator's.
• Played above its own pitch, a sample loses the top of its spectrum that no longer fits, gradually as the pitch rises, instead of turning harsh (aliasing). At its own pitch and below, it's played as recorded.
• Everything but the switches can be modulated: a start that moves with the velocity, a loop that an LFO shifts.
• A note on a voice that's still sounding (every voice in use): what it was playing fades out for a few milliseconds, then the sample starts. With legato, a tied note doesn't start the sample again: only its pitch moves.
• A ONE SHOT note keeps its voice until the sample ends, even with the key up.
• No sample chosen: silence.
• The panel: the sample drawn, between its switches (symbols, no text), under the sample button and over its tuning and RELEASE.
• Left of the sample, MODE's three buttons, one lit: an arrow (NORMAL), an arrow leaving a line (ONE SHOT), two arrows going round (LOOP). Under them, an arrow pointing left: reverse.
• Right of the sample, only in LOOP mode: two arrows the same way (FORWARD), one each way (BACK AND FORTH), and two arrows pointing left (loop reverse).
• On the sample: an upright bar at START and at END, with a flag at its top. Drag a bar sideways to move it.
• In LOOP mode: two more bars, in the accent colour, at the loop's start and end, their flags a row lower and joined by a line. They stay between START and END. Where a loop bar is on START or END (as at first), dragging the bar moves the loop's; drag START or END by its flag, in the top row.
• Along the bottom, small heads: where FADE IN ends, where FADE OUT starts, and in LOOP mode going forward, where the loop's XFADE starts. Drag one away from its end of the sample (or of the loop) for a longer fade. A faint dashed line goes up the sample from each.
• Only what the mode uses is shown: no loop bars outside LOOP mode, no fade out in it, no XFADE going back and forth.
• Double-click a bar or a head: its default.
• The tint behind the sample is its level: even where it plays at full level, fading along a fade, none where nothing is played, where the sample is also greyed. The loop's fade is a lighter tint at each end of the loop.
• A line shows where the note played last is in the sample.
• The button left of the sample's name (three sliders) shows the same settings as values in the sample's place: there each can be modulated and typed in. Click it again for the sample.
• Under the sample: COARSE, ST, CT and ROOT (a note's name: C3, F#2), dragged up and down, and the keyboard button for key follow. Then RELEASE.
• A value its mode doesn't use reads -- and can't be set: ROOT with key follow off, RELEASE in ONE SHOT mode.
• The sample isn't zoomed: in a long one, set a very short fade or loop from the values.

the sampler's panel, in LOOP mode

the sampler's panel showing its values

Samples and patches

• As with wavetables: a patch remembers which sample each processor plays by its package, folder and name, and doesn't hold the audio.
• So a preset using one of your samples needs that sample where it's loaded: Export... the preset's package in the preset browser, which puts the samples its presets play in with them, and ADD FOLDER it on the other computer.
• A sample that isn't found: the processor plays nothing and its button says (missing); the patch keeps the name, so it's back once the sample is there.
• However many processors and instances use a sample, it's in memory once.

Grooves #

• A groove: the feel of a rhythm. For each step, how early or late its note starts: a swing (every second step late), a push, a drag, a player's unevenness.
• It holds timing only, in shares of a step (half a step early to half a step late), and no note length: whatever plays it says how long a step is, so one groove fits every rate.
• It's a row of steps (64 at most) that goes round: after its last step comes its first.
• Grooves are files, kept in packages and folders like presets, tunings, wavetables and samples, and chosen like a wavetable: for one processor, from that processor's panel. Which processors play one (the arpeggiator), and which grooves come with the synth, depend on the synth.

A processor's groove

• A processor that plays a groove has a button with the groove's name in its panel (also in its expanded graph node): Straight while it has none, and then every note starts on its step.
• The arrows at its ends: the previous and the next groove, without opening the browser. Same order as the browser's list (Straight first, then every package), going round at the ends. Each is a step to undo. While the groove browser is searched, only through what the search leaves, like a wavetable's.
• Click the name: a menu of every groove, names only, like a wavetable's: Straight, then Favourites if there are any, then each package opening its folders and its grooves, the one in use ticked. Picking one is one step to undo.
• The button with a folder, after the arrows: that processor's groove browser opens over the whole view.
• The name is followed by (missing) when the patch names a groove this computer doesn't have (Grooves and patches).
• Drag a MIDI file (.mid), or a groove file (.swgroove), from the computer onto the processor's panel, anywhere on it: the panel is framed while the file is over it, and dropping it imports the file and makes it the processor's groove. No browser opens. Kept in the User package, named after the file; several files are all imported and the first is played, as one step to undo.
• A file that can't be read as a groove: a notification, and nothing changes.

the arpeggiator's panel with its groove button

The groove browser

• Works like the preset browser and the wavetable browser: packages, folders, info (name, author, tags, description, the package's version, website and image), search and favourites, right-click menus, copy/cut/paste, drag and drop, NEW PACKAGE, ADD FOLDER, REFRESH.
• Its title says whose groove is being chosen. It opens with the groove in use selected.
• First row: Straight, no groove.
• A picture of the selected groove under its info, before it's used: its steps and how early or late each note starts, as the arpeggiator draws it.
• Double-click a groove (or LOAD, or Enter) to use it, as one step to undo; the browser closes. To try one after another while playing: the arrows of the groove button.
• SAVE: saves a copy of the groove the processor plays, under a name, where the selection is.
• IMPORT: a MIDI file as a groove.
• Export... (right-click a groove, a folder or a package): asks whether as groove files, to use in another SynthWorld synth, or as MIDI files, for anything else.
• Closing it: CLOSE, Escape, or a click outside its panel.
• Grooves built into the synth: read-only; use them, or copy them to a package of your own.
• Grooves aren't tied to one synth: your packages show up in every synth of the same maker.
• Files: .swgroove, in the grooves folder of the maker (Show in file manager). Small text files: a number per step, its shift, from -0.5 (half a step early) to 0.5 (half a step late).

the arpeggiator's groove browser open over the synth

MIDI files as grooves

• How to make a groove of your own: in the DAW, play or draw one note per step, each as early or late as it should be, and export the clip as a MIDI file. IMPORT it, or drop it on the processor.
• Only where the notes start counts: not their keys, lengths or velocities. Every track is read. Notes that start together (a chord) are one.
• The steps are counted from the file's start, and are the longest of quarter notes, eighths, sixteenths and thirty-seconds that give every note a step of its own. So swung eighths are read as eighths, the late ones late. Play the rhythm at least twice over, or a whole bar: with only a couple of notes the step can't be told, and a longer one is taken.
• The groove is as long as its last note's step. A step with no note isn't moved. A rhythm that's the same steps several times over is kept once.
• "Import MIDI file..." and "Import MIDI folder..." in a package's or a folder's right-click menu: one file, or every MIDI file of a folder.
• Exported as MIDI: one note per step, as sixteenth notes, the groove at least twice over so it reads back as itself. A first step that's early is written on time.
• A MIDI file whose times are in seconds (SMPTE), or with no notes, isn't read.

Grooves and patches

• As with wavetables: a patch remembers which groove each processor plays by its package, folder and name, not its timing.
• So a preset using one of your grooves needs that groove where it's loaded: Export... the preset's package in the preset browser, which puts the grooves its presets play in with them, and ADD FOLDER it on the other computer.
• A groove that isn't found: the processor plays straight and its button says (missing); the patch keeps the name, so it's back once the groove is there.

Chord maps #

• A chord map: which chords each note of the octave plays. A scale's triads, its seventh chords, anything.
• It has a number of notes per octave (12, or as many as the tuning it's made for), and for each of them a list of chords: none, one, or up to eight to choose from.
• A chord is its notes counted in keys from the key played: 0 is that key, 0 4 7 a major triad on it. So a map is the same in every key and every octave: what plays it says the key (the chord maker's KEY).
• Each chord has a chance (how often it's the one played when its note has several) and a name (IIm7), which is only shown.
• Chord maps are files, kept in packages and folders like presets, tunings, wavetables, samples and grooves, and chosen like a wavetable: for one processor, from that processor's panel. Which processors play one (the chord maker), and which maps come with the synth, depend on the synth.

A processor's chord map

• A processor that plays a chord map has a button with the map's name in its panel (also in its expanded graph node): None while it has no map.
• The arrows at its ends: the previous and the next map, without opening the browser. Same order as the browser's list (None first, then every package), going round. One step to undo each. The quick way to try a progression in another scale.
• Click the name: a menu of every chord map, names only, like a wavetable's.
• The button with a folder, after the arrows: that processor's chord map browser opens over the whole view.
• The name is followed by (missing) when the patch names a map this computer doesn't have (Chord maps and patches).
• Drag a chord file (.swchord) from the computer onto the processor's panel, anywhere on it: the panel is framed while the file is over it, and the file dropped is kept in the User package and played. One step to undo, which leaves the file.
• A file that isn't a chord map: a notification, and nothing changes.

The chord map browser

• The preset browser's, for chord maps, over the whole view: packages, folders, search, favourites, copy, move, export.
• Its title says whose map is being chosen. It opens with the map in use selected.
• First row: None, no map.
• A picture of the selected map under its info, before it's used: a column per note of the octave, with a cell at each key its first chord sounds above the key played (the key's own row marked).
• Double-click a map (or LOAD, or Enter) to use it, as one step to undo; the browser closes.
• SAVE: saves a copy of the map the processor plays, under a name, where the selection is. Changing a map's chords is the editor's.
• No IMPORT: a chord map isn't made from another kind of file.
• Maps built into the synth: read-only; use them, or copy them to a package of your own.
• Chord maps aren't tied to one synth: your packages show up in every synth of the same maker.
• Files: .swchord, in the chords folder of the maker (Show in file manager). Small text files: the octave's number of notes, and for each chord its note, its name, its notes and its chance.

The chord map editor

• EDIT, in the panel of a processor that plays a chord map: the editor opens over the whole view, on the map that processor plays.
• On the left, the notes of the map's octave, each with the names of its chords. They're counted from the note the map is played in: a twelve-note map's are named as in C (D is the second note of a major scale, whatever KEY is), another's are numbered.
• On the right, the selected note's chords, a row each (eight rows: as many as a note can have):
  • NAME: what the chord is called. Only shown; can be left empty.
  • NOTES: its notes as keys up from the key played, with spaces between them. 0 4 7: a major triad; 0 3 7 10: a minor seventh chord; a number below zero: a note under the key; past the octave is fine (0 7 16). In a map of another number of notes than twelve: that tuning's steps.
  • CHANCE %: how often it's the one played when the note has several, from 0 to 100, weighed against the others'. Empty: 100.
• Enter, or leaving a field, takes what's typed. Tab: the next field. Escape in a field: back to what it was. What can't be read is refused with a notification.
• A row without notes is no chord: empty a chord's notes to remove it.
• Every change is heard at once, on the keys played, while the editor is open. It's only kept once saved.
• The keyboard under the rows: the notes of one chord on keys, from an octave under the key played to two above; the marked keys are the key played and its octaves. Click a key to put that note in the chord or take it out: the same as typing it. It shows the chord of the row last typed or clicked in (its number is over the keys), so click in an empty row, then on keys, to add a chord.
• Right-click a note: copy its chords, paste them over another note's, clear them.
• FILL: gives every note of a scale its chord, in the place of what the map had. It asks for the scale's notes of the octave (0 2 4 5 7 9 11: the major scale), the notes per chord (3: triads; 4: seventh chords) and how many of the scale's notes apart they are (2: thirds; 3: fourths). The chords are named after their place in the scale, and in a twelve-note map after what they are (IIm7). A quick start for a map of your own, in any tuning.
• NEW: asks for the number of notes per octave (the patch's tuning's at first) and starts an empty map.
• SAVE: writes the map over the file it came from. For a file of your own; every processor playing that file, in every patch, plays it as saved.
• SAVE AS: a name, and optionally an author, tags and a description. Saved as another file, next to the one it came from or, if that's built into the synth, in the User package. The processor then plays the new file: one step to undo.
• CLOSE, Escape or a click outside: closes it. With changes not saved, it asks first; discarding them brings back the map the patch names.
• To change a chord's chance in a map built into the synth: edit it, SAVE AS a map of your own.

the chord map editor open on a scale's seventh chords

Chord maps and patches

• As with wavetables: a patch remembers which chord map each processor plays by its package, folder and name, not its chords.
• So a preset using one of your maps needs that map where it's loaded: Export... the preset's package in the preset browser, which takes the maps along.
• A map that isn't found: the processor plays none and its button says (missing); the patch keeps the name, so it's back once the map is there.
• A map saved over changes every patch that names it.

Effects #

• Stock processors that change the sound going through them: a synth has them or not, per voice or on the global side.
• On the global side an effect is there once, for all the notes together, and goes on sounding after the last note has ended. Per voice, each note has its own, and the note's voice lasts as long as the effect has something to give.

The delay

• The sound comes back after a time (an echo), each channel through its own delay.
• TIME L and TIME R: each channel's time, from 1 ms to the synth's longest (2 s unless it chose another).
• LINK (the button with a chain on it, on at first): both channels take the left one's time; the right one's isn't used.
• TEMPO: the times are note lengths at the tempo instead, LENGTH L and LENGTH R, from 1/64 to 4/1, dotted (.) and triplets (T) included. They follow the tempo as it changes.
• OFFSET L and OFFSET R (-50 % to +50 %): make a channel's time that much shorter or longer, whatever sets it. With LINK on, they're what makes the two sides differ: a few percent widen the echoes.
• FEEDBACK: how much of each echo goes back in, so how loud the next one is. 0 %: one echo. 100 %: they never fade (unless a filter is on).
• MIX: from the sound as it came (0 %) to the echoes alone (100 %).
• PING PONG: the echoes go from side to side. Both channels' sound goes in together; it's heard left after the left's time, then right after the right's time more, then left again, each quieter by FEEDBACK.
• Two filters on the way back: LP (low-pass) and HP (high-pass), each with its switch and its cutoff. The first echo is the sound as it was; each later one is darker (LP) or thinner (HP) than the last. Off, a filter takes no CPU.
• FILTER SLOPE (the button at the filters' right, which reads 6 dB, 12 dB or 24 dB: each click goes to the next, and from the last back to the first): how steeply both filters cut beyond their cutoffs, each time round: 6 dB per octave (gentle, as at first), 12 or 24. One setting for the two. At any of them a filter leaves what's on its side of the cutoff as it is, takes 3 dB at the cutoff and has no peak there, so the echoes never grow, not even with FEEDBACK at 100 %. Changing it while the delay sounds fades to the new slope. Not modulated.
• Everything but the switches and the filters' slope can be modulated. A time that moves bends the pitch of the echoes, as a tape's speed does: slowly and slightly, a chorus; a lot, a siren.
• Switching LINK, TEMPO or a note length, or a change of tempo, doesn't click: the time glides to the new one in a moment, with a short bend of the pitch.
• A time longer than the delay can hold (twice its longest TIME: a long note length at a slow tempo) is held at that.
• Turned off and on again, the delay starts empty.
• Its panel: one knob per channel's time, which reads TIME or, with TEMPO on (the note between the two), LENGTH. With LINK on (the chain under the note) the left one is both channels' and the right one is dimmed, with nothing to set. Each side's OFFSET is under its time, PING PONG's switch between them, and FEEDBACK is over MIX.
• The filters are in a row over the display: each one's switch (its shape: falling for LP, rising for HP, lit while on) beside its cutoff, a value dragged up and down like the value bars' (modulated, typed in and reset the same way), which reads -- until its switch is on; then the slope's button.
• Its display: what the delay answers to a click. Time goes left to right, each line is as tall as what's heard then is loud, in decibels.
  • First the click itself, as MIX lets it through; then the echoes, as loud as MIX and FEEDBACK make them.
  • The left channel's lines are in one colour and the right's in another (L and R in the corner, in those colours), side by side when they come together.
  • The glow under them follows the level the whole answer dies away along.
  • Levels are in decibels: the top is the click as it came (0 dB), each line across is 12 dB quieter (-12 dB, -24 dB, -36 dB), and the bottom is -48 dB: what's quieter isn't drawn. Echoes that each lose the same share of the one before step down evenly.
  • While LP or HP is on, a curve in another colour is drawn over the echoes: what the filters let through of each frequency every time an echo goes round, on the same levels. Its frequencies are along the top, from 20 Hz at the left to 20 kHz at the right (100 Hz, 1 kHz, 10 kHz, in the curve's colour), as the times are along the bottom. With both off there's no curve and no frequency.
  • The time shown is a round one with room for a few echoes, marked along the bottom in ms or s. Turning a time moves the echoes along it; now and then it changes to another round time.
  • With TEMPO on the marks are note lengths, as fractions of a whole note (1/4, 3/2), the same picture at any tempo. A time held at the delay's longest isn't shown as held.
  • It follows the knobs as they're turned, automated or modulated (the note played last).
  • The echoes' lines don't show what the filters take from them, which depends on the sound: the first echo is as it came, the second has been through the curve once, and so on.
  • Greyed while the delay is off.
• Per voice with FEEDBACK at 100 % and no filter, a note never ends: its voice is only freed by being taken for another note.

The saturator

• The sound is made louder (DRIVE) and pushed into a curve that bends it, which adds harmonics: from a little warmth to a heavy distortion.
• FUNCTION: the curve.
  • SOFT CLIP: rounded into its limit.
  • HARD CLIP: left as it is up to the limit, then cut flat. The harshest.
  • TUBE: one side squashed gently from the start, the other rounded late. Not the same on both sides, so it adds even harmonics too.
  • TAPE: both sides squashed from the start, over a long way. The gentlest.
  • DIODE: one side squashed early, to half the level; the other clipped. Strongly lopsided.
  • RECTIFY: what's under zero is turned over, which sounds an octave up; clipped at the limit.
• DRIVE (0 to 48 dB): how much louder the sound goes into the curve. At 0 dB a quiet sound comes out as it went in (but through RECTIFY); the same setting drives every function alike.
• OUTPUT (-24 to +12 dB): the level of the saturated sound, to make up for the drive. However hard it's driven, what a function gives is never above the limit, which OUTPUT at 0 dB puts at full level.
• MIX: from the sound as it came (0 %) to the saturated one alone (100 %, as at first).
• Two filters in the saturated sound's way: LP (low-pass) and HP (high-pass), each with its switch and its cutoff, off at first. They filter the saturated sound only: what MIX lets through of the sound as it came isn't filtered. Off, a filter takes no CPU.
• FILTER POSITION (the button that reads PRE or POST: a click moves both filters): where they are.
  • POST (as at first): after the curve, so they shape what it made. LP takes the edge off the harmonics; HP thins the sound out, and takes away what TUBE, DIODE and RECTIFY put off-centre.
  • PRE: before the curve, so they choose what's saturated. HP keeps the bass from being what gets clipped; LP gives the curve less to make harmonics of, but leaves the ones it makes.
  • The path: DRIVE, then (PRE) the filters, the curve, then (POST) the filters, OUTPUT, MIX.
  • Moving them while a sound goes through fades from one place to the other, with no click.
• FILTER SLOPE (the button that reads 6 dB, 12 dB or 24 dB: each click goes to the next, and from the last back to the first): how steeply both filters cut beyond their cutoffs, as the delay's: one setting for the two, no peak at the cutoff, a fade when changed.
• DRIVE, OUTPUT, MIX and the filters' cutoffs can be modulated; FUNCTION, the filters' switches, their slope and their position are chosen. Changing FUNCTION while a sound goes through fades to the new curve, with no click.
• Per voice, each note is saturated alone: a chord stays a chord. On the global side the notes go through the curve together, and distort each other, as through a guitar amplifier.
• It's made so that little of what it adds folds back as tones that don't belong to the note (aliasing), which is plainest on high notes driven hard. Oversampling takes off more.
• The saturated sound loses a little of its top end, less with oversampling on.
• TUBE, DIODE and RECTIFY aren't the same on both sides, so what they give isn't centred around zero while a sound goes through.
• Its panel: FUNCTION's box, over its filters, over its display, over its three knobs. Where the panel is narrow: the box has no name before it, and MIX goes under DRIVE.
• The two arrows at the right of FUNCTION's box choose the function before the one chosen and the next one; past the last comes the first. Each click is one step to undo.
• The filters are in a row over the display, as the delay's: each one's switch (its shape, lit while on) beside its cutoff, a value dragged up and down like the value bars', which reads -- until its switch is on; then the slope's button and the position's, neither with a label: its text is what's chosen. In a narrow panel the two buttons are one over the other.
• Its display: the curve the sound is bent by. What goes in is left to right, with silence in the middle and a sound at full level reaching both ends; what comes out is upwards, around the middle line.
  • The faint straight line is a sound coming out as it went in: where the curve leaves it, the sound is bent. The flatter the curve, the more it's squashed there.
  • DRIVE pushes the sound further along the function (the curve gets steeper in the middle and flat sooner), OUTPUT makes it taller or lower, and MIX brings it back towards the straight line.
  • TUBE, DIODE and RECTIFY are seen lopsided: the two halves aren't each other's mirror.
  • The levels marked at the right are full level (0 dB, the top and the bottom) and, once OUTPUT makes the curve go over it, +6 dB and +12 dB: the picture then shows more, so the curve gets smaller.
  • It follows the controls as they're turned, automated or modulated (the note played last). A modulation doesn't change the levels shown: a curve it takes over them is drawn along the edge.
  • While LP or HP is on, a second curve in another colour is drawn over it: what the filters let through of each frequency. It has axes of its own, in its colour: the frequencies along the top, from 20 Hz at the left to 20 kHz at the right (100 Hz, 1 kHz, 10 kHz), and its levels at the left, in decibels: the top of the picture is all of it (0 dB), then -12 dB, -24 dB, -36 dB, and the bottom is -48 dB. With both off there's no such curve and no such marks.
  • The filters' curve is the same in PRE and in POST: the button says where they are. The saturator's own curve doesn't change with the filters.
  • It doesn't show how loud the sound going through is, so how far along the curve it reaches.
  • Greyed while the saturator is off.

The bitcrusher

• The sound as an old, cheap converter would give it: taken only now and then and held in between, and rounded to a few levels. Gritty, ringing, noisy.
• RATE (250 Hz to 48 kHz): how many times a second the sound is taken, each one held until the next. What's above half of it folds back under it as frequencies that weren't there, which move against the notes played. At the top nothing is held.
• BITS (1 to 16): how many levels the sound is rounded to; each bit less halves them. At 16 nothing of it is heard; around 8 quiet sounds get grainy; at 1 a sound at full level is only up, down or silent. It needn't be a whole number.
• A sound quieter than half a step between two levels is silenced: with few bits, a note's quiet end is cut off.
• DITHER (0 %, none, as at first, to 100 %): noise added to the sound before it's rounded, up to a step either way.
  • The rounding then hisses rather than distorts: the grain that followed the sound becomes a steady noise under it.
  • A sound quieter than half a step comes through in that noise instead of being silenced: a note's quiet end fades into hiss.
  • The fewer the bits, the louder the noise. Silence gets none: with nothing going in, nothing comes out.
  • Each note, and each side of the stereo, has noise of its own.
• MIX: from the sound as it came (0 %) to the crushed one alone (100 %, as at first).
• All four can be modulated.
• Per voice, each note is crushed on its own. On the global side the notes are crushed together: with few bits, quiet notes come and go with the louder ones.
• Off, it takes no CPU.
• Its panel: its display over its four knobs (BITS, RATE, DITHER, MIX): in a row, or where the panel is narrow, in two rows of two.
• Its display: what it makes of a sound. The faint curve is one cycle of a sine at full level going in, over 4 ms (the time in the corner); the stair is what comes out of it.
  • A step is as long as the sound is held: RATE makes the steps shorter or longer. At its lowest the whole picture is one step; at its highest the steps are too short to see.
  • A step is on one of the levels BITS rounds to: fewer bits, taller steps. While the levels are few (up to about 4 bits) each is a faint line.
  • With DITHER some steps are on the level next to theirs. The picture holds still: it shows which levels the noise reaches, not the noise as it's heard, which is never twice the same.
  • With MIX under 100 % the steps lean along the sine they're blended with.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • It doesn't show the sound going through: the sine is the display's own.
  • Greyed while the bitcrusher is off.

The chorus

• The sound comes back a few milliseconds late, several times over (its voices), each at a time that sweeps slowly back and forth, so each wavers in pitch a little, and never as the others do. Mixed with the sound as it came, it's several players of the same part: thicker, wider, shimmering.
• DELAY (0.5 to 30 ms): the shortest time a voice comes after the sound. A few milliseconds thicken it; a very short one hollows it out as the times sweep.
• DEPTH (0 to 20 ms): how far over DELAY each voice's time sweeps. The further, the more its pitch wavers. At 0 the voices stand still, all at DELAY.
• RATE (0.01 to 20 Hz): how fast the times sweep. Slow is a gentle drift, fast a vibrato's waver; the faster, the more a given DEPTH bends the pitch.
• VOICES (1 to 4): how many delayed copies each channel has, evenly apart in the sweep: while one's time gets longer another's gets shorter. More of them are a thicker, steadier chorus, as loud as one.
• WIDTH: how far the right channel's voices are in the sweep from the left's. At 0 % both channels get the same chorus; at 100 % (as at first) the right's voices are halfway between the left's, the widest.
• MIX: from the sound as it came (0 %) to the voices alone (100 %), which with one voice is a vibrato. Half of each, as at first, is the usual chorus.
• RATE, DEPTH, DELAY, WIDTH and MIX can be modulated; VOICES is chosen, the same for every note. Changing it while a sound goes through fades to the new voices, with no click.
• Per voice, each note has its own chorus, whose sweep starts with the note. On the global side the notes share one, whose sweep runs on.
• Off, it takes no CPU and no memory. Turned on again, it starts empty.
• Its panel: its display, over a value bar (DELAY, WIDTH, VOICES: values dragged up and down), over its three knobs (RATE, DEPTH, MIX).
• Its display: how long after the sound each voice comes, over one cycle of the sweep. Time goes left to right through the cycle; a curve is as high as its voice is late, from no time at the bottom.
  • A curve per voice: the left channel's in one colour, the right's in another (L and R, in the bottom left corner). Where both channels sweep alike (WIDTH at 0 %), the left's cover the right's.
  • The upright line is where the sweep is now, with a mark on each voice. A voice's pitch is bent down while its curve climbs under its mark, and up while it falls.
  • DELAY lifts the curves, DEPTH makes them taller, VOICES adds curves evenly apart and WIDTH slides the right channel's away from the left's.
  • The times marked at the right are in milliseconds, up to a round one that holds DELAY and DEPTH together: the picture shows more as they're turned up, so the curves get smaller.
  • Along the bottom, the cycle's length and half of it, from RATE.
  • It follows the controls as they're turned, automated or modulated (the note played last). A modulation doesn't change the times marked: a curve it takes over them is drawn along the edge.
  • It doesn't show MIX, nor the sound going through.
  • Greyed, with no line and no marks, while the chorus is off.

The flanger

• The sound comes back a very short time late, a time that sweeps slowly back and forth, and is mixed with the sound as it came. The two cancel out at evenly spaced frequencies (a comb's teeth), which slide up and down with the sweep: a jet's whoosh.
• DELAY (0.1 to 10 ms): the shortest time the sound comes back after. The shorter, the higher and the further apart the frequencies that cancel out: 1 ms puts them 1000 Hz apart.
• DEPTH (0 to 10 ms): how far over DELAY the time sweeps: how wide a stretch the teeth slide over. At 0 the teeth stand still.
• RATE (0.01 to 20 Hz): how fast the time sweeps. Slow is a long whoosh, fast a wobble.
• FEEDBACK (-95 to 95 %): how much of what comes back goes in again. At 0 % the teeth are smooth; further from it they sharpen into a ringing, metallic pitch that follows the sweep, and the sound gets louder. Under 0 % it goes in turned over: the ring is an octave lower and hollower.
• WIDTH: how far ahead of the left channel's sweep the right's is. At 0 % both sweep together; at 100 % one is at its shortest time while the other is at its longest. Half of that at first.
• MIX: from the sound as it came (0 %) to what comes back alone (100 %), which with no feedback is a vibrato. Half of each, as at first, cancels the most.
• All six can be modulated.
• Per voice, each note has its own flanger, whose sweep starts with the note. On the global side the notes share one, whose sweep runs on.
• A note lasts until its flanger has stopped ringing: longer with more FEEDBACK.
• Off, it takes no CPU and no memory. Turned on again, it starts empty.
• Its panel: its display, over a value bar (DELAY, DEPTH, WIDTH: values dragged up and down), over its three knobs (RATE, FEEDBACK, MIX).
• Its display: how much of each frequency it lets through right now. Frequencies go left to right, low to high (20 Hz to 20 kHz, marked at 100 Hz, 1 kHz and 10 kHz); a curve is as high as that frequency is loud.
  • It moves by itself: the teeth slide up as the sweep shortens the time and down as it lengthens it (the sweep of the note played last). With no note sounding it stands still.
  • The left channel's is in one colour, the right's in another (L and R, in the bottom left corner), the left's drawn over the right's.
  • The levels are in decibels, from -24 dB at the bottom to 24 dB at the top, marked every 12; the stronger line in the middle, 0 dB, is the sound as it came. They don't change: a dip or a peak beyond them is drawn along the edge.
  • To the right, and with long times, the teeth are too close together to draw one by one: they're a band, from their dips to their peaks.
  • DELAY and DEPTH say between where the teeth slide, FEEDBACK sharpens them and lifts their peaks, MIX deepens them and WIDTH takes the right channel's away from the left's.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • It doesn't show RATE (but by how fast it moves), nor the sound going through.
  • Greyed while the flanger is off.

The phaser

• The sound goes through stages that leave every frequency as loud as it came but turn each by a different angle, and is mixed with the sound as it came. Where the two are opposite they cancel out: notches, which slide up and down as the stages sweep. A swirl, softer than the flanger's: a few notches, not a whole comb.
• FREQ (20 Hz to 20 kHz): the middle of the sweep. The notches lie evenly about it, low and high.
• DEPTH (0 to 4 octaves): how far the notches sweep each way from FREQ. At 0 they stand still.
• RATE (0.01 to 20 Hz): how fast they sweep.
• NOTCHES (1 to 6): how many frequencies cancel out. More are a thicker, more vocal sweep, and take a little more CPU. The same for every note, and not modulated; changing it while the phaser sounds fades to the new ones.
• FEEDBACK (-95 to 95 %): how much of what the stages give goes in again. At 0 % the notches are smooth dips; further from it what's between them sharpens into ringing peaks that follow the sweep, and the sound gets louder. Under 0 % it goes in turned over: the peaks are where the notches were.
• WIDTH: how far ahead of the left channel's sweep the right's is. At 0 % both sweep together; at 100 % one is at its lowest while the other is at its highest. Half of that at first.
• MIX: from the sound as it came (0 %) to what the stages give alone (100 %), which with no feedback sounds all but the same. With no feedback, half of each gives the deepest notches; with some, a little more than half.
• All but NOTCHES can be modulated.
• Per voice, each note has its own phaser, whose sweep starts with the note, at its lowest. On the global side the notes share one, whose sweep runs on.
• A note lasts until its phaser has stopped ringing: longer with more FEEDBACK and a low FREQ.
• Off, it takes no CPU. Turned on again, it starts empty.
• Its panel: its display, over a value bar (FREQ, DEPTH, WIDTH, NOTCHES: values dragged up and down), over its three knobs (RATE, FEEDBACK, MIX).
• Its display: how much of each frequency it lets through right now, drawn like the flanger's: frequencies left to right (20 Hz to 20 kHz), levels from -24 dB to 24 dB, which don't change, the left channel's curve over the right's.
  • It moves by itself: the notches slide up and down with the sweep (the sweep of the note played last). With no note sounding it stands still.
  • The dips are the notches, as many as NOTCHES on each channel; with FEEDBACK, the peaks are between them (under 0 %, where the notches would be).
  • FREQ slides them all, DEPTH says how far they sweep, FEEDBACK sharpens the peaks, MIX deepens the notches and WIDTH takes the right channel's away from the left's.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • It doesn't show RATE (but by how fast it moves), nor the sound going through.
  • Greyed while the phaser is off.

The reverb

• The sound comes back as from a room's walls: a few reflections, then more and more of them, until they're a wash that dies away.
• SIZE (25 to 200 %): the room's size, which is how far apart its reflections are. Small, they come at once and close together, and with a long DECAY can ring like a pipe; large, the reverb takes longer to build up. It doesn't change how long the reverb lasts. The same for every note, and not modulated; changing it while the reverb sounds fades to the new one.
• DECAY (0.1 to 30 s): how long the reverb lasts: the time it takes to fall by 60 dB, whatever SIZE.
• DAMP (200 Hz to 20 kHz): the frequency above which the reverb dies away sooner, as in a room whose walls take the highs: far above it, in a quarter of DECAY. Low, the reverb darkens as it fades; at the top it stays bright to the end.
• MIX: from the sound as it came (0 %) to the reverb alone (100 %).
• PRE DELAY (0 to 250 ms): how long the reverb waits after the sound. A little keeps the sound clear of it. Modulating it bends the pitch of what goes in.
• MOD: how much the room's reflections drift, slowly, each its own way. A little keeps any frequency from ringing on and makes the reverb smoother; a lot wavers in pitch like a chorus. It takes a little of the highs away, and a little more CPU. At 0 % the room stands still.
• WIDTH: how far apart the reverb's two channels are. At 100 % each has its own reflections; at 0 % both get the same reverb, in the middle.
• Its filters, over the display: a low-pass (LP) and a high-pass (HP) on what goes into the reverb, each with its switch and its cutoff, and the button that steps through their slope (6, 12, 24 dB per octave), as the delay's. They keep the high or the low end of the sound out of the whole reverb; the sound as it came isn't filtered. A cutoff is blocked while its filter is off.
• All but SIZE and the filters' switches and slope can be modulated.
• Nothing grows, even at the longest DECAY.
• Per voice, each note has its own reverb (and its own memory: a lot of it with many voices). On the global side the notes share one.
• A note lasts until its reverb can no longer be heard.
• Off, it takes no CPU and no memory. Turned on again, it starts empty.
• Its panel: its filters, over its display, over a value bar (PRE DELAY, MOD, WIDTH: values dragged up and down), over its four knobs (SIZE, DECAY, DAMP, MIX).
• Its display: what it answers to a click, drawn like the delay's: time left to right, marked in ms or s, and level upwards in decibels, the top being the click as it came and each line across 12 dB quieter.
  • First the click itself, as MIX lets it through.
  • Then a line where each of the room's first reflections comes, the left channel's in one colour and the right's in another: PRE DELAY moves them all, SIZE spreads them.
  • The glow is the level the reverb dies away along, which DECAY makes longer: all of it for the reverb's body, and the denser part, under the fainter line, for its highs, which are gone sooner the lower DAMP is.
  • The curve over them, while LP or HP is on, is what the filters let into the reverb of each frequency, the frequencies along the top (20 Hz to 20 kHz), on the same levels. What they keep out also lowers where that part of the glow starts.
  • Its time axis is a round time, as long as the reverb takes to reach the bottom of the display; it follows DECAY's own value, not its modulations.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • These are the levels the reverb falls along, not each reflection's own: a reverb is thousands of them, each far quieter than the click.
  • It doesn't show MOD, WIDTH, nor the sound going through.
  • Greyed while the reverb is off.

The compressor

• It turns the sound down while it's loud, so its loud and quiet parts end up closer together; brought back up, the quiet parts are louder than they came. Evens out a part, adds sustain, glues several notes into one sound, or pumps on purpose.
• THRESHOLD (-60 to 0 dB, -18 at first): the level over which the sound is turned down, in decibels under full level. Lower, more of the sound is over it.
• RATIO (1:1 to 20:1, 4:1 at first): how many decibels over the threshold go in for one that comes out. At 1:1 nothing is done; 2:1 to 4:1 evens a sound out; at 20:1 it hardly gets over the threshold (a limiter).
• KNEE (0 to 24 dB, 6 at first): the stretch of levels around the threshold over which the ratio sets in. At 0 it's a corner at the threshold; wider, the sound starts being turned down a little before it, and the compressor is heard less.
• ATTACK (0.1 ms to 1 s, 10 ms at first): how fast the sound is turned down once it's over the threshold. Short, it catches every hit; longer, the start of each hit gets through first, which keeps its punch.
• RELEASE (1 ms to 5 s, 150 ms at first): how fast the sound comes back up once it's quieter. Short, the level follows the sound closely, and can waver with a low note's own cycles or pump; longer, it's smooth, and a quiet part after a loud one stays down for a while.
• Both times are how long it takes to cover most of the way (63 %), in decibels.
• MAKEUP (0 to 36 dB, none at first): how much louder the compressed sound is made, to bring back the level the compression took. It doesn't change the dry sound of MIX.
• MIX: from the sound as it came (0 %) to the compressed one alone (100 %, as at first). In between (parallel compression), a hard-compressed sound under the dry one fills in its quiet parts and leaves its hits as they were.
• A sound under the threshold (and under the knee) is left exactly as it came.
• It listens to the louder of the two sides and turns both down alike, so the sound stays where it is between them.
• SIDECHAIN (the button after the knobs, lit while on; off at first): whose level it follows.
  • Off: the sound going through it.
  • On: the sound connected to its side-chain input in the graph. The sound going through is turned down while that other one is loud (a bass ducked under a kick, a pad under a voice), and that other one isn't heard here.
  • On with nothing connected there, nothing is turned down.
  • Switching it moves the level over ATTACK or RELEASE, with no click. Off, nothing is computed for what's connected to the side chain.
  • Part of the patch, not a parameter the DAW automates, and not modulated.
• The seven knobs and values can be modulated.
• Per voice, each note is compressed on its own. On the global side the notes are compressed together: a loud note turns the quiet ones down with it.
• Off, it takes no CPU.
• Its panel: its display, over ATTACK, RELEASE and KNEE as values in a bar (drag one up or down, or double-click to type), over its four knobs and the SIDECHAIN button (THRESHOLD, RATIO, MAKEUP, MIX).
• Its display: what it does to a sound's level. How loud a steady sound goes in, left to right, against how loud it comes out, upwards, both from -60 dB to full level, with a line every 12 dB.

  The compressor's display: the curve leaving the straight line at the threshold, and the reduction meter at its right

  • The faint straight line is a sound coming out as it went in. The curve leaves it at THRESHOLD, marked by an upright line.
  • KNEE rounds the corner there; RATIO sets how flat the curve goes on.
  • MAKEUP lifts the whole curve (over full level it's drawn along the top); MIX brings it back towards the straight line.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • The meter at its right: how much the sound is being turned down right now, in decibels, from none at the top down to 24 (marked at 6, 12 and 18). Watch it to set THRESHOLD, and ATTACK and RELEASE, which are how fast it moves down and back up. Per voice, it's the note played last.
  • It doesn't show how loud the sound going through is, nor MAKEUP and MIX in the meter.
  • Greyed while the compressor is off, the meter empty.

The gate

• It lets the sound through while it's loud and turns it down while it's quiet, which leaves out what's under a sound between its notes: a noise, a hum, the end of a reverb. Set fast, it cuts notes short on purpose.
• THRESHOLD (-72 to 0 dB, -40 at first): the level the sound must reach for the gate to open, in decibels under full level. Set it just over what should be left out: higher, the quiet ends of the notes are cut with it.
• HYSTERESIS (0 to 24 dB, 6 at first): how far under the threshold the sound must fall for an open gate to close. At 0 it opens and closes at the same level, and a sound hovering around it makes it chatter; with more, a note that opened the gate is let through until it has faded well under the threshold.
• RANGE (0 to 96 dB, 96 at first): how much the sound is turned down while the gate is closed. At the top it's as good as silence; less, what's under the threshold is only made quieter, which is heard less as a gate. At 0 nothing is done.
• ATTACK (0.1 ms to 1 s, 1 ms at first): how fast the gate opens once the sound reaches the threshold. Short, the start of each note gets through whole; longer, it fades in.
• HOLD (0 to 2 s, 50 ms at first): how long the gate stays open once the sound is too quiet to keep it so, before it starts closing.
  • It's also what keeps the gate open through a low note's own cycles, which go through silence twice each: too short (under half a cycle of the lowest note, 25 ms for 20 Hz), the gate wavers with them.
  • A sound that comes back within it starts it again.
• RELEASE (1 ms to 5 s, 100 ms at first): how fast the gate closes once the hold is over. Short, the sound is cut; longer, it fades out.
• ATTACK and RELEASE are how long it takes to cover most of the way (63 %).
• While it's open the sound is exactly as it came.
• It starts closed: the very start of the first sound through it is its attack.
• It listens to the louder of the two sides and turns both down alike, so the sound stays where it is between them.
• It follows the sound going through it, not another one.
• All six can be modulated.
• Per voice, each note opens its own gate. On the global side the notes open it together: one loud note lets the quiet ones through with it.
• Off, it takes no CPU.
• Its panel: its display, over ATTACK, HOLD and RELEASE as values in a bar (drag one up or down, or double-click to type), over its three knobs (THRESHOLD, HYSTERESIS, RANGE).
• Its display: what it does to a sound's level. How loud a steady sound goes in, left to right, against how loud it comes out, upwards, both from -72 dB to full level, with a line every 12 dB.

  The gate's display: the curve jumping onto the straight line at the threshold, the way back beside it, and the reduction meter at its right

  • The faint straight line is a sound coming out as it went in.
  • The curve is a sound getting louder: RANGE under the straight line (along the bottom when that's under what the display shows) up to THRESHOLD, marked by an upright line, where the gate opens and the curve jumps onto the straight line.
  • The thin line is the way back: a sound getting quieter is let through down to the fainter upright line, HYSTERESIS under the threshold, where the gate closes.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • The meter at its right: how much the sound is being turned down right now, in decibels, from none at the top, the gate open, down to 96 (marked at 24, 48 and 72). Full while the gate is closed, so with no sound too. Watch it to set THRESHOLD, and ATTACK, HOLD and RELEASE, which are how fast it empties, how long it waits and how fast it fills. Per voice, it's the note played last.
  • It doesn't show how loud the sound going through is.
  • Greyed while the gate is off, the meter empty.

The limiter

• Nothing of the sound ever comes out louder than its ceiling: what would is turned down just enough, and just before, so a peak is rounded rather than cut. With GAIN it makes a sound as loud as it can be without clipping.
• GAIN (0 to 36 dB, 0 at first): how much louder the sound is made going in. At 0 only what's already over the ceiling is turned down; with more, more of the sound reaches the ceiling and is held there: louder and denser, and with a lot of it, squashed.
• CEILING (-36 to 0 dB, 0 at first): the level nothing comes out over, in decibels under full level.
• RELEASE (1 ms to 2 s, 100 ms at first): how fast the sound comes back up once a peak is past (the time to cover most of the way, 63 %).
  • Short, the sound is as loud as can be, and a low note's own cycles can make the level waver, which distorts.
  • Longer, it's smoother, and a quiet part after a loud one stays down for a while.
  • Every peak is held for 10 ms from when it goes in before the release starts, which keeps most notes' cycles from being followed one by one.
• LOOKAHEAD (a box of times: 0, 0.5, 1, 2, 3, 5 or 10 ms, 3 at first): how long before a peak the sound starts being turned down.
  • The longer, the more gently, and the less it's heard.
  • The sound comes out that much later, which is latency: the DAW is told and compensates, so the synth stays in time with the other tracks. Played live, it answers that much later.
  • At 0 there's no wait (the sound is one sample late): each peak is turned down the moment it comes, which is heard as a clipped edge.
  • A DAW takes another latency by stopping and starting the synth, which cuts what's sounding: choosing another time does that, and so does turning the limiter on or off, or connecting it. So it's a choice to make once, not a knob.
  • The same for every note; not modulated and not automated. Part of the patch.
  • With several limiters one after another their times add up; side by side, the longest counts, and the sound that goes round a limiter is ahead of the one that goes through it.
• A sound that stays under the ceiling comes out exactly as it came, later by LOOKAHEAD.
• It listens to the louder of the two sides and turns both down alike, so the sound stays where it is between them.
• SIDECHAIN (the button after the knobs, lit while on; off at first): whose level it follows.
  • Off: the sound going through it.
  • On: the sound connected to its side-chain input in the graph. The sound going through is turned down by as much as that other one, GAIN added, is over CEILING, and that other one isn't heard here.
  • The ceiling is then the other sound's: the sound going through can come out over it.
  • On with nothing connected there, nothing is turned down.
  • Switching it fades from one to the other. Off, nothing is computed for what's connected to the side chain.
  • Part of the patch, not a parameter the DAW automates, and not modulated.
• GAIN, CEILING and RELEASE can be modulated. Moving CEILING turns the sound up and down with it, with nothing cut.
• Per voice, each note is limited on its own. On the global side the notes are limited together, which is where it keeps a chord under full level.
• Off, it takes no CPU, and the sound isn't late.
• Its panel: its display over its three knobs (GAIN, CEILING, RELEASE), the LOOKAHEAD box and the SIDECHAIN button.
• Its display: what it does to a sound's level. How loud a steady sound goes in, left to right, against how loud it comes out, upwards, both from -60 dB to full level, with a line every 12 dB.

  The limiter's display: the curve over the straight line up to the ceiling and flat from there, and the reduction meter at its right

  • The faint straight line is a sound coming out as it went in.
  • The curve is GAIN over it, up to where that reaches CEILING, marked by an upright line, and flat from there on.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • The meter at its right: how much the sound is being turned down right now, in decibels, from none at the top down to 24 (marked at 6, 12 and 18). The more often and the further it moves, the harder the sound is being limited. Watch it to set GAIN and RELEASE. Per voice, it's the note played last.
  • It doesn't show how loud the sound going through is, nor RELEASE and LOOKAHEAD.
  • Greyed while the limiter is off, the meter empty.

The parametric equaliser

• Four bands, each changing the sound around a frequency, then an output level: what shapes a sound's tone, or takes out a part of it that's in the way.
• Each band has a SHAPE, chosen in the box over its knob:
  • LOW CUT: what's under its frequency is taken out, 12 dB per octave.
  • LOW SHELF: everything under its frequency is turned up or down. The first band's at first.
  • BELL: a peak or a dip around its frequency. The second and third bands' at first.
  • HIGH SHELF: everything over its frequency. The fourth band's at first.
  • HIGH CUT: what's over its frequency is taken out, 12 dB per octave.
  • Any band can be any shape. Two cuts at one frequency are 24 dB per octave.
  • The same for every note, and not modulated. Changing it while a sound goes through fades to the new one.
• And three values:
  • FREQ (20 Hz to 20 kHz; 100 Hz, 400 Hz, 2 kHz and 8 kHz at first): where the band is. A bell's gain is reached there; a shelf is half way there, and all the way well past it; a cut is 3 dB down there. Any band can be anywhere: they needn't stay in order.
  • GAIN (-18 to 18 dB, 0 at first), the band's knob: how much louder, or under 0 quieter, a bell or a shelf makes the sound. At 0 the band does nothing. A cut has none: its knob is there, dimmed, with nothing to set.
  • Q (0.1 to 10, 0.71 at first): how narrow a bell is, from several octaves wide to a sharp peak or notch. For a shelf, how sharp its step: over 0.71 the sound just before the step is pushed the other way, a bump on one side and a dip on the other. For a cut, how much it rings at its frequency: 0.71 is the flattest, more is a peak there.
• OUTPUT (-24 to 24 dB, 0 at first): the level of the whole sound after the bands. What makes up for the level the boosts added or the cuts took.
• The bands come one after another, so what they do adds up (in decibels) where they overlap.
• With every bell and shelf at 0 dB, no cut and OUTPUT at 0, the sound is exactly as it came. A bell or a shelf at 0 dB takes no CPU.
• All but the shapes can be modulated.
• Per voice, each note is shaped on its own (and a modulation can give each its own curve). On the global side the notes are shaped together.
• Off, it takes no CPU.
• Its panel: its display, over the bands, a column each: its FREQ and under it its Q as values in two bars (drag one up or down, or double-click to type), over its GAIN knob under its shape's box. OUTPUT's knob is at their right.
• Its display: what the equaliser does to each frequency, from 20 Hz at the left to 20 kHz at the right.

  The equaliser's display: the curve over and under the line of the sound as it came, a dot for each band, one of them lit under the mouse

  • The line across the middle is the sound as it came. Over it, that frequency comes out louder; under it, quieter. The display goes 24 dB each way; what adds up to more is drawn along its edge.
  • Lines mark 100 Hz, 1 kHz and 10 kHz.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • A dot marks each band, at its frequency and its gain (a cut's on the middle line). It's the band's handle, lit while the mouse is on it:
    • Drag it sideways to move the band's frequency and, for a bell or a shelf, up and down for its gain. One step to undo.
    • The mouse wheel over it is its Q: up, narrower.
    • A double-click puts its gain back to 0 dB. On a cut's, its Q back to 0.71.
    • The dots are where the bands are set, not on the curve: the curve is all the bands together, and the output's level.
    • Two dots at the same place: the mouse has the later band's. Move it away to reach the other.
  • It doesn't show the sound going through.
  • Greyed while the equaliser is off.

The ring modulator

• The sound multiplied by a sine of the effect's own. Every frequency of the sound comes out as two, one that far under it and one that far over it, and is itself gone: the sound's harmonics are no longer in tune with each other. Bells, metal, a robot's voice.
• FREQ (1 Hz to 20 kHz; 261.6 Hz, middle C, as at first): the sine's frequency.
  • Under some 20 Hz the sound's level goes up and down with it: a tremolo.
  • Over that, the sound is moved down and up by that many hertz: the higher, the further from the notes played.
  • The sine itself is never heard: with nothing going in, nothing comes out.
• KEY (the piano's keys; off as at first): key follow. On, the sine moves with the note played, an octave for each octave, glide and bend included, so what comes out keeps its place among the note's harmonics along the keyboard.
  • FREQ is then the sine's frequency at middle C: left at 261.6 Hz, the sine is at the note played, and the note comes out an octave up.
  • On the global side it follows the note the global notes give it.
• WIDTH (0 %, as at first, to 100 %): how far the right channel's sine is ahead of the left's, up to half a cycle. In between the two channels differ, which is wide; with a slow sine the sound goes from side to side.
• MIX: from the sound as it came (0 %) to the sound times the sine alone (100 %, as at first).
• FREQ, WIDTH and MIX can be modulated.
• Per voice, each note has its own sine, which starts with the note. On the global side one sine, which runs on, multiplies the notes together.
• Off, it takes no CPU.
• Its panel: its display over FREQ, WIDTH, MIX and KEY: in a row, or where the panel is narrow, in two rows of two.
• Its display: what it makes of a tone. Frequencies go left to right, 20 Hz to 20 kHz; a line is as tall as that frequency is loud, in decibels (full level at the top, a line every 12 dB).
  • The thin faint line is a tone going in: the A at 440 Hz, at full level.
  • The lines with a dot are what comes out of it: one FREQ under it and one FREQ over it, each at half its level. They move apart as FREQ goes up; the lower one turns back up once FREQ passes the tone.
  • With MIX under 100 % the tone itself comes out too, and the other two get quieter.
  • The coloured upright line is FREQ: the sine, which isn't in the sound.
  • A frequency beyond either end of the display is drawn at its edge.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • It doesn't show the sound going through (the tone is the display's own), the key follow, nor WIDTH.
  • Greyed while the ring modulator is off.

The stereo width

• A stereo width and mid-side utility. A stereo sound is its mid, what both channels have alike (heard in the centre), and its side, what they differ by (heard at the sides): it sets how much of each is left.
• WIDTH (0 % to 200 %; 100 % at first): how much of the side is left.
  • 0 %: both channels the same (mono).
  • 100 %: the sound as it came.
  • Over it: wider than it came, up to twice its side.
  • A sound that's the same in both channels has no side: nothing widens it.
• MID and SIDE (silence to +6 dB; 0 dB at first, the knob's middle): the level of each part.
  • MID all the way down: the side alone, each channel the other turned over.
  • SIDE all the way down: mono. The side comes out at WIDTH and SIDE together.
• MONO (the switch; off at first): mono bass. On, the two channels are made the same under BELOW's frequency, whatever the rest is.
  • BELOW (20 Hz to 1 kHz; 120 Hz at first): that frequency. What's lower loses its side, 24 dB of it an octave down; what's higher keeps its width.
  • Its knob reads -- while MONO is off: nothing reads it.
  • Switching it fades: no click.
  • The mid isn't touched: the bass is as loud as it was.
• WIDTH, MID, SIDE and BELOW can be modulated.
• With everything as at first, the sound goes through unchanged.
• Off, it takes no CPU.
• Its panel: its display over WIDTH, MID, SIDE, MONO's switch and BELOW, in a row.
• Its display: how wide the sound comes out at each frequency. Frequencies go left to right, 20 Hz to 20 kHz; the band is as tall as the sound is wide there, towards the left channel (L) at the top and the right (R) at the bottom.

  The stereo width's display: the band closed to the middle line under MONO's frequency and open over it

  • The middle line is mono.
  • The faint line each side of it, marked 100 %, is the sound as wide as it came; the display's edges are twice that, and what's wider is drawn along them.
  • WIDTH and SIDE open the band. So does turning MID down: what's left is then more at the sides.
  • With MONO on, the band closes under BELOW's frequency, which the coloured upright line marks.
  • It follows the controls as they're turned, automated or modulated (the note played last).
  • It doesn't show the sound going through (how wide it is to begin with), nor how loud MID and SIDE leave it: only how they compare.
  • Greyed while the stereo width is off.

Sending sound to a global processor

• In the graph, drag a per-voice node's output to a global node: the global node gets that node's sound from every voice, added together.
• Nothing goes back from the global side to the per-voice side.
• Each global node fed that way gets a small node on the dashed line, where its per-voice cables arrive. Drag it up and down the line to tidy the cables.
• A global node's side-chain input is fed the same way (drop the cable on its arrow), and gets a small node of its own on the line, with one arrow where the other has two: a global node fed both ways has two of them.
• Drop a per-voice node's cable on that small node to connect it too.
• The one cable on its right stands for everything it receives. Drag it: to another global node (sends everything there, joining what that node already gets), to the same node's other input (its side-chain arrow, or its body), to the output (skips the global side), onto the background (disconnects it all).

Global notes

• Global processors exist once, so they have no note of their own. GLOBAL NOTES chooses which notes they follow: what starts and releases a global envelope, or restarts a global LFO.
• ALL NOTES: every note played starts them, every note released releases them.
• HELD NOTES: they start with the first note held and are released with the last one, as one long note.
• FIRST VOICE: only the notes of the first voice.

Global modulators on the voices

• A global modulator can modulate per-voice parameters: every note then gets the very same movement (one LFO in step across a chord), instead of one per note.
• The condition: nothing of the voices may reach that modulator (no sound sent into it or into what feeds or modulates it). Otherwise its modulations of the voices stay in the patch but do nothing, and their cables are greyed in the graph.
• So a global envelope follower with a per-voice node sent into it modulates global parameters only: the effects after the voices, not the voices themselves.
• A sound from outside isn't the voices': a global envelope follower fed by the audio input alone modulates the voices too (a filter on every note opened by a drum loop).
• The other way round isn't possible: a per-voice modulator never modulates a global parameter.

Presets #

• A preset is a saved patch: every parameter, connection and modulation.
• A synth chooses where its preset browser is, or whether it has one.

The browser

• Left: the list. Init (the synth's default patch), then every package with its folders and presets.
• Right: the info of what's selected (a preset's name, author, tags, description; a package's author, version, website, description, image).
• Packages and folders start closed, except the ones holding what's selected. Click the arrow before a package or folder, or double-click it, to show or hide what's in it. Left/Right do the same on the selected one.
• The status line: what was just done.
• The buttons: LOAD, SAVE, NEW PACKAGE, ADD FOLDER, REFRESH.
• A package's website is a link: click to open it in the web browser.
• Over the list, the search box and a star: finding presets.

the preset browser with a preset selected, parts labelled

Searching and favourites

• Type in the search box over the list: only the presets with every word typed are listed, under the packages and folders they're in. A word is looked for in a preset's name, its author, its tags and the names of the folders it's in. Its description isn't searched.
• Capitals and accents don't matter: eclat finds Éclat, and the other way round. Greek and Cyrillic letters are found whatever their case.
• While the list is searched, packages and folders are all shown open, with no arrow: nothing is collapsed. Empty the box and the list is as it was.
• Enter in the box selects the first preset found and moves to the list (Enter again loads it). Escape empties the box.
• Favourites: the star beside LOAD, or right-click a preset → Add to favourites / Remove from favourites. A favourite has a star at the end of its row. Presets built into the synth can be favourites too.
• The star beside the search box lists the favourites alone; with something typed, the favourites that match.
• Favourites are yours, not the patch's or the preset's: every instance of the synth has them (after REFRESH, or the next time its window opens), and a preset given to someone else doesn't carry its star. A favourite stays one when it's renamed or moved, in the browser or outside the synth (in a file manager: found again at the next REFRESH). Not if the file was also changed outside the synth, or copied by a program that doesn't keep the time it was written.
• A CLAP host with its own preset browser is told which presets are favourites, as of when it last looked for presets.
• A preset can also be given a MIDI program number, to load it from a controller: right-click it → MIDI program... (Program changes).
• The default preset, the one the synth starts with in place of Init: right-click a preset → Set as default. One preset at a time: setting another takes its place. Remove as default, in the menu of the one that is, goes back to starting with Init. Presets built into the synth can be the default too.
• The default preset is yours, like the favourites: every new instance of the synth starts with it, in any project, and it follows its preset when that's renamed or moved. The selected preset's details say Default preset. A DAW project still opens with the patch it was saved with, and the application with its last run's.
• Whatever is saved, imported, pasted or moved is always shown: if the search would hide it, the search is emptied.
• The same box and stars are in every browser: tunings, wavetables and samples.
• Wavetables have one search for the whole synth, and samples another: what's typed in one processor's browser is in every other processor's, and the arrows of the wavetable and sample buttons step through what it leaves.

Loading

• Double-click a preset, or select it and press Enter or LOAD, or right-click → Load.
• It replaces the whole current patch, but for what's locked; undo brings the previous one back.
• Init loads the default patch: the one the synth was made with, whatever the default preset is.
• A preset made for an older or newer version of the synth still loads: what it doesn't mention keeps its default, what the synth no longer has is ignored.
• A file that isn't a readable patch: a notification, and the current patch is left untouched.
• From the host's own preset browser (CLAP hosts that have one, such as Bitwig Studio): the same presets, with their author, description and tags, a package shown as a sound pack. Loading one there is the same load, one step to undo.
• The host lists what was there when it last looked for presets: a preset saved or a folder added since may need the host to scan again.
• The two browsers follow each other: a preset loaded in the host's is selected in the synth's, with Loaded on its status line, and the host is told which preset the synth's loaded. Not after Init, an undo or a project loaded: those are no preset's.

Locks: keeping parts of the patch

A locked processor's panel, the small lock in its bottom right corner, with the right-click menu showing the unlock item.

The lock button with its menu of the synth's parts, two of them ticked.

• For going through presets while keeping a part of the sound: the effects, the arpeggiator, the macros, the tuning.
• Right-click anywhere in a processor's panel or on its graph node: Lock <name> on preset load; Unlock <name> on preset load while it's locked.
• A container is locked as a whole (the item names it): the processor it has chosen, its power button, and every processor in it.
• Whole parts of the synth: the lock button (where the synth places it). A click lists them under Keep on preset load, ticked while locked; pick one to lock or unlock it.
  • Note processors, Per-voice processors, Global processors: every processor of that part, with its modulations and its cables. Only the parts the synth has.
  • Tuning: the scale, the keyboard mapping, A =, REF KEY and ROOT KEY. Also Lock tuning on preset load, by right-clicking in the tuning editor.
  • Settings: the synth-wide settings that are part of a patch (voices, unison, voice stealing, legato, glide, oversampling, internal tempo, master gain).
  • Graph view: where the graph's nodes are, its zoom and what it shows, not its cables. Also Lock graph view on preset load, by right-clicking the graph's background.
  • Unlock everything: the parts and every processor locked on its own.
• The lock button is lit while anything is locked, a single processor included.
• A processor of a locked part shows the lock too, and its own item reads Locked with the per-voice processors (or its part's), greyed: its own lock is kept for when the part is unlocked.
• A small lock shows what's locked: in the bottom right corner of a panel, of the tuning editor or of the graph, and in the top right corner of a graph node.
• What a locked processor keeps when a preset is loaded:
  • its controls, whether it's on, the file it plays (wavetable, sample, groove, chord map) and its curve;
  • the modulations of its controls and the ones it's the modulator of (a locked macro keeps what it moves), with the modulations of their amounts;
  • its cables, at both ends: what's sent to a locked effect still is, even if the preset connects that processor elsewhere too;
  • a note processor: the order of all the note processors and what's connected to the note output.
• The preset gives everything else. Its modulations and cables that involve a locked processor are left out.
• With the modulation matrix full, a locked processor's modulations that find no free row are left out.
• Locks also hold for Init, a preset file dropped on the synth and a preset loaded from the host's browser.
• They don't hold for undo and redo, A and B or a DAW project being opened: those bring back a whole patch.
• Not part of the patch: no preset locks or unlocks anything, and locking isn't a step to undo. Saved with the DAW project (in the application: kept for next time).

Saving

• SAVE, or right-click a package or folder → Save preset here...: a name, and optionally an author, tags and a description.
• Tags: words to find the preset by (bass, deep), typed with commas between them. Spaces around one, an empty one and one repeated are dropped. They're shown over the preset's description, changed with Edit info..., and given to a host's own preset browser, which makes its categories of them. The search box finds presets by them.
• Where it goes: into the selected package or folder, or next to the selected preset. With nothing selected, or something built into the synth, into the User package.
• Saving under the name of a preset already there replaces it.
• Two presets can share a name if they're in different folders.

Packages and folders

• A package is a folder on disk holding presets, with a name and optionally an author, a version, a website, a description and an image.
• The User package: always there, the default place for your presets.
• NEW PACKAGE (or right-click → New package...): makes one.
• ADD FOLDER (or Add package from disk...): adds a folder of your computer to the list, e.g. a package you downloaded. Seen by every instance of the synth.
• Right-click a package: Edit package info..., Set image... (PNG or JPEG) / Remove image, Show in file manager, Remove package... (only from the synth's list, or from disk too).
• Folders inside a package (up to 8 deep): New folder..., Rename folder..., Delete folder... (with its presets).
• REFRESH: reads the folders again, for presets saved from another instance or files copied in by hand.
• Order: folders first, then presets, both by name. Start names with numbers to set an order.
• Sharing presets: right-click a preset, a folder or a package → Export..., then pick a folder of your computer: a copy goes there (a preset as its .swpreset file; a folder or a package as a folder of its name, a package with its info and image, ready for ADD FOLDER on another computer). Presets built into the synth can be exported too. Nothing in the destination is replaced: a name already taken gets a number.
• A package exported whole takes along the wavetables, samples, grooves and chord maps its presets play (copies, in the same folder, each where it was in its own package; the ones built into the synth aren't needed). The exported presets play those copies, so the folder is all someone else needs. The status line says how many went with it; a notification, which ones weren't found.
• A package can hold presets, tunings, wavetables, samples, grooves and chord maps together: one folder, one name. ADD FOLDER in any browser adds it to every browser it has files for, so a downloaded package's presets find their wavetables and samples. Each browser shows only its own kind of files, and removing or deleting the package in one leaves the others' files.
• A preset that plays wavetables or samples this computer doesn't have: a notification on loading it names them and the package each is from (Wavetables and patches).
• A package is also just its folder (Show in file manager). Where the User package is on each system.
• Removing never deletes anything but this synth's own presets and what's left empty.

Dropping preset files on the synth

• Drag a preset file (.swpreset) from the computer (a file manager, the desktop) onto the synth's window, anywhere: the whole window is framed while the file is over it, and dropping it imports the preset and loads it. The preset browser needn't be showing.
• Where it's kept: the User package, under the preset's name (a name taken there gets a number, so nothing is replaced). A preset file that's already in one of your packages isn't copied: it's loaded from where it is.
• Several files at once: all are imported, the first is loaded.
• Over the panel of a processor that takes files itself (a wavetable, a sample) only that panel is framed, but a preset file dropped there is loaded as a preset all the same.
• Loading it is a load like any other: one step to undo. The imported file stays in the User package.
• A folder can be dropped too. A package (a folder with its package.json, as Export... makes one) is added to the list where it is, like ADD FOLDER (Packages and folders), in every browser it has files for. Any other folder is copied into the User package as a folder of its name, with the folders inside it that hold presets.
• Nothing of a dropped folder is loaded: it's selected in the preset browser, and the status line says what was done. Dropped together with preset files, the first file is still loaded.
• What's refused, with a notification: a file that isn't a preset, a preset of another synth, a folder with no preset in it. The notification is still there once a preset dropped with them has loaded.
• In a plugin on Linux, whether files can be dropped depends on the host and on the program they're dragged from. ADD FOLDER in the preset browser always works.

Moving, copying, deleting

• Drag a preset or a folder onto a package or a folder to move it there.
• Right-click a preset: Edit info..., Copy, Cut, Export..., Show in file manager, Delete preset....
• Then right-click a package, a folder or a preset → Paste: the copied or cut preset goes there (next to the preset, if it was one). A cut preset is pasted once.
• A copy landing where its name is taken gets a name of its own; a move there is refused.
• Deleting asks first.

Presets built into the synth

• Packages that come inside the plugin: always there, nothing to install, updated with the synth.
• They can't be changed, renamed or deleted.
• To change one: load it and save it in a package of yours, or copy it and paste it there, or drag it there (it's copied, not moved).

Forms and the file picker

• Forms (save, new package, edit info, confirmations): see Text fields for the keys. A refused form stays open and says why.
• The file picker (a folder to add or to export into, a package's image): double-click a folder to go into it, or the first row to go up. The path shown is what will be chosen; when choosing a folder with none selected, it's the folder you're in.
• NEW FOLDER (when choosing a folder, e.g. where to Export...): type its name under the list, then Enter or CREATE; Escape gives up on the name. It's made in the folder you're in and selected, ready to be chosen. A name already taken, or one that can't be a folder's, is told in place of the path.
• Right-click a folder in the picker → Rename folder... (when choosing a folder): its name goes in the same field, to change; Enter or RENAME. A name something else has there is refused: nothing is replaced. Renaming a folder that is one of the synth's packages takes it off the list, like renaming it outside the synth: add it again with ADD FOLDER.
• It's the synth's own picker, not the system's.

the save form, and the file picker

Settings #

• Settings of the whole synth. Where each one is depends on the synth: all on one page, or each next to what it's about.
• The master volume, voices, unison and its four settings, voice stealing, legato, glide, oversampling (while playing and while bouncing) and the internal tempo are part of the patch (saved in presets); the view size, the skin, the keyboard shortcuts, which tempo is followed, the DC filter, the safety clip and the MIDI setup (MIDI channel, MIDI mappings) aren't.
• Changing the voices or the oversampling silences the synth for an instant.
• Run on its own, the synth has more of them: its audio output, its MIDI inputs, the MIDI clock it follows and its MIDI output.

the settings

Master volume

• A knob with no label and no value beside it: how loud the whole synth comes out, from silence to +6 dB. At 0 dB the sound is left as it is.
• Where it is depends on the synth; usually always in view, whatever the page shown (in a corner, next to the tabs).
• Its value is shown in decibels in the value popup while it's dragged, and typed in there (right-click → Enter value). Double-click: back to 0 dB.
• It comes after everything else in the patch: no processor can be put after it, and it isn't in the graph.
• It can't be modulated; the DAW can automate it (see Automation).
• A change is reached smoothly, without a click.
• Saved with the patch, so each preset has its own.

Voices

• NUM VOICES: how many notes can sound at once.
• More voices use more CPU only while they're playing.
• 1 voice: a monophonic synth (see Legato and Glide).
• A note keeps its voice until its sound has ended (its release finished), not just until the key is up.

Unison

• UNISON (1 to 8): how many voices play each note. 1: no unison.
• Each is a whole voice, with its own oscillators, filters, envelopes and modulations: the classic polysynth unison. Four settings say how they differ; the notes sounding follow them smoothly, the DAW can automate them and a MIDI controller can be mapped to them. None does anything at 1.
• UNISON DETUNE (0 to 100 cents): how far the outermost voices are from the note's pitch, one below and one above, the others in between. Everything that follows the note follows it: every oscillator, a filter's key follow.
• UNISON WIDTH: how far they're spread between left and right. 0: all in the centre; 100 %: the outermost fully to a side. The lowest is on the left.
• UNISON BLEND: their levels. Above 0 the central voices are louder, below 0 the outer ones (the most detuned and the widest); the overall level stays the same. Needs 3 voices or more.
• UNISON BIAS: how the voices are laid out between the middle and the outermost. 0: evenly apart; above 0 gathered towards the middle, below 0 towards the outermost. Their detune and their side follow. Needs 4 voices or more.
• The display over them shows a note's voices: each is an upright line, where it is between left (L) and right (R) and as tall as its level. A line's colour is its detune: the further the voice is from the note's pitch, above or below, the more of the accent colour it takes. With 1 voice: one line in the centre.
• For anything else (a cutoff, a wavetable's position, one oscillator's pitch alone) there are the unison spread and unison distance modulators (Playing as modulation).
• The number of notes that can sound at once (Voices) stays the same; voice stealing, legato and glide treat a note's unison voices as one.
• A note uses about as many times the CPU as it has unison voices (up to four cost little more than one in most synths).
• The level is evened out: a note is about as loud whatever the unison, once its voices are detuned. Voices that aren't told apart at all add up instead, and are louder.
• Where each unison voice's oscillators start their cycle is the oscillators' own setting (RANDOM, The stock processors): anywhere in the cycle, or, at 0, all together, a loud peak at each note's start.
• Not the same as an oscillator's own unison (The stock processors), which copies only the oscillator and is much cheaper; the two can be used together.
• Changing UNISON cuts the notes sounding.

Voice stealing

• Which voice a new note takes over when every voice is busy.
• Always first: a voice playing the same note, then voices already released, then those held only by a pedal.
• Among them: OLDEST (the note that started first), NEWEST (the one that started last, so earlier notes keep sounding), QUIETEST (lowest level), LOWEST / HIGHEST (pitch).
• Or the note closest to the new one, counted in keys: CLOSEST ABOVE (the closest above it; the closest below if there's none above), CLOSEST BELOW (the other way round), CLOSEST, HIGHER / CLOSEST, LOWER (the closest either side; of two as close, the higher or the lower).
• A stolen voice carries on from where it was, to avoid a click.
• A note stolen while its key is down isn't lost: when a voice stops being held (a key or a pedal let go) and that key is still down, the voice goes back to its note, the one stolen last first. With 1 voice: play a note, hold it, play another, let it go, and the first one is back.

Legato

• LEGATO: what a note does when it takes over a voice whose note is still held (key down, or a pedal holding it).
• ON: the voice moves to the new pitch and nothing starts again (envelopes carry on, LFOs keep their cycle). OFF: the note starts like any other (envelopes attack again, from where they were).
• With 1 voice: notes played overlapping are joined, notes played apart start anew. With more voices it only happens once every voice is held.
• A note taking over a voice that was already released is never joined to it.
• Going back to a key still held (see Voice stealing) follows the same setting.
• What "starts again" means depends on the synth's processors.

Output meters

• They show the synth's output as it leaves it: after the master volume, the DC filter and the safety clip. A synth may have some or none of them, wherever its layout puts them; the meter is usually next to the master volume, with the button of the output panel beside it.
• The meter: a bar per channel, the left one first (on top when the bars lie, at the left when they stand), from -48 dB to +6 dB.
• The line across the bars is 0 dBFS, full level. A bar is in the skin's accent colour, and turns whole to its warning colour when the output goes past full level: there are no green, yellow and red zones, the colours are the skin's.
• A bar goes up at once and falls slowly (30 dB a second). The mark ahead of it is the loudest it got: it stays a second, then falls too. The bar keeps the warning colour for as long as its mark is past 0 dBFS, so a short peak is seen.
• Beyond 0 dBFS: with the safety clip on, the sound is being bent (up to +1.3 dB); with it off, it's as loud as shown, and whatever comes after the synth has to cope. Turn the master volume down.
• The scope: the last 20 milliseconds of the output as a wave, both channels together. The lines at its top and bottom are full level; the wave is held there if it goes past.
• A steady note stands still on the scope (it's drawn from where the wave rises through the middle); notes below about 33 Hz, chords and noise drift or jump.
• When the DAW stops running the synth, the meter falls to nothing and the scope goes flat.
• None takes any CPU from the sound while it isn't on screen (the window closed, on a page that isn't shown, the output panel closed).
• Hovering one shows what it is in the info box; there's nothing to click or set.

The meter next to the master volume, its bars in the warning colour past 0 dBFS; the scope with a steady wave.

Output panel

• The output's meters together, in a panel that floats over the synth: a spectrum, the scope under it, and a level meter with its scale at the right. They show the same output as the meters above. In its top right corner, the CPU meter.
• Opening and closing: the button drawn as three bars, the middle one the tallest, lit while the panel is open (where the synth places it, usually next to the meter). The cross in the panel's corner closes it too.
• No key until you give it one: Output meters on / off in the keyboard shortcuts.
• Moving it: drag it from anywhere on it, its meters included. It stays whole inside the window.
• The synth is used as always while it shows: turn a knob, change page, play, and watch the output change. Only what's under the panel can't be reached: move it aside.
• It stays open, where you left it, from one page to another and when a preset is loaded; closed and opened again, it's back in the same place.
• The spectrum: how loud the output is at each frequency, both channels together. Left to right, 20 Hz to 20 kHz, each octave as wide (lines at 100, 1k and 10k); bottom to top, -78 dB to +6 dB (a line every 24 dB down from 0). A sine at full level reaches 0 dB.
• It rises at once and falls slowly (60 dB a second), so a short sound can be read.
• The level meter: the bars standing, against a scale in dBFS (+6, 0, -12, -24, -36, -48). The number over them is the loudest either channel got in the last moments (-- for nothing), in the warning colour while that's past 0.
• Limitations: in the lowest octaves the spectrum is coarse (a low note's peak is wide, and two notes close together there aren't told apart); it lags a sound by about a tenth of a second; with the DAW at a sample rate over 48 kHz, what the output has above 24 kHz shows folded back among the lower frequencies.

The output panel over the synth: the spectrum of a chord, its wave, the level meter with its scale.

CPU meter

• Where: in the top right corner of the output panel, CPU, a bar and a percentage. A synth may have another somewhere in its layout.
• What it measures: how much of the time it has the synth takes to make its sound. 100% is all of it: beyond, the sound arrives late and breaks up (crackles, dropouts).
• The bar and the number: the average of the last tenth of a second. The mark ahead of the bar: the slowest moment of the last two seconds, which is what a crackle comes from; it can be well ahead of the bar when notes start.
• The bar and the number turn to the skin's warning colour while the mark is at 100% or beyond.
• It's this synth alone: not the DAW's meter, not the other plugins, not the whole computer. One instance per meter.
• What brings it down: fewer voices or notes held, less unison, less oversampling, processors turned off (an off processor takes nothing). A larger buffer in the DAW doesn't lower the number, but leaves more room for the slow moments.
• While the DAW renders offline (a bounce), it can read well over 100% with nothing wrong: the render just takes longer than the music lasts.
• With no audio running (the DAW stopped the synth) it shows --.
• Measuring takes next to nothing, and nothing at all while the panel is closed.
• Hovering it shows what it is in the info box; there's nothing to click or set.

The output panel's header with the CPU meter: the bar, its mark ahead of it, the percentage.

DC filter

• DC FILTER, OFF by default: turned ON, it takes away an offset the patch leaves in the synth's output (a wave that isn't centred, a shaper bending one side more than the other, feedback stuck at a level). It works after the master volume and before the safety clip.
• An offset isn't heard, but it takes headroom from what is, thumps when it comes and goes, and is no good to what comes after the synth (a compressor, a speaker).
• It's a gentle high-pass at 2 Hz, far under the lowest notes: a 20 Hz note keeps its level. An offset is gone within about half a second.
• Very low notes with flat tops (a square wave in the bass) come out slightly tilted, their peaks up to about a dB higher.
• OFF: the output as the patch gives it, offset included. For when the wave's exact shape matters, or the synth is used for something other than sound (a control signal).
• Switching it doesn't click. It adds no latency and next to no CPU.
• It isn't part of the patch: it's kept with the plugin instance in the DAW project (or with the application on its own), loading a preset never changes it, and it's no step to undo.

Safety clip

• SAFETY CLIP, ON by default: a safety net on the synth's output, after the master volume and the DC filter, for a patch that gets out of hand (a feedback loop of modulations, a resonance, a delay or a comb feeding itself).
• Up to 0 dB the sound is left exactly as it is. Nothing comes out louder than +1.3 dB: what's over 0 dB is bent into that room, and flat beyond it.
• It isn't an effect: a sound pushed into it is distorted, harshly (it works after the oversampling). Turn the patch down (the master volume, a gain) so it only works by accident; to distort on purpose, use a saturator in the patch.
• A patch broken to the point of giving no sound at all (a filter blown up) comes out as silence instead of noise, until what broke is back in order.
• OFF: the output as the patch gives it, however loud. For when the level is looked after further down the DAW's chain.
• Switching it doesn't click. It adds no latency and next to no CPU.
• It isn't part of the patch: it's kept with the plugin instance in the DAW project (or with the application on its own), loading a preset never turns it off, and it's no step to undo.

Glide

• GLIDE (portamento): how long a note takes to slide to its pitch, from 0 (none) to 10 s.
• Where it slides from: the pitch its voice was at if the voice was sounding (even halfway through another glide); otherwise the pitch of the last note played. The first note has nothing to slide from.
• The slide is even in pitch: an octave takes the same time whatever the notes.
• GLIDE NOTES: which notes slide. ALL: every note. OVERLAPPING: only a note played while another is still held (key down, or a pedal holding it), so notes played apart don't slide and notes played joined do ("fingered" portamento).
• It doesn't depend on Legato: a note can slide and start again, or be joined without sliding.
• It applies to whatever follows the note's pitch (oscillators, a filter's key tracking).
• The glide time and the master volume are the only settings the DAW can automate (see Automation). A new time applies to the notes played from then on.

Oversampling

• UPSAMPLE FACTOR: the synth runs at this many times the DAW's sample rate.
• Higher values reduce aliasing (harsh, inharmonic tones from bright or heavily modulated sounds) but use more CPU.
• At 1, a wavetable's brightness can step as its pitch moves (a slide, a bend); from 2 up it doesn't.
• Adds a very short latency, which the DAW compensates. None at 1. A limiter in the patch adds its own.
• A sound coming in through the audio input is that much later again, which the DAW doesn't compensate.
• When to raise it (FM, fast modulation, high notes).
• RENDER UPSAMPLE FACTOR (RENDER UPS.): the oversampling used instead while the DAW bounces (renders offline), so a patch can be played at a low factor and rendered at a high one. LIVE, the default: rendered as it's played. Otherwise 2 to 16; a factor lower than UPSAMPLE FACTOR does nothing.
• A render at a higher factor takes that much longer and has less aliasing than what was heard while playing: for a harsh, bright sound on a high note at 1, a render at 16 leaves about a sixtieth of it. It's the same patch, but not the same sound to the sample.
• It's part of the patch, and the latency is the same while playing and while bouncing (the longer of the two factors'), so setting it can add a few samples of latency, which the DAW compensates.
• Limits: it needs a DAW that tells the synth when it bounces (see Bouncing and the tail). It isn't shown in a synth running on its own as an application.

Tempo

• What the synth keeps time with (the arpeggiator's rate).
• TEMPO, in a DAW only: HOST (the default) follows the DAW's tempo and song position; INTERNAL uses the synth's own tempo instead, whatever the DAW's is.
• INTERNAL TEMPO (TEMPO when the synth runs on its own): 20 to 300 BPM, 120 by default.
• On its own, the synth can follow a MIDI clock instead: this tempo is then used only while no clock comes.
• Drag the slider for whole beats (double-click: back to 120), or click the number, type a tempo (decimals allowed: 93.5) and press Enter. Escape leaves it as it was.
• On the internal tempo nothing follows the DAW's beat: the arpeggiator's SYNC and the LFO's do nothing.
• The internal tempo is part of the patch: saved in presets (loading one brings its tempo) and undone like any other value. The DAW doesn't automate it.
• Which tempo is followed (HOST or INTERNAL) isn't: it's kept with the plugin instance in the DAW project, and loading a preset doesn't change it.

View size

• Scales the whole window, from 25 % to 200 %. Applied when the drag ends.
• A new size must be confirmed: Keep (or Enter). Revert (or Escape), or waiting a few seconds, goes back. So a size too big or small to use never sticks.
• Kept with the plugin instance in the DAW project (between runs, on its own), not in the patch or its presets.

Skin

• How the synth looks: choose one of its skins in the SKIN box. The whole window changes at once; the sound doesn't.
• Which skins there are depends on the synth. A synth with a single skin has no SKIN box.
• Kept with the plugin instance in the DAW project (between runs, on its own), not in the patch or its presets: loading a preset doesn't change it, and it's not undone.

Keyboard shortcuts

The shortcuts editor, one shortcut waiting for its keys

• SHORTCUTS → EDIT (where the synth places it) shows the editor over the window: a row per action, each with two shortcuts (the second an alternative, empty for most).
• The actions every synth has: undo, redo, modulator on / off, panic, switch A / B, randomise and output meters on / off (these three start with no key) and, in the application, the computer keyboard on / off (their original keys: reference). A synth may add its own.
• To change one: click it (it reads PRESS A KEY), then press the new keys: a key on its own, or with Ctrl (Command on macOS), Alt, Shift. Backspace or Delete leaves it empty; Escape, or a click elsewhere, leaves it as it was.
• Keys another action had are taken from it: the line at the bottom says which action lost them.
• RESET: an action's original shortcuts back (taken from whichever action has them now). RESET ALL: every action's.
• Closing: CLOSE, Escape or a click outside the panel. No shortcut is triggered while the editor shows.
• Fixed keys can't be changed: Escape, and the keys of lists, menus and text fields, which come before any shortcut while one of those has the keyboard.
• A key the computer keyboard plays works as a shortcut only while that keyboard is off: prefer keys with Ctrl.
• Yours, not the patch's or the DAW project's: kept on this computer for every instance of the synth, in every project, and in the application. An instance already open shows them the next time its window opens. Not undone.

MIDI controllers #

• Two ways a controller (a wheel, a pedal, a knob, the pressure on the keys) reaches the sound, for two different jobs. They can be used together, even on the same controller.
• As modulation: part of the sound. The controller is a modulator like an LFO, with an amount, and it's saved in the patch and its presets. For what is played with expression: bend, vibrato depth, brightness under the fingers.
• MIDI learn: part of your setup. A knob of your controller turns a knob of the synth, which moves on screen and stays where it's left. It isn't in the patch: a preset loaded keeps your knobs' assignments.

Playing as modulation

• Which of these a synth has is its author's choice; they're listed in The stock processors. They are processors with a modulator button and a curve: drag the button onto a control, or select it and use the control's + (Adding a modulation).
• Pitch bend: the pitch wheel, down, centred or up. Bending the pitch is a modulation of an oscillator's pitch control by it, and the modulation's amount is the bend range. A synth usually comes with it already routed.
• MIDI controller (the mod wheel by default): any controller, chosen by its CC number in the box of its panel (1 mod wheel, 2 breath, 11 expression, 74 often brightness). The number is saved with the patch.
• Aftertouch: how hard the keys are pressed while held. On a keyboard with polyphonic aftertouch, or an MPE controller, each note follows its own key; otherwise all follow the keyboard's pressure.
• Timbre: where the finger is along its key on an MPE controller (slide), each note following its own. On an ordinary keyboard it follows controller 74.
• Velocity: how hard each note's key was struck, held for the whole note.
• Release velocity: how fast each note's key was let go. Nothing while the key is held, there at once when it's released, and kept while the note fades. For what the release should depend on: an envelope's release with a negative amount makes the notes let go fast end short. A keyboard that doesn't sense how its keys are released gives every note the same, and a note the pedal was holding is released at the highest. Since it arrives at once, a level routed to it is heard jumping.
• Note: the key played, from the lowest key a keyboard can have (0) to the highest (1), held for the whole note. For what should follow the keyboard: a filter that opens towards the top, envelopes that get shorter, a level that falls. It's the key itself, so a glide, a bend or another tuning doesn't move it; with its curve it can do something on a part of the keyboard only.
• Random: not something played either: a value drawn for each note when it starts, anywhere from -1 to 1, held for the whole note. With a small amount on a level, a cutoff or a pitch, no two notes are quite alike; on a pan, each note is somewhere else. A note's unison voices all get their note's value, and a note played legato keeps the one of the note it continues. Its curve says which values come up most (flat in the middle: mostly small ones; only the right half rising: never below 0).
• Unison spread: not something played, but used the same way: where each voice is among its note's unison voices, from one end (-1) to the other (1), as UNISON BIAS lays them out; the middle one, and a note with no unison, at 0. For what the unison's own settings don't spread: the modulation's amount is how far the outermost voices go.
• Unison distance: the same without the side: how far each voice is from the middle of its note's unison voices, 0 for the middle one (and a note with no unison) and 1 for the outermost on both sides. For what should grow towards the outside whatever the side (more noise in the outer voices), where the unison spread would add on one side and take away on the other. A synth may have only one of the two.
• They move smoothly: a controller's steps aren't heard. A note starts where the controller already is.
• They cost nothing until a modulation uses them.

the MIDI modulators' panels, one being dragged onto a knob

Their curve

• Each of them has a curve on its panel: how it answers to what's played. What comes in goes from left to right, what its modulations get from bottom to top. For the pitch wheel, the unison spread and the random, which go both ways, the left end is all the way down (-1), the middle is the centre and the right end all the way up.
• The default is a straight line from the bottom left corner to the top right one: the value as it comes. Left that way it costs nothing.
• What it's for: a velocity curve (soft playing that still opens the filter, or only hard notes that do), a mod wheel that does little at first and a lot at the end, a wheel that only bends upwards, a dead zone.
• Points: double-click anywhere away from a point to add one there (up to 32 with the two ends); drag a point to move it, no further sideways than its neighbours; double-click a point to remove it. The two ends are always there: they only move up and down, and a double-click puts one back in its corner.
• Two points at the same place across make a step.
• Bending: drag up or down anywhere away from the points to bend the line between the two points around the mouse, which follows the mouse's way. The part the mouse would bend is lit.
• The right-click menu: Remove point (Reset point on an end), Add point, Straighten segment, Reset curve (back to the straight line).
• Behind the curve, the value being played, as it comes from the keyboard: 0 to 127 for a velocity, a controller, the aftertouch and the timbre, the key's name for the note (C4 is middle C), -8192 to +8191 for the pitch wheel, the place from -1.00 to +1.00 for the unison spread (0.00 to 1.00 for the distance), the value drawn from -1.00 to +1.00 for the random. A mark on the curve shows where it is: straight down from it what comes in, straight left what the modulations get. It's the note played last, and there's nothing while its key isn't held; the release velocity is the other way round, shown from the key's release until the next note.
• The curve is part of the patch: saved with it and its presets, and each change is a step to undo. It isn't a parameter: the DAW can't automate it, and nothing modulates it.
• Changing it while a note sounds is heard at once, smoothly.
• Limits: the mark and the value are the first of a note's unison voices; after every sound is cut at once (a DAW's panic), the last value stays shown until the next note.

a velocity curve with two added points and a bent segment, a note held

MIDI learn

• MIDI learn, in the right-click menu of a control: a message asks to move a knob or fader of the controller; the first one moved is assigned to the control. Cancel MIDI learn in the same menu calls it off.
• From then on the controller sets the parameter's value: the control moves on screen, the DAW is told (it can record it as automation), and it stays where it's left.
• Which controls: the ones the DAW can automate, on a processor's panel or its graph node (Automation). Not on/off buttons of processors, nor the settings.
• One controller can be assigned to several controls; a control has one controller (learning again replaces it).
• Remove MIDI CC n in the control's menu unassigns it. MIDI MAPPINGS in the settings shows how many there are; CLEAR removes them all.
• Not undone by undo, neither the assignments nor what a controller sets (like the DAW's automation).
• Saved with the plugin instance in the DAW project (What's saved with a project), not in presets.
• Pedals (sustain, sostenuto) can't be learnt: they hold notes.

a knob's right-click menu with the MIDI items

A controller's range

• A learnt controller covers the control's whole travel at first.
• To cover part of it: put the control where the controller's lowest position should be and choose MIDI CC n: lowest here; the same for highest here.
• Lowest above highest reverses the controller.
• MIDI CC n: whole range goes back to the full travel.
• For a controller that adds to a control instead of setting it, or that moves several controls by different amounts, use a MIDI controller modulator.

MIDI channel

• MIDI CHANNEL, in the settings: OMNI (every channel), MPE (MPE) or one of the 16.
• Omni treats all channels as one keyboard: a controller box on another channel than the keys moves the same sound, and a pedal on any channel holds every note.
• With one channel chosen, everything on the others is ignored: notes, pedals, wheels and controllers (learnt ones too).
• Changing it releases the notes sounding.
• Which MIDI devices reach the synth is the DAW's to set (its track's input); on its own, the synth's (MIDI inputs).
• Saved with the plugin instance in the DAW project, not in presets.

Program changes

• A MIDI program change loads the preset given that program number. Without one for that number, nothing happens.
• Giving a preset its number: in the preset browser, right-click the preset → MIDI program..., type a number from 1 to 128 and APPLY. Emptied, the preset has no program again. Presets built into the synth can have one too.
• Programs are counted from 1 here, as most controllers show them. One that counts from 0 sends its program 0 for the synth's 1.
• A program is one preset's: given to another preset, it's taken from the one that had it (the status line says which). A preset has one program at most.
• A preset's program is shown with its details when it's selected in the browser.
• The preset is loaded as from the browser: a step to undo, with the browser showing it as the one loaded, and with what's locked left as it is. It works with the synth's window closed.
• Only program changes on the MIDI channel listened to count. If several arrive before one is loaded, the last one is.
• Loading a preset cuts the notes sounding, as from the browser.
• Program numbers are yours, like favourites: not in the patch or the preset file, the same for every instance of the synth, kept when a preset is renamed or moved.
• In a DAW, whether program changes reach the plugin is the DAW's to decide (some keep them for themselves).

MPE

• For an MPE controller (one that plays each note on a MIDI channel of its own, so each has its own bend, pressure and timbre): choose MPE as the MIDI channel, and set the controller, and the DAW's track if it has such a setting, to MPE.
• Each note's bend is its pitch: every oscillator (and a filter following the key) follows it by itself, in semitones, with nothing to route. The range is 48 semitones each way unless the controller sets another (its pitch bend range setting).
• Each note's pressure and slide are the Aftertouch and Timbre modulators: route them like any modulator. A synth's presets may or may not use them.
• The controller's shared channel (the first one) is everything's: its pitch wheel is the Pitch bend modulator, its controllers and pressure reach every note, its pedals hold every note.
• MIDI learn only listens to the shared channel there: what the fingers send all the time isn't learnt.
• The lower zone over all 16 channels is assumed; a controller that announces other zones (lower, upper, their sizes) is followed.
• With more notes than voices, notes are taken over as usual.
• In a DAW that sends notes with their own expressions instead of MPE (CLAP), a note's tuning, pressure and brightness are followed the same way, whatever the MIDI channel setting.

In your DAW #

Automation

• What the DAW can automate: every processor parameter, the amounts of the modulations (32, plus the global 32), the glide time and the master volume.
• What it can't (changed in the synth only, saved with the patch): processors' on/off, containers' choices, connections, which modulator goes to which parameter, a modulation's direction and on/off, the other settings.
• Moving a control in the synth is recorded by the DAW as automation, when it's recording.
• Automation is applied where the DAW puts it, to within a millisecond or two, and glides from there to the end of the audio block (or to the next automation point in it).
• Values are shown in the DAW with the same texts as in the synth.

Modulation by the DAW

• In a DAW with modulators of its own (Bitwig Studio, with the CLAP plugin): its LFOs, envelopes and the like can move the synth's parameters without changing their values. The knob stays where it is; the sound moves around it.
• Which parameters: the ones that can be automated and have a continuous value (not choices, switches or whole numbers), except the glide time, the unison's settings and the master volume.
• Per note: a parameter of the per-voice part of the synth can be moved differently for each note sounding (a DAW's polyphonic modulators), each note with its own amount from its first sample. A parameter of the global part is moved once, for all notes.
• It adds to the synth's own modulations, and the result is held within the parameter's range.
• Not shown in the synth: its controls and displays show the parameter's own value and the synth's own modulations.
• Not saved with the patch or its presets: it's the DAW's, kept in its project.
• Only the CLAP format carries it. As VST3 or AU the DAW can only automate.
• With an arpeggiator on: every note it plays for a held key has that key's per-note modulation, for as long as the key is in the pattern.

What's saved with a project

• The whole patch, graph view included, plus the view size, the skin, which tempo is followed, whether the DC filter and the safety clip are on, whether an MTS-ESP master is followed, the MIDI channel and the MIDI mappings.
• Not saved: the selected modulator, the clipboards, the undo history (a project or DAW preset loaded clears it).
• DAW presets (the DAW's own preset files) hold the same thing as the project does.

Notes, pedals and pitch wheel

• Notes are placed exactly where they are in time, not at block edges.
• Sustain and sostenuto pedals hold notes.
• The pitch wheel, the other controllers, aftertouch and velocity: see MIDI controllers.
• Some plugin formats don't hand controllers to a plugin as they are; the synth still gets them (the DAW maps them), but which ones may depend on the DAW.
• A note repeated while it still sounds replaces the previous one, which plays its release.
• All notes off / all sound off (the DAW's stop or panic).
• Stereo output. An audio input only in synths that have one: Audio input.

Audio input

• Only in synths with an audio input among their processors. The plugin has a stereo audio input next to its output: what the DAW sends there (the sound of the track it's on, another track as a side chain: how depends on the DAW) comes into the patch.
• Where it comes in, and what to do with it: the audio input node. Until that node is turned on and connected, the incoming sound is ignored and costs nothing.
• A DAW that sends one channel: it's heard on both sides.
• With oversampling on, the incoming sound is later than the notes by the oversampling's latency, which the DAW doesn't compensate.
• Whether a DAW lets an instrument take audio at all depends on the DAW and on the plugin format.

Notes out

• Only in synths with note processors. The plugin has a note output next to its audio output: route it to another instrument's track in the DAW (how depends on the DAW).
• What comes out of it: Sending the notes on.
• The notes are placed exactly where the synth plays them.
• Whether a plugin's notes can be routed at all depends on the DAW and on the plugin format.

Tempo and song position

• The synth follows the DAW's tempo where it keeps time (the arpeggiator's rate, an LFO's or an envelope's times with TEMPO on). Without one, 120.
• Unless TEMPO in the settings is INTERNAL: the synth then keeps its own tempo and ignores the DAW's tempo and song position.
• While the DAW plays, the arpeggiator's SYNC follows its song position too, so its steps fall on the beat; so does an LFO's, for its cycle.
• A tempo change or a jump is taken at the start of the next block of audio.
• Limit: nothing else locks to the DAW's bars and beats (envelopes).

Automation and mapping marks

• In a DAW that tells the plugin (CLAP only), a knob, slider or value box shows what the DAW does with its parameter, as small marks at its top left corner.
• A dot: the parameter has automation in the DAW. A ring while it has some that isn't being played; filled while it's played, recorded, or overridden (the control was moved since).
• A square: the DAW has something mapped to the parameter (a knob of a hardware controller, one of its own macros).
• Their colours are the DAW's (an automation lane's colour, a controller page's). Where it gives none: the skin's accent colour, and its warning colour for automation being recorded or overridden.
• What is mapped, in words, is the title over the DAW's entries in the parameter's right-click menu.
• Only on knobs, sliders and value boxes: switches and boxes of choices show no marks.
• Nothing of it is saved with the patch: the DAW says it again for each project.

three knobs: one with a ring, one with a filled dot in a lane's colour, one with a dot and a square

The DAW's menu entries

• In a DAW that offers them (CLAP only), the right-click menu of a parameter's control ends with the DAW's own entries for that parameter, after the MIDI learn items: showing its automation lane, mapping it to a controller, whatever that DAW has.
• They are the DAW's: their names, which are ticked or greyed out, their submenus, and what they do.
• Only for the parameters the DAW can automate.

The track's name

• In a DAW that says it (CLAP only), the synth shows the name of the track it's on, after a mark of the track's colour: what tells one instance's window from another's. Where it is depends on the synth.
• It follows the DAW: renamed or recoloured there, it changes here.
• Empty in a DAW that doesn't say, and when the synth runs on its own.

Key names

• With a tuning other than the standard one (twelve equal notes to the octave, one per key), a DAW that shows key names in its piano roll (CLAP only) gets them from the synth.
• A key's name is the note of the scale it plays, as the tuning has it (3/2, 701.955; the scale's first note is 1/1, or 0.0 in a scale of cents), then the period it's in when that isn't the root key's (3/2 +1, 9/8 -2).
• A key the keyboard mapping leaves silent has no name.
• They follow the tuning, its mapping and the root key as they're edited, and the patch loaded.
• With the standard tuning the DAW's own names (C3, F#4) are left alone.
• Limit: while an MTS-ESP master tunes the synth, the names are still the patch's tuning's.

Bouncing and the tail

• The synth tells the DAW how long the patch rings after the last note (CLAP only), for a DAW that uses it to know how long to keep rendering: the longest release of the envelopes in use, then the time each delay and reverb in use after the voices takes to fall silent.
• It's an estimate from the parameters as they're set, whatever modulates them. It follows the patch as it's edited, automated or loaded.
• A delay with its FEEDBACK at 100 % never ends: the DAW is told so.
• The synth also knows when the DAW is bouncing (rendering offline) rather than playing: it then runs at the patch's render oversampling if one is set. As the bounce starts and ends the synth is silent for an instant and any note still sounding is cut.
• Limit: a DAW that bounces in real time, or doesn't say it's bouncing, gets the sound as it's played.

Several processor cores

• In a DAW that lends its plugins its thread pool (CLAP only), a heavy patch is worked out on several cores at once: its voices in groups, each group on a core. The sound is exactly the same.
• It happens by itself, only while it's faster: several notes sounding in a patch heavy enough (many processors, unison, a high oversampling). A light patch stays on one core.
• Limits: the effects after the voices stay on one core; there's no switch for it (the DAW decides whether to lend its cores); the VST3 and AU versions and the synth as an application use one core.

On its own, without a DAW #

• A synth may also come as an application: the same synth in a window of its own, played from the computer's MIDI devices (or from its keyboard) and heard through its audio output. Nothing else is needed.
• Everything in this manual applies, but for In your DAW: there's no automation and no DAW tempo.
• Its devices are chosen in the settings: ones the plugin doesn't have.
• With no DAW to follow, the tempo is the synth's own, set in the settings and saved with the patch, or another device's: a MIDI clock it follows.
• The window can't be dragged to another size: its size is the view size.
• Closing the window quits.

the standalone window showing the devices' settings

Audio output

• AUDIO SYSTEM: which of the computer's audio systems is used (Linux: PulseAudio, ALSA, JACK; Windows: WASAPI…; macOS: Core Audio). Only the ones that work on this computer are listed. Changing it goes to that system's default output.
• OUTPUT: the device. Default is the one the system uses for everything else.
• SAMPLE RATE: 44100 to 96000 Hz. Higher costs more of the processor.
• BUFFER SIZE: 32 to 2048 samples. Smaller: less delay between a key and its sound. Bigger: fewer crackles when the computer is busy. 512 to start with.
• AUDIO: what the output is running at and the delay its buffer adds, or why it isn't running. Without an output the synth still works, in silence.
• Any change stops the sound for an instant.
• A device chosen that isn't there at the next start (unplugged) gives way to the default one, and is used again once it's back.
• INPUT: only in synths with an audio input. The device the audio input node takes its sound from: a microphone, an interface's input. None at first: with a microphone near the loudspeakers the sound feeds back, so turn the synth down or use headphones before choosing one. Default is the one the system uses for everything else.
• If the input can't be opened, the output plays on without it and AUDIO says why.
• Stereo output and input. The input is a device of the same audio system as the output.

MIDI inputs

• Every MIDI input the computer has is listened to, as one keyboard: nothing to set up to start playing.
• MIDI INPUTS shows how many are listened to, of the ones there are. Click it for their list: a tick marks the ones listened to; picking one turns it on or off.
• A device plugged in while the synth runs is listened to within a couple of seconds.
• Notes, pedals, pitch wheel, controllers, aftertouch: as in Notes, pedals and pitch wheel. The MIDI channel, MIDI learn and program changes work the same.
• What's played sounds from the start of the next audio buffer (see Known limitations).

Playing from the computer's keyboard

• The keyboard button (where the synth places it), lit while it's on; Ctrl+K does the same (Command on macOS; a shortcut you can change). Only in the application: a plugin has no such button, its DAW has its own.
• The keys, laid out as a piano's: the home row is the white keys (A S D F G H J K L: C D E F G A B C D), the row above it the black ones (W E, T Y U, O). A starts at middle C.
• Z: an octave down; X: an octave up. Notes already held keep sounding where they are.
• A note lasts as long as its key is down. Every note has the same velocity; no pedals, wheel or aftertouch.
• While it's on, those keys are notes only: no shortcut on one of them is triggered. Keys with Ctrl (undo...) and the other keys (M, Escape) work as always.
• While typing in a text field, the keys type; they play again once the field is left.
• Clicking another window ends the notes held.
• The keys are the letters, wherever the keyboard's layout puts them: the piano shape is a QWERTY keyboard's.
• It's heard whatever the MIDI channel, alongside the MIDI inputs.
• Whether it's on, and its octave, are kept between runs.

the keyboard button lit, and the keys' layout over a computer keyboard

MIDI clock

• MIDI CLOCK: the MIDI device whose clock the synth keeps time with (a drum machine, a sequencer, another program), in place of its own tempo. None at first: the synth's own tempo.
• Any of the computer's MIDI inputs, whether it's among the MIDI inputs listened to for notes or not. One at a time; the other devices' clocks are ignored.
• What follows the tempo follows the clock's: the arpeggiator's rate, an LFO's, an envelope's or a delay's times with TEMPO on.
• While the clock plays (its device was started), the synth has its song position as in a DAW that plays: the arpeggiator's SYNC puts the steps on its beat and an LFO's its cycle. Start, stop, continue and a song position sent by the device are followed.
• CLOCK, under it: the clock's tempo and whether it's playing or stopped; No clock while none comes from the device chosen; Not followed with None.
• While no clock comes (the device is off or unplugged, or sends none), the synth's own tempo is used, and nothing follows a beat. When the clock is back it's followed again.
• A device that sends its clock only while it plays: the synth's own tempo until it starts.
• The tempo is steadied: after a change on the device, the synth is on the new tempo within a beat or two, and back on the device's beat a few seconds after.
• Any tempo the device sends is followed, also outside the synth's own 20 to 300 BPM.
• Choosing one doesn't interrupt the sound. A device unplugged shows as None, and is followed again when it's back.
• The choice is kept between runs. The synth sends no clock of its own.

MIDI output

• Only in synths with note processors. MIDI OUTPUT: the MIDI device the synth's own notes are sent to (Sending the notes on), to play another instrument with them. None at first: nothing is sent.
• One device at a time. Choosing one doesn't interrupt the sound.
• A device unplugged shows as None, and is used again when it's back.
• When the device is changed, the audio output is changed or the synth quits, the notes left sounding on the device are ended.
• The synth's own MIDI ports aren't offered, as an input or as an output: its notes would come back and play for ever.
• The choice is kept between runs.

What's kept between runs

• On quitting, everything is saved; the next start carries on from it: the whole patch, the view size, the skin, the MIDI channel, the MIDI mappings, the devices chosen and the computer keyboard (on or off, its octave).
• The first start, or one with --reset, is the default preset if there's one, else Init. Every other start is the last run's patch, whatever the default preset.
• What a DAW project keeps for the plugin, the application keeps for itself. The plugin and the application don't share it; presets are how a sound goes from one to the other (they have the same presets).
• Not kept: the undo history, the selected modulator, the clipboards.
• If the application doesn't get to quit (a crash): session recovery.
• Where: a file named standalone.json in the synth's settings folder (Linux: ~/.config/<vendor>/<synth>).

After a crash: session recovery

• While it runs, the application keeps a copy of the session aside every 15 seconds, whenever something changed: the whole patch, the MIDI channel and the MIDI mappings.
• If it closes unexpectedly (a crash, the computer switching off), the next start opens as the last normal quit left it and asks: RESTORE THE LAST SESSION?
• RESTORE (or Enter): brings back what the copy had, in place of the current patch. Up to the last 15 seconds before the crash may be missing.
• DISCARD: forgets the copy for good.
• The question has to be answered before anything else is done; closing the window instead keeps it for the next start.
• Not brought back: the devices chosen, the view size and the skin (they stay as the last normal quit left them), the undo history.
• Nothing is asked if nothing was changed before the crash, or after quitting normally. Moving things about in the graph alone doesn't count as a change.
• Several copies of the synth running at once each keep their own session; one that's still running is never offered to another.
• Only when the synth runs on its own: in a DAW, the project is what keeps the plugin's state.

the question asked after a crash, over the window

Starting it from a terminal

• --reset: start from the synth's default patch and settings, ignoring what the last run left (if a saved setting keeps it from starting well).
• --recover: after a crash, restore the last session without asking (what to use when it runs without a window, where it only says a session can be restored).
• --list-devices: print the audio systems, their outputs, the MIDI inputs (the ones a MIDI clock can be followed from) and the MIDI outputs, and quit.
• --help: list the options.
• Without a display it runs without a window, until interrupted (Ctrl+C), which also saves.

Notifications #

• Appear over the UI when something asked for couldn't be done (a full mod matrix, a preset that can't be read or written…).
• Go away by themselves after a few seconds, or at once with a click.
• What worked is reported quietly instead, e.g. on the preset browser's status line.

Mouse and keyboard reference #

• macOS: Command in place of Ctrl for the text shortcuts.
• M, undo, redo, panic's Ctrl+P and the computer keyboard's Ctrl+K are the original keys of shortcuts you can change.

The stock processors #

• Processors that come with SynthWorld; a synth may use some, all or none of them, under names of its own. One subsection each, with its controls:
• Wavetable oscillator: follows the played note; plays a wavetable, a set of waveforms it can morph between, chosen with the button over its display (Wavetables). POS goes from the first waveform (0) to the last (1) and PHASE is a fraction of the cycle, whatever the wavetable; it starts at 0, the beginning of the cycle. RANDOM (PHASE RANDOM in the DAW's parameter list): how much of the cycle each note's start is drawn at random over, from PHASE on. At 100 %, where it starts, a note begins anywhere in the cycle, so oscillators playing together don't all begin at once and add up to a loud peak; at 0 every note begins exactly at PHASE, the same every time (a bass or a pluck whose attack must not change). It's taken when a note starts: turning it changes the next note, not the ones sounding, and it can't be modulated (a macro can still move it). Limits: oscillators at exactly the same pitch keep their distance for the whole note, so with RANDOM up their level and tone differ from note to note; two exports of the same notes don't begin alike unless it's at 0. PM, the input socket after them: what modulates its phase, as the simple oscillator's. SYNC (0 to 48 semitones): hard sync. It plays the waveform that far above the note and starts it again at each cycle of the note, so the pitch stays the note's and the tone gets brighter and more hollow or vocal as it's turned up; modulated (an envelope, an LFO), it's the classic sync sweep. 0: no sync. PHASE is then where the waveform starts again. Limits: the display doesn't show it; on very high notes a lot of sync can sound harsh (aliasing), which oversampling reduces. Its tuning (COARSE, ST and CT, as the simple oscillator's) is a value bar under its display, there in the unison's view too. Its panel shows the wavetable's waveforms one behind another, with the one played drawn over them: where POS and its modulations put it, in the note played last. While PHASE is dragged, a mark on that waveform shows where in its cycle the oscillator starts, modulations included.
  • The unison button, in the top left corner of the display, shows the unison in the same room: its knobs take the place of the oscillator's own, and the display shows the unison's voices in the place of the wavetable. Clicking it again brings the oscillator's own back. It changes what's shown, not the sound, and isn't saved with the patch. In an oscillator with other processors built in after it (a gain, a panner), their knobs make room for the unison's too, and come back with the oscillator's own.
  • The unison display: each voice is an upright line, where it is between left (L) and right (R) and as tall as its level. A line's colour is its detune: the further the voice is from the pitch, above or below, the more of the accent colour it takes. A faint fill under the lines' tops shows the shape they make. It follows DETUNE, WIDTH, BLEND and BIAS with their modulations, in the note played last.
  • Unison: VOICES (1 to 16) plays that many copies of the oscillator at once, each a little off the others, for a wider and thicker sound at a fraction of what as many oscillators would cost. 1 is the oscillator alone, and the other unison knobs then do nothing. It's a setting for every note: it can't be modulated, and changing it while notes sound fades the voices in and out.
  • DETUNE: how far the outermost voices are from the pitch, in cents, the others in between (not quite evenly, so their beating doesn't pulse).
  • WIDTH: how far the voices are spread between left and right; 0 puts them all in the centre.
  • BLEND: above 0 the central voices are louder, below 0 the outer ones (the most detuned and the widest). Needs three voices or more.
  • WT SPREAD: gives each voice its own wavetable position around POS, which the central voice plays; at 100 % the outermost are the whole table away, one before and one after.
  • BIAS: how the voices are laid out between the central one and the outermost. At 0 they're evenly apart; above 0 they gather towards the centre, below 0 towards the outermost. Their detune, their place between left and right and their wavetable position all follow it; the central voice and the outermost ones stay where they are, so DETUNE, WIDTH and WT SPREAD still say how far the unison reaches. Needs four voices or more.
  • The level stays that of one voice whatever the number of voices, the width and the blend. Each note starts its voices at phases of its own, so no two notes begin alike, as far apart as the oscillator's RANDOM lets them: at 0 they all begin together, a loud peak at each note's start. With DETUNE at 0 the voices don't drift apart, and how loud a note is then depends on those phases.
• Simple oscillator: follows the played note, with a choice of basic waveforms; light on the CPU and often used as a modulator, e.g. for FM. Its waveform (WAVEFORM in the DAW's parameter list) is chosen with the row of buttons over its display, one per waveform, each under a drawing of its waveform and with no name: SINE, TRIANGLE, SAW (rising), SQUARE or PULSE, in that order. The buttons are thin slots: only the chosen one's is lit, in the accent colour like its drawing; clicking the drawing counts as clicking its button. It's chosen, not modulated, and nothing morphs between them; changing it while a note sounds fades to it. COARSE: its frequency as a ratio of the note's, from 0.125 to 4. ST and CT: transpose it by semitones (two octaves either way) and cents. The three are its tuning: a value bar beside its waveform's buttons, with PM as its fourth value. It has no knobs: PHASE, RAND and PW are a second value bar under the tuning, so the whole panel is a strip two bars high, made for several to go one under the other. In an oscillator with other processors built in after it (a panner, a gain), their knobs are more values of that second bar. PHASE: where in its cycle each note starts, as a fraction of the cycle; every waveform starts its cycle going up through the middle. RAND (PHASE RANDOM in the DAW's parameter list): how much of the cycle each note's start is drawn at random over, from PHASE on, as the wavetable oscillator's RANDOM. It starts at 0 here: as a modulator, where its cycle is against the oscillator it modulates is part of the tone, which would then change from note to note. Turn it up for an oscillator that's heard. PM (PHASE MOD): where modulations of the oscillator's phase go, for FM sounds. It has no value of its own: it's an input socket, not a knob, so it can't be dragged, typed or pasted into, and the DAW doesn't list it. It only takes modulations (the usual +): modulate it with another oscillator, whose tuning sets the tone and the modulation's amount how bright it gets, up to 4 cycles each way (far more than PHASE, which only has one cycle in all and cuts off what goes past its ends). An amount of 0.5 already reaches all of it; past that the modulation is cut off at 4 cycles. An oscillator modulating its own PM is feedback. Limit: deep modulation of a high note adds harsh, out-of-tune partials (aliasing), which oversampling reduces. PW (WIDTH in the DAW's parameter list): the part of each cycle the PULSE is up for, from 1 to 99 %; 50 % is a square, and modulating it is pulse width modulation. The other waveforms ignore it, so it can only be set while PULSE is chosen: otherwise it's still there under its name, dimmed, reading --, and can't be dragged, typed or pasted into (it keeps its value for the next time; its modulations stay, can still be edited there, and are still listed in the modulation matrix). The pulse stays between the same two levels whatever its width, so away from 50 % it isn't centred: as a modulator it pushes its target more one way than the other. Limits: on very high notes the triangle, saw, square and pulse can still sound a little harsh (aliasing), which oversampling reduces. With unison, a note's unison voices each start where their own draw puts them: together with RAND at 0, a peak at each note's start.
• The simple oscillator's display, under its waveform's buttons and beside its values: one cycle of the waveform chosen. The PULSE is drawn as wide as PW makes it, following its modulations in the note played last. While PHASE is being set (dragged, or its value typed), a line shows where the cycle starts, modulated likewise. It doesn't show where the oscillator is in its cycle.
• Sampler: plays a recording at the played note's pitch, all of it or a part, forwards or backwards, once or in a loop: The sampler.
• Velvet noise: silence with a pulse now and then, up or down at random, spread evenly in time. DENSITY: how many pulses per second, from 1 to 20000. From some 2000 up it sounds like white noise, only smoother; lower, the pulses are heard one by one, as a crackle. The sparser, the quieter, unless LEVEL is on. JITTER: how irregular the pulses are; at 100 % each falls anywhere in its turn (noise), and at 0 % they're evenly spaced: a buzz at the density's pitch, each pulse still up or down at random. PW (pulse width): how long each pulse lasts, as a share of the time between pulses; at 0 % a single click, the brightest noise, and wider pulses make it darker and louder. A pulse ends when the next one comes. KEY, the switch drawn as a piano's keys: on, the density follows the played note as an oscillator's frequency does, and the DENSITY knob becomes RATIO: the pulses per second as a ratio of the note's frequency, from 0.125 to 4 (at 1 a pulse per cycle of the note; with JITTER at 0 % the pulses then play the note, at 2 an octave up, at 0.5 an octave down); the density in Hz is then unused. Off, it ignores the played note. LEVEL, on unless turned off: the sparser or narrower the pulses the bigger they are, so the noise stays as loud whatever DENSITY and PW are (as loud as at 2000 Hz with no width); pulses grow up to 8 times, so below some 30 Hz it gets quieter again. MONO, on unless turned off: the same noise in every channel, in the middle. Off, the left and the right channel each have their own noise, as dense and as loud, so it sounds wide; it then takes about twice the CPU. Each note has its own noise.
• The velvet noise's display, over its knobs: the noise as a picture, cut into squares with one pulse each, up or down at random (two colours), as the noise cuts time into turns with one pulse each. DENSITY: how many squares, which grow and shrink from the picture's centre; with very few pulses per second a square is as tall as the picture, and stays so. With KEY on, the squares are those of RATIO at middle C, whatever the note played. JITTER: how far from rows and columns the pulses fall. PW: how much of its square a pulse fills. It follows the knobs' modulations, in the note played last; it doesn't show LEVEL or MONO.
• Colour noise: noise of a chosen colour, which is how much it has of each frequency. Its colour (COLOUR in the DAW's parameter list) is chosen with the buttons under its display, one per colour, each a small glow of that colour and with no name: WHITE (as much of every frequency: a hiss), PINK and BROWN (darker: a waterfall, a rumble), BLUE and VIOLET (brighter), GREY (more of both ends, where the ear hears less, so every frequency sounds as loud) and GREEN (the middle alone, around 500 Hz), in that order. Under each glow a thin slot: only the chosen one's is lit, in the accent colour; clicking the glow counts. It's chosen, not modulated; changing it while a note sounds fades to it. Every colour is as loud as measured, though the bright ones sound louder. MONO, the last button of the same strip, its name over its slot (on by default): the same noise in every channel, in the middle; off, each channel has its own, which sounds wide and takes about twice the CPU. It has no knobs. In a narrow panel the buttons are in two rows or more.
• The colour noise's display, over its buttons: a round glow of the chosen colour, fading out from its centre. It changes at once, without the fade heard, and shows nothing of the sound itself.
• LFO: a slow wave for vibrato, tremolo or filter sweeps. SHAPE: sine, triangle, saw up, saw down, square, RANDOM (a new random level held for each cycle: sample and hold) or SMOOTH (gliding from one random level to the next). Changing it while a note sounds fades to the new shape. REVERSE: plays the shape backwards (a saw falls instead of rising, a sine dips first), fading to it the same way; the random shapes ignore it. The random levels are each note's own; with SYNC following the DAW they are the song's instead, the same for every note and every time that part of the song plays. FREE: off, its cycle restarts with every note; on, it keeps going between notes. TEMPO, the switch drawn as a musical note with no text, lit while on: its speed follows the tempo (the DAW's, or the synth's own), and the knob becomes RATE, the length of a cycle as a note length, from 4/1 (four whole notes) to 1/64 (a dot: dotted, T: triplet, as in the arpeggiator); the frequency in Hz is then unused. Each note-shaped TEMPO switch is for the knobs it's next to. The one beside DELAY and RISE (FADE TEMPO in the DAW's parameter list) is a switch apart: they become note lengths at the tempo (OFF for none) whatever the rate's TEMPO is. RATE can be modulated like any knob: it then steps from one note length to the next, per note. SYNC (shown with TEMPO on): while the DAW plays, the cycle is counted from the song's start, so it falls on the beat, the same for every note and every time that part of the song plays; notes no longer restart it. With the DAW stopped or on the synth's internal tempo, it does nothing. On its own, the song is a MIDI clock's while one is followed and playing. After a loop or a jump the LFO goes back on the beat smoothly. DELAY: how long the LFO does nothing after a note starts; its cycle starts when the wait ends (with FREE on, or on the beat, it runs on underneath). RISE: how long it then takes to grow, in a straight line, from nothing to its full depth: a vibrato that comes in gradually. Every note starts both again, except a legato one. PHASE: where in its cycle the LFO starts, in degrees (180: half way); it shifts a free running cycle and one on the beat as well, and the random shapes ignore it. SMOOTH: rounds the wave off (a square's edges and RANDOM's steps become slopes), by the same share of a cycle at any speed; the more, the shallower the wave, and at 100 % it's flat: the LFO does nothing. Its knobs have no text of their own: their value shows in the value popup while they're turned. Its panel shows one cycle of the wave the LFO gives: the chosen shape, reversed or not, rounded off as SMOOTH and its modulations leave it (for a random shape, the cycle it's in), with a mark where the LFO is now, in the note played last; the mark stands still while nothing uses the LFO (no note sounding, or no modulation reading it). The part between the wave and the middle line is lit, the brighter the further the wave is from the line, fading out towards it. Behind the wave, faint lines mark the cycle's start and its quarters, with their time from the start, each text ending at its line: in ms (seconds from one up) at the frequency, or with TEMPO on, as fractions of a whole note (at 1/4: 1/16, 1/8, 3/16, 1/4). The last one, brighter, at the wave's end, is how long a cycle lasts; they follow the knob, not its modulations. While PHASE is dragged, a line shows where in the cycle the LFO starts, modulations included (not for the random shapes, which ignore it).
• Envelope (AHDSR): attack, hold, decay, sustain while held, release. Typically modulates a gain or a filter's cutoff. It's also what decides when a note's sound has ended. A knob per stage, under its letter: A, H, D, S and R (ATTACK, HOLD, DECAY, SUSTAIN and RELEASE in the DAW's parameter list). SLOPE: a smaller knob under the attack's, the decay's and the release's (ATTACK SLOPE, DECAY SLOPE and RELEASE SLOPE in the DAW's parameter list), from 0.1 to 10, which bends that stage. At 1, in the middle, the stage is a straight line; above it, it covers most of its way early and eases into the level it ends at (a snappy attack, a decay or release that falls fast and fades slowly); below it, it starts slowly and does most of its way at the end. The time the stage takes doesn't change. The knobs have no text of their own: their value shows in the value popup while they're turned. CURVE, beside the slope knobs: the kind of bend they give. POWER: a rounded curve, even from one end to the other. ANALOG: the exponential curve of an analog envelope, which turns more sharply and then runs flatter (a release that drops and then lingers). Both are a straight line at 1, and half way through the stage at the same level for slopes above 1. TEMPO, the musical note under H, lit while on: the stages' times follow the tempo (the DAW's, or the synth's own). The attack, hold, decay and release knobs become note lengths (ATTACK LENGTH, HOLD LENGTH, DECAY LENGTH and RELEASE LENGTH in the DAW's parameter list), shown as fractions: from 1/64 to 4/1 (a dot: dotted, T: triplet, as in the LFO), or OFF for a stage with no length (they start as a short pluck: attack 1/32, no hold, decay 1/16, release 1/16); the times in seconds are then unused, and kept for when it's switched off. Modulated, a length steps from one note length to the next. A stage's time (with the tempo it's worked out at), slope and curve are taken when the note enters it: changing or modulating them reaches the next one, not a stage under way.
• The envelope's display, over its knobs: its shape, time across and level up, as the knobs have it (not their modulations), the part under it lit, fading out downwards. The sustain lasts as long as the note is held, so it has no width: it's the corner where the decay ends and the release starts. A dot on the shape marks each stage's end, with a faint line through it and its time under it, each text ending at its line: counted from the envelope's start, the sustain taking no time (at 100 ms each: 100 ms, 200 ms, 300 ms, 400 ms), so the last one, brighter, is how long all the stages take together; with TEMPO on, as fractions of a whole note (1/8, then 3/16…; 0 where every stage so far is OFF). A text with too little room since the line before is left out: zoom in to see it. A coloured mark shows where the envelope is, in the note played last, at the level it gives (on the sustain's corner while the note is held): off the shape when a note is released before reaching its sustain, or starts while the last one still sounds; none while it's off.
• Editing on the display (the cursor and what lights up say what a drag would do): drag a stage end's line sideways for that stage's time, or with TEMPO on its note length, the nearest to where you drag. Drag the dot where the decay ends up and down for the sustain level, which follows the mouse (the decay's time is its line's, above and below the dot). Drag up and down inside the attack, the decay or the release to bend it (its slope: the shape follows the mouse). Double-click any of them for its default: a line for its stage's time or note length, the decay's dot for the sustain, inside a bent stage for a straight line. The hold's inside does nothing. Lines at the same place (stages of no length): the drag moves the last stage's. Each drag or double-click is one undo step.
• The display's time axis: the whole envelope always fills it, so its scale follows the envelope's length: a longer stage squeezes the others rather than leave the display, and dragging the release's line, at the right edge, changes the scale under it. A drag is worth the time it covered at the scale it started at. The mouse wheel zooms in around the mouse (up to 64 times) and back out, for stages too short to see or grab; a scroll bar then shows under the times, to drag along the envelope (also: Shift + wheel, or a sideways wheel). Double-click where there's nothing to set (inside the hold, on the times or the scroll bar) to see the whole envelope again. Limit: the zoom isn't saved, and goes back to the whole envelope when a preset is loaded.
• Filter (state variable): resonant, 12 or 24 dB per octave. SLOPE, a box: how steeply it cuts beyond the cutoff, 12 or 24 dB per octave (a band-pass: half that on each side). It's chosen, not modulated; changing it while a note sounds fades to it. CUTOFF, in Hz. Q (RESONANCE in the DAW's parameter list), from 0.1 to 10: the emphasis around the cutoff, as loud there at either slope. TYPE: morphs from low-pass (0) through band-pass (0.5) to high-pass (1). DRIVE, from 0 to 36 dB: pushes the sound into saturation before the filter. Quiet sounds get louder by that much and loud ones are squashed rather than made louder, so the sound gets denser and brighter; at 0 dB the sound is left as it is. A hard-driven high note can sound harsh unless the synth is oversampled. MIX: from the sound as it came in, before the drive (0), to the filtered one alone (1). The piano keys beside the knobs, lit while on (KEY in the DAW's parameter list): the cutoff moves with the played note, an octave for each octave (following its glide and its bend), so high and low notes keep the same brightness; middle C leaves it where the knob is. Its panel shows what the filter lets through of each frequency, from 20 Hz at the left to 20 kHz at the right (faint lines at 100, 1k and 10k): the higher the curve, the louder that frequency comes out, and the lit part under it, fading out downwards, is what passes. The line across is the level of the sound as it came in; the resonance's peak rises above it. The curve follows CUTOFF, Q, TYPE, MIX and their modulations, in the note played last, and the SLOPE. Limits: it doesn't show the key follow (which moves the cutoff with each note) nor the drive, and what's cut by more than 54 dB is drawn at the bottom.
• The filters' panels follow their width: their knobs in a row, in two rows of three, or, two knobs wide, in rows of two under a shorter display (the ladder's two boxes then one over the other, its key follow button after the second; the formant filter's after its box; the comb filter's beside its display).
• Filter (ladder): four stages in a row with feedback around them, low-pass or high-pass. TYPE, a box: LOW-PASS or HIGH-PASS. SLOPE, a box: how steeply it cuts beyond the cutoff, 6, 12, 18 or 24 dB per octave. Both are chosen, not modulated; changing one while a note sounds fades to it. CUTOFF, in Hz. RES (RESONANCE in the DAW's parameter list), from 0 to 1: the emphasis around the cutoff, strongest in the low-pass; the sound away from the cutoff stays as loud whatever the resonance. Limit: it doesn't whistle on its own at full resonance. DRIVE, MIX and the key follow button: as in the state variable filter. Its panel shows what it lets through of each frequency, as the state variable filter's does, following CUTOFF, RES, MIX and their modulations, and the TYPE and SLOPE chosen.
• Filter (comb): the sound comes back one cycle of FREQ late and is mixed with the sound as it came, so the frequencies FREQ apart add up and the ones halfway between cancel out: a comb's teeth. FREQ, in Hz (20 to 20000): how far apart the teeth are, and the pitch it rings at. FEEDBACK, from -98 % to 98 %: how much of what comes back goes in again. At 0 % the teeth are smooth; further from it they sharpen into a pitch that rings on, like a string or a tube, louder and longer; under 0 % it rings an octave lower and hollower, on the odd harmonics. DAMP, in Hz: each trip round takes more off the frequencies above it, so a ringing comb loses its highs first, as a plucked string does; at the top (20000 Hz) nothing is taken off. MIX: from the sound as it came in (0) to what comes back alone (1); halfway, with little feedback, is the deepest comb. The piano keys, lit while on (KEY in the DAW's parameter list): the comb's frequency moves with the played note, following its glide and its bend; with FREQ at 261.6 Hz (middle C, where it starts) it rings at the note played: a short noise or a click through it is a plucked string. All but the key follow can be modulated. Its panel shows what it lets through of each frequency, 20 Hz to 20 kHz, in decibels (a line every 12, the sound as it came at 0 dB), as the flanger's does: a curve where the teeth are apart, and to the right, where they're too close together to draw, a band from their dips to their peaks. It follows the knobs and their modulations, in the note played last. Limits: with a lot of feedback it's loud at its peaks, which isn't made up for; a damped comb rings slightly under FREQ; the display doesn't show the key follow, and a peak over 24 dB or a dip under -24 dB is drawn along the edge; a note lasts while the comb rings (seconds, with the most feedback at a low frequency); in a panel two knobs wide the key follow button is beside the display.
• Filter (formant): gives the sound the resonances of a sung vowel (its formants), so it seems to say it. VOWEL, from 0 to 4: A (0), E (1), I (2), O (3), U (4), and anywhere between two of them; moving it (by hand, an LFO, an envelope) is what makes the sound speak. VOICE, a box: whose vowels they are, BASS, TENOR, ALTO or SOPRANO; chosen, not modulated, and changing it while a note sounds glides to it. SHIFT, from -24 to 24 semitones: moves all the resonances up (a smaller, brighter voice) or down (a bigger, darker one). Q (RESONANCE in the DAW's parameter list), from 0.25 to 8: how sharp the resonances are against the voice's own (1); higher values ring and leave less of the sound between them. MIX: from the sound as it came in (0) to the filtered one alone (1). The piano keys, lit while on (KEY): the resonances move with the played note, an octave for each octave, middle C leaving them where they are; off, as in a voice, they stay put. Its panel shows what it lets through of each frequency, as the state variable filter's does: each peak is one of the vowel's four resonances, following VOWEL, SHIFT, Q, MIX and their modulations, and the VOICE chosen. Limits: the sound comes out quieter than it went in (only what's around the resonances is left), which isn't made up for: follow it with a gain; it works best on sounds rich in harmonics (a saw, a pulse, noise); the display doesn't show the key follow.
• Envelope follower: a modulator that follows how loud a sound is, so one sound can move a parameter. Connect what it should follow to its side-chain input in the graph, the only input it has (on the global side: a send from a per-voice node), then route it like any modulator: to a gain, with a negative amount, to duck one sound under another; to a filter's cutoff for an auto-wah. ATTACK: how fast it rises when the sound gets louder (0.1 ms to 1 s: the time to cover most of the way). RELEASE: how fast it falls back (1 ms to 5 s); short, it wavers with a low note's own cycles, longer it's smooth. GAIN, in dB (-24 to 48): how much louder the sound is taken to be; at 0 dB a sound at full level gives the whole modulation, and nothing gives more than that. It changes nothing of what's heard.
• The envelope follower reads the sound and passes nothing on: its node has no output, and what's connected to it isn't heard through it. It does nothing, and costs nothing, until something is connected to its side-chain input and a modulation reads it.
• The envelope follower's display, over its knobs: the level it gives, over the last three seconds, now at the right, from nothing at the bottom to the whole modulation at the top line. On the global side it's the one follower's; per voice, the note's played last. It doesn't show the sound going in.
• Audio input: the sound the DAW sends into the plugin (Audio input), as a node of the graph. GAIN: how loud it goes on into the patch, from silence to +12 dB; 0 dB is the sound as it came. Connect its output like any node's: into effects, and the synth is an effect for that sound; into an envelope follower's side-chain input, and that sound moves a parameter (the synth ducked under a kick drum, a filter opened by a voice); into a compressor's, and the synth is turned down by it.
• The audio input passes on what's connected into it with the incoming sound added. Turned off, the incoming sound is muted. It does nothing, and costs nothing, until its output goes somewhere.
• On the global side the incoming sound is heard whether notes play or not. Per voice, every note gets it through its own processors (a filter and an envelope per note): it's heard only while notes sound, and once per note.
• Gain: amplifies or attenuates; modulated by an envelope, it shapes a note's volume.
• Stereo panner: moves the sound between left and right. PAN: one knob, its value in its centre: C in the centre, where the sound goes through unchanged, then how far to a side, up to 100, and the side (50L, 100R). Towards a side the other channel fades out, down to silence at the end, and that side's gets louder (3 dB at the end), so the sound stays as loud all the way. Typed: a number and its side (30L), or a signed number (-30: left). Each channel only gets louder or quieter: what's on the left is never moved to the right. Modulated by an LFO, the sound travels between the speakers. Limit: stereo only; with any other number of channels it does nothing.
• Pitch bend, MIDI controller, aftertouch, timbre, velocity, release velocity and note: what's played, as modulators (Playing as modulation). Only the MIDI controller has a control: its CC number.
• Random: a value drawn for each note, as a modulator with no control (Playing as modulation).
• Unison spread: where each voice is among its note's unison voices, as a modulator with no control; and unison distance, how far each is from the middle, both sides alike. The wavetable oscillator also starts each unison voice but the first at a random point of its cycle, so they don't all peak together.
• Why a patch with no envelope on its volume is cut short or never ends: Voices.

Known limitations #

• Moving a modulation cable to another parameter removes the modulations of its amount (Modulation cables).
• A muted modulation still keeps its modulator computed (CPU).
• A global modulator stops modulating the voices once their sound reaches it (Global modulators on the voices).
• Automation points closer than 64 samples to each other are applied together, up to that much early or late (Automation).
• At most 32 parameters can be moved per note at once by the DAW's own modulation (Modulation by the DAW).
• No system file dialog: the synth's own file picker.
• Wavetables: a patch names its wavetables and doesn't carry them (Wavetables and patches); of audio files, only WAV ones are imported (Audio files as wavetables); a wavetable's brightness can step as its pitch moves when oversampling is at 1, and one whose phase or frequency is modulated hard can still sound harsh (aliasing) on high notes, which oversampling reduces.
• Grooves: timing only (no accents or note lengths), drawn nowhere and edited nowhere in the synth: a groove of your own is a MIDI file imported. Only the arpeggiator plays one. A patch names its grooves and doesn't carry them (Grooves and patches).
• Chord maps: a patch names its chord map and doesn't carry it, and a map's chances are changed in the editor and saved with the map, not set per patch (Chord maps and patches). A map counts in keys: it must have as many notes per octave as the tuning played, with one note per key. The chord maker's controls are read when a key is played and can't be modulated; its roll at the tempo starts with the key, on no beat; its strict leading only tells parallel fifths and octaves between chords of as many notes.
• Samples: a patch names its samples and doesn't carry them (Samples and patches); of audio files, only WAV ones are imported, up to 30 seconds (Audio files as samples); a bright sample played above its own pitch can still sound slightly harsh between octaves, which oversampling reduces; choosing another sample ends the notes sounding; a library of many long samples makes the sample browser slow to open.
• MIDI: 14-bit controllers and NRPNs aren't read; in MPE, a note still fading out follows its channel if a new note takes it, and the zones a controller announced aren't remembered by a project (the controller says them again when it's switched on or set up); a learnt controller sets its parameter from wherever the knob is (no pick-up), a block of audio late (MIDI learn).
• MIDI mappings are kept per plugin instance: a new instance starts without them.
• The default preset: no row of the browser shows which it is, only the selected preset's details; if its file is gone or can't be read, the synth starts with Init without saying so; and the synth's browser doesn't show it as the preset loaded when a new instance opens.
• Program changes: 128 programs, no banks (bank select isn't read); the preset browser has no list of the programs given, only each preset's own; a preset renamed or moved outside the synth answers its program again once the browser has been opened or refreshed; a program change is loaded a moment after it arrives, not at its place in time.
• MTS-ESP: a master's per-channel tunings and MIDI tuning messages (MTS SysEx) aren't followed; the tuning panel doesn't say that a master's tuning is the one heard (the MTS-ESP MASTER setting does).
• The DC filter tilts the flat tops of very low notes (a square wave in the bass), which raises their peaks by up to about a dB; its 2 Hz can't be changed.
• The safety clip distorts what goes over 0 dB, with aliasing: it's a safety net, not a limiter.
• An undone removal of a modulation brings its row back at the end of the mod matrix, not where it was.
• On its own: MIDI is played at the start of the next audio buffer, not at its exact time (a smaller buffer tightens it); there's no transport of its own, so the arpeggiator's SYNC and the LFO's do nothing without a MIDI clock followed, and with one the synth is behind the clock's beat by the delay of its audio output (no setting makes up for it); a new audio output device shows up in the list only after a device setting is changed or at the next start; an output that's unplugged while playing isn't replaced by itself; the computer's keyboard doesn't play notes; two running copies of the same synth can't play each other through the MIDI output.
• Platforms: only Linux is tested so far; drawn mouse cursors only there.
• On Intel and AMD computers, the processor must have SSE4.1 (any from about 2011 on) unless the synth was built for older ones.