Playing

MPE

Off by default, and under Settings. With it on, a controller that gives every note its own MIDI channel can bend and press each note separately, which on this instrument means each finger gets its own copy of all 33 aftertouch destinations. Press into one note in a chord and only that note's upper partials come in.

Slide, the forward and back axis under a finger, has a destination of its own under Settings. By default it moves that note's brightness, thinning or opening the top of the series, which is what a player's hands expect from that axis. It can be aimed at the tuning instead, so that pushing one finger forward slides that note alone between equal temperament and just intonation while the rest of the chord stays where it is. A keyboard with no slide axis sends nothing and changes nothing.

It is read as movement rather than as a position: the first value a note is given becomes that note's nought, and what reaches the sound is how far the axis has travelled from it since. Controllers disagree about what the axis is, which is why. A Seaboard reports where the finger sits and rests at the centre, while an Expressive E Osmose spends the first part of the key travel on pressure and only then starts sending slide, from the bottom of its range upward. Read as a position that second kind lurches the moment the axis engages. Read as movement both start the note on the patch and follow the finger from there.

The travel is scaled so that the longer of the two directions away from the rest is the whole of the slide, and both directions move at that one rate. A rest near the bottom of the axis leaves very little travel underneath it, and giving that sliver a scale of its own would make it a hair-trigger: pressing in would feel smooth while coming back out snapped to dark. The short side keeps the long side's sensitivity and simply reaches less far, which is honest, since there is less of it.

It is a setting rather than something permanently on because it changes what a channel number means. With it off, a channel is ignored: a key-up on channel 7 releases a note that went down on channel 1, which is what a single-channel keyboard needs. With it on, the channel is part of who the note is, so the same key can be held twice on two channels, bent in two directions, as two voices rather than one retriggering the other.

Legato

A switch in Settings, beside the voice count, and what it does depends on that count. With one voice the instrument is not polyphonic at all: pressing a key while another is still held moves the note rather than starting it again, so the envelopes carry on and a run keeps the shape the first key gave it rather than restarting the attack under your fingers. Letting a key go while others are still down falls back to the most recent of them, which is what makes a trill work, and only the last key coming up releases the note.

With more than one voice it plays a top line. A key that lands above everything you are holding, except the highest note, carries that note's phrase on: it starts a voice of its own that takes over the highest note's envelopes where they stand and glides from its pitch, so it does not start again. Every other key plays as it always does. A legato melody over a staccato chord does what it sounds like: the chord retriggers each time its keys come up and go down again, and the melody slides from note to note over it, up or down, as long as it stays above the chord.

The note a melody carries on from is never taken away, so a key that turns out to be part of the chord always sounds. It goes on sounding while its key is down and fades quickly when the key comes up, rather than ringing out its release under the melody, and so does the upper note of a trill lifted back onto the key below.

A key going down under the highest note could be the melody stepping down or a chord struck under a held melody, and only what the old melody key does afterwards would tell them apart, so the instrument goes by what is likely. Above the highest note, a key always carries on. Under it, with a chord held and no other key just struck, it carries on if it is above that chord, however far down it goes. Alone under a lone melody, or among keys struck together, it carries on only if it is within a major third and the highest of the keys struck with it, since a melody moves by seconds and thirds and the top of a chord usually sits further down. A wrong guess costs no more than an attack. No key ever waits.

When the voices run out, the melody is the last thing taken. The sustain pedal holds the sound, not the line, so a note only the pedal is keeping does not carry anything on. With MPE on it works the same, and a note carried onto a key of its own channel brings that key's bend, pressure and slide with it.

Legato moves the note in a single step. With a glide time on the channels, it slides there instead.

Glide

Every channel has its own glide, the last knob under PITCH MOD. It is how long that partial takes to travel from the note before to the new one, from Off, where every channel starts, to five seconds. Because each partial travels on its own time, the series does not have to arrive together. Short times on the top and a long one on the fundamental land the brightness first and let the pitch follow. The other way round, the note arrives and its colour catches up. Somewhere between, the partials stop agreeing on a fundamental while they move and settle into one as they land, which no single portamento can do.

A partial glides in a straight line in pitch, and everything that follows the pitch follows the glide: the oscillator character is chosen at the frequency the partial has reached, and the Bulb character's lamp sags behind a long slide as a real one would. Vibrato, drift and both kinds of bend ride on top of it. The needle on each partial's PITCH MOD rule follows the glide, swinging back to centre as the partial arrives, so a series given different glide times can be watched landing a partial at a time.

Two switches govern every channel at once, on the two lit letters beside the GLIDE caption on the left of the mixer, L and T, and again at the end of the menu a right-click on any GLIDE knob opens, under Glide (every channel). A letter is lit when its switch is on. They belong to the patch, so a preset saves them and a legato lead can say so, and the letters show it the moment one loads.

Played polyphonically, each new note glides from the one played last. In Legato the voice carrying the phrase glides from wherever it is, the one voice with one voice set and the top line with more, so a key pressed halfway through a slide turns the slide round from where it had got to rather than jumping to either end. A trill falling back to a key still held glides back to it too.

GLIDE is a row like any other, so LINK sets it across the series in one drag and a macro can sweep every channel's glide time from one control.

The Settings menu

Everything set once and then left. The menu is where the instrument is set up and the panel is where the sound is made, and a control's place says which of the two it is.

Almost none of it travels with a preset, and that is a rule rather than a list that might fall behind one: how you play the instrument and what it is tuned to are properties of the session rather than of any one sound in it. The zoom, the window size and which groups of rows you have folded away are in the same position for the same reason.

The safety clip is the exception in this table, and the master fader is the exception outside it. Both used to sit with the rest on the grounds that how loud the instrument is belongs to the session, and both moved when the clipper became five shapes you choose between: the fader is the drive into that stage, so how hard a patch leans on it is part of the patch rather than part of the desk. The cost is worth knowing before you go looking for it: loading a preset now moves your output level.

Automating it from the host

Eight hundred and thirty-nine parameters reach the host: thirty-seven global ones, twenty-three on each of the 32 partials, eighteen on the noise channel, and six for each of the eight macros. They are listed in the order the panel reads, the globals first and then channel by channel with the noise channel last.

Every per-channel name carries its channel, so a parameter list answers both ways of looking for something. Type H7 and you have that partial, type Tune and you have that row across the whole mixer. The same names are what a screen reader reads, which is why they are words rather than numbers.

The noise channel has eighteen rather than twenty-three because it has no tuning, pitch modulation, drift or glide, and gains COLOUR in their place.

Everything in Settings is in the list too, even though no preset will ever move it. The temperament, the polyphony and the bend range are all automatable, so a piece that changes temperament partway through is something a host can do, whatever the panel thinks a setting is for.

A parameter is identified by an id of its own rather than by its position, in the Audio Unit as in the VST3. That is what lets a release add controls without breaking what you have already written: 1.6.0 put 65 new parameters among the existing ones rather than after them, and every automation lane written before it still points at the control it was pointing at.

LINK is not a parameter, since it is a tool for the mouse rather than part of the sound. A drag under it opens a gesture on every channel it reaches, writes each one, and closes them together, so what a host records is what actually moved: thirty-two lanes if the drag moved thirty-two channels.

Macros

A macro is one parameter that moves a whole row, and the reason it exists is automation. A Link drag across the tuning knobs moves 32 parameters, and a host catches only the last one you touched, so a relationship you can edit by hand cannot be recorded at all. A macro gives that relationship a single control a host can draw.

The icon of three faders on the top bar, beside the settings cog, opens them over the mixer. Each macro is one line: a colour, the row it drives, which channels of that row it reaches, how it is shared out across them, and the amount. Add a macro takes one of eight, and the cross at the end puts it back.

Eight is a pool rather than a limit you are meant to feel. A plugin declares its parameters once when it is created and neither VST3 nor AU can add one later, so making a macro takes one of eight that already exist and removing it puts it back. The panel shows the ones you have made, and your host's parameter list shows all eight whether or not you use them.

Scope is All, Odd, Even, or any one interval: choosing fifths reaches harmonics 3, 6, 12 and 24. Curve is Uniform or Taper, and a taper leans hardest on a channel you pick and fades with distance either side of it. Link's third curve, Spread, is not offered and could not be: it scatters each strip along the direction your drag gave it relative to the strip under the mouse, and a macro has neither a direction nor a strip under the mouse.

A macro offsets what the patch says rather than writing it. Nothing in your sound is overwritten, the host has one lane to draw instead of thirty-two to catch, and the knobs do not move when a macro does. That last part is normal for anything that works this way and is surprising the first time, so the controls say both at once: the ring around a knob lights to where the macro has taken the value, in the macro's colour, while the pointer on the cap stays at what the patch holds, in the channel's. The gap between them is the macro. A fader, which has no ring to spare, says it in the gutters either side of its track: two narrow columns of segments appear there, cut from the same cloth as the meter behind the fader, and light from the foot of the track up to the level it has been taken to. The glass of the cap takes a wash of the same colour.

With a macro at rest the two coincide, so a macro you have assigned but not moved looks like no macro at all.

The amount runs from -1 to +1 whatever it drives, so a macro can always push a control either way from where the patch left it, and it moves a proportion of each control's own travel rather than a number of its units. That matters most on a row like decay, which runs from a thousandth of a second to twenty: the same number of seconds added would be a twentyfold change at one end and a rounding error at the other.

A full sweep reaches the ends of the range, which means a patch already sitting near one end has a stretch of fader that does nothing in that direction. That is deliberate. One macro reaches up to 32 controls that may be set anywhere, so there is no smaller fraction of the range that would be right for all of them.

Right-click a macro's amount to bind a controller to it, the same MIDI Learn the mixer's controls use. Only the amount: the row, the scope and the curve say what a macro is rather than what it is doing. The macros icon lights whenever any are made, since a macro's work is otherwise invisible until you look at a ring.

Playing it from a controller

Right-click any control and choose MIDI Learn, then move something on your controller. A violet ring marks the control while it waits, so you can see what is listening without opening the menu again, and only ever one control carries it. Right-click the same control afterwards and the menu says which controller moves it, with the way to forget it or to bind a different one. In the mixer these items sit under the LINK settings that right-click already opens. On the bar and the noise channel they are a menu of their own, since neither is part of the series LINK gangs.

Your host almost certainly offers the same thing. What this does that a host cannot is let you make the map by right-clicking the knob itself rather than finding H17 AM Rate in a list of 836, and behave the same way wherever you open it.

Eleven controllers are refused, because something is already listening for them and losing it would be felt while playing rather than noticed afterwards. Controller 1 is the mod wheel, which can feed aftertouch. 64 is the sustain pedal. 74 is the slide axis, the one most easily bound by accident since it moves whenever a finger does on an MPE keyboard. 120 and 123 stop notes. Controllers 6, 38, 98, 99, 100 and 101 are how a keyboard tells the instrument what it is. Arming a knob and then reaching for the slide to steady a note leaves the note steadied and the knob still waiting.

It works the same with MPE switched on, which is not free: with MPE on, nearly every message belongs to the part of the instrument that reads your keyboard's layout, so a binding is looked for before that rather than after it. Otherwise it would be dead on exactly the controllers this is usually played from.

The map is saved with your project, beside the window size and the folded sections, and a preset never touches it. Which controller moves which knob describes the desk you are plugged into rather than the sound, and a patch that rearranged it would be rearranging your hardware. A binding records the parameter by name, so a later version adding controls cannot silently repoint it at something else.

Two things to expect. A control jumps to wherever the controller is standing rather than waiting to be picked up, so a knob turned to the top will take the parameter there the moment you touch it. And a controller moving a control is not something undo reverses, for the same reason automation is not.