Switch-Accessible MIDI: One, Two and Four Switch Rigs

Most music software assumes you have eighty-eight keys, or at minimum sixteen pads. Switch-accessible music making starts from a different number — one, two, maybe four reliable inputs — and asks what can be built from there. The answer is a great deal more than people expect, provided the design starts from the input count rather than trying to shrink a keyboard down to fit.

This guide is about that design problem specifically. Not which switch to buy, and not how to wire an Xbox Adaptive Controller — those are covered in the XAC MIDI guide. This is about what you do once the switches reach the computer: how a one-switch rig differs from a two-switch rig, which trigger behaviours earn their place, how to stop a switch firing twice, and where switch-driven MIDI genuinely runs out of road.

What does a one-switch music rig actually do?

It makes the single switch control a running process rather than play a note, so one press can select any of many outcomes depending on when it lands.

A switch has one bit of information in it. To get more than one outcome from one bit, you need a second variable, and the only one available is time. Every workable one-switch design uses timing in some form.

The three patterns that recur:

• Scanning. The software steps through a list of options — sounds, clips, notes — at a fixed rate, highlighting each in turn. The switch selects whatever is highlighted when it is pressed. This is the dominant pattern in switch-accessible computing generally and it transfers directly to music. The scan rate is the single most important setting and it should be adjustable per person.
• Step-advance. Each press moves to the next item in a prepared sequence. The musical content is pre-arranged; the player controls when each thing happens. This is how a lot of accessible live performance actually works, and it is far more expressive than it sounds, because the timing of the advance is a genuine performance decision.
• Trigger-in-time. A loop runs continuously and the switch fires a sound into it. The player is effectively a one-note percussionist in an ensemble, which is a real musical role and always has been.

The gamepad bridge's part in all three is small but necessary: it turns the switch closure into a MIDI note that your DAW, sequencer or scanning software can respond to. OneSwitch.org.uk's music section is the best public archive of one-switch music approaches and has been accumulating them for two decades.

One-switch latency tolerance is different

Worth knowing: in a scanning rig, the scan interval dominates everything. If the scan steps every 800 ms, a few milliseconds of MIDI transport latency is irrelevant. In a trigger-in-time rig it is the opposite — the player is playing rhythmically and latency matters as much as it does for any drummer. Our latency benchmark covers the wired versus wireless difference, and the short version is that a cable is worth using when the switch is being played rhythmically.

Why is two switches the real turning point?

Because two switches give you advance and confirm, which is enough to navigate any menu, select any sound, and play with intent rather than waiting for a scan.

The jump from one switch to two is bigger than the jump from two to four. With one, the player is always waiting for the software to offer the right option. With two, the player drives: one input moves the selection, the other commits. Nothing has to be timed to a machine's scan rate, and the whole experience shifts from reactive to active.

A good two-switch music layout:

Switch	Short press	Long press	Behaviour
Switch A	Advance selection	Jump back to start of list	One-shot note, with a separate note on hold
Switch B	Trigger selected sound	Latch selected sound as a sustained note	One-shot on tap, latch on hold

That uses a long-press distinction to double the effective input count. It is a genuine accessibility trade-off: long press costs time and requires sustained actuation, which is exactly what some people cannot do. Offer it, do not require it. If long press is not viable, three switches with no long press beats two switches with it.

Universal Controller MIDI handles the short-versus-long distinction as a per-input hold-time threshold in the mapping editor, which is a Pro feature. The same mechanism is what powers shift layers on a standard gamepad, covered in our banks and layers guide.

How many switches do you need for real production work?

Four to six, if they are the right four to six — and the right ones are transport, two sounds, undo, and stop.

• Transport, two sounds, undo
• Stop-all and scene step

Once you can start and stop the transport, play two different sounds, undo a mistake and kill everything, you can produce music. Everything beyond that is convenience, and convenience is worth having but it is not the threshold. A lot of adaptive rigs fail because they chase input count — nineteen jacks populated on day one — rather than getting six inputs genuinely right.

A six-switch layout that holds up:

Input	Function	Behaviour	Why it earns a slot
1	Play / stop	Toggle	Used more than anything else in a session
2	Record arm and start	One-shot	Separating record from play avoids accidental overwrites
3	Sound A	Momentary note	The primary musical voice
4	Sound B	Momentary note	Two voices is the minimum for a musical relationship
5	Undo	One-shot	Without undo, every mistake becomes a support request
6	All notes off and master mute	One-shot	The safety input. Always separate, always reachable.

Note what is not there: no filter sweeps, no scene navigation, no tempo control. Those get added later, by the person using the rig, when they ask for them. Adding them pre-emptively costs the player memory and gains nothing.

Which trigger behaviour should each switch use?

Use momentary only when hold duration carries musical meaning, and use toggle or latch everywhere else.

The default behaviour on any gamepad is momentary — note on when pressed, note off when released. That is correct for a drum pad and wrong for almost everything else on an adaptive rig, because it means holding a note requires holding a switch, and sustained actuation is precisely the thing many switch users are trying to avoid.

A decision rule that works:

• Does the sound need to sustain for more than about two seconds? Use latch. One press on, separate control off.
• Is it a binary state — mute, solo, loop on, effect bypass? Use toggle.
• Is it an event with no meaningful duration — scene launch, undo, next track? Use one-shot.
• Does how long you hold it change how it sounds? Only then use momentary.

We go through each of these in much more depth, including dwell-based triggering and the ergonomics of hold thresholds, in adaptive MIDI mapping design.

The latch trap and how to avoid it

Latch has one failure mode and it is severe: a note starts, the player cannot reach the stop, and the sound continues indefinitely. This is distressing in a way that no other mapping bug is, and it is entirely preventable.

Three defences, use all three:

1. A dedicated panic switch on a physically distinct actuator, mapped to all-notes-off plus master mute, placed where it is always reachable but cannot be hit by accident.
2. A maximum latch duration. Set a timeout so any latched note releases itself after a fixed period. Thirty seconds is generous and still finite.
3. A visible state indicator. If the controller has a light bar or rumble available, use it to signal "something is latched". The DualSense light bar is handy for exactly this, and our piece on bidirectional MIDI feedback covers driving it from the DAW side.

How do you stop a switch firing twice?

With a debounce window — a short period after each trigger where further presses on that input are ignored.

Double triggers come from two different places and they need different fixes. Mechanical chatter is the switch contact bouncing as it closes, and it produces two triggers within a handful of milliseconds. Physical double-actuation is a hand, foot or head bouncing off the switch, and it produces two triggers maybe 80 to 200 ms apart. A debounce window long enough for the second will feel sluggish if the first is all you have.

Symptom	Gap between triggers	Suggested debounce	Cost
Mechanical chatter	Under 20 ms	40 to 60 ms	Negligible
Hand or limb bounce	80 to 200 ms	150 to 250 ms	Caps that input at roughly four presses per second
Tremor-driven repeats	Variable, often rhythmic	See the tremor guide	Needs smoothing, not just debounce

For the third case, debounce alone is the wrong tool and will fight the player rather than help them. Our tremor-friendly setup guide covers the smoothing, hysteresis and quantisation settings that actually address it.

What music suits a switch rig?

Music built to need few triggers — loop-based, drone-based, ensemble parts and percussion — rather than music that happens to use fewer notes than intended.

This is the single biggest creative decision and it gets far less attention than the hardware. A two-switch rig playing a piano patch sounds like a piano being played badly. The same rig playing a well-chosen percussion voice over a running loop sounds like a person playing percussion, because that is what it is.

Approaches that work well with low input counts:

• Loop-based arrangement. Prepare the bed, perform the top. Our live looping workflow covers the structure, and it adapts straight across.
• Drone and texture. Latched sustained sounds with slow parameter movement. One switch can hold a drone that does something interesting for a minute. The drone sustain rig is built on exactly this idea.
• Percussion over a bed. One or two voices played rhythmically against a running loop. Musically complete, low input count.
• Scene and section launching. The player conducts the arrangement rather than playing every note. In an ensemble context this is a real and respected role.
• Ambient and generative. The switch nudges a system that is already producing sound. Ambient soundscapes covers the patch design.

Open Up Music, which runs the National Open Youth Orchestra, has done a lot of thinking about what it means to write parts that are genuinely playable on accessible instruments rather than simplified versions of existing parts. That distinction — composing for the instrument versus reducing an existing part — is the right frame for switch-based music too.

Where does switch-driven MIDI run out of road?

Anywhere the music needs simultaneous continuous control, fast polyphony, or nuance in velocity.

Being honest about this is more useful than selling past it:

• Velocity. A switch is binary. It has no force sensing, so every note arrives at the same velocity unless something else varies it. You can randomise velocity slightly in the DAW to avoid machine-gun uniformity, but that is a cosmetic fix, not expression. An analog trigger — covered in our adaptive trigger velocity curves piece — is the real answer if any pressure-sensitive input is viable.
• Simultaneous continuous control. Two parameters moving at once needs two axes, and switches do not provide them. A joystick into an XAC USB port, or any analog stick, does.
• Fast polyphonic playing. Chords from switches means one switch per chord, prepared in advance. That is a legitimate way to make music but it is not keyboard playing.
• Anything where the barrier is not actuation. If the obstacle is seeing the screen, reading the plugin, or operating a mouse, switches do not touch it. The screen-reader workflow addresses a different problem entirely.

Where a switch rig genuinely is not the answer, the field has better ones. Drake Music's accessible musical instrument collection documents a decade of purpose-built instruments. The OHMI Trust competition funds adaptations of conventional orchestral instruments so a disabled player can take an ordinary seat in an ordinary ensemble. Soundbeam has built movement and proximity-triggered instruments for decades. None of these are competitors to a gamepad bridge in any meaningful sense; they solve different problems and they solve them better.

Who should you ask for help?

An assessment service first, a disabled musicians' network second, and a software guide like this one a distant third.

SpecialEffect assesses people one to one and loans equipment free of charge in the UK. AbleGamers runs peer counselling and publishes a knowledge base in the US. Drake Music works specifically on music with disabled musicians and its resources index is the most useful starting point in the field. We link to them because they know more than we do about the parts of this that are not software. None of them has endorsed this product.

FAQ

Can you make music with a single switch?

Yes, but the single switch has to control a process rather than play notes directly. The usual pattern is to have the DAW or a sequencer running a loop or a scanning selection, and the one switch confirms, advances or triggers at a chosen moment. One-switch music making is well established in the accessible technology field and OneSwitch.org.uk has documented approaches to it for over twenty years.

How many switches do you actually need for a usable music rig?

Two is the point where a rig stops feeling constrained and starts feeling playable, and four to six covers most practical production work. One switch is viable with the right software design. Beyond about eight switches the limiting factor is usually the player remembering the layout rather than the hardware.

What is the difference between momentary, toggle and latch?

Momentary means the note sounds only while the switch is held and stops on release. Toggle means the first press turns something on and the second press turns it off. Latch means one press starts a sustained note that keeps sounding until a separate stop command arrives. Toggle and latch both exist to remove the need for a sustained physical hold.

How do you avoid a switch firing twice on one press?

Use a debounce window, which is a short period after each trigger during which further presses on that input are ignored. Around 60 ms is a reasonable starting point for mechanical chatter. If the double triggers come from hand movement rather than the switch itself, a longer window of 120 to 200 ms usually settles it, at the cost of limiting how fast that input can be played.

Do I need the Pro version to build a switch-accessible mapping?

For a custom layout, yes. The free tier covers one controller, one MIDI port and three saved presets with the stock mapping, which is enough to confirm your switches reach the computer. Assigning specific notes and control changes per input, and setting momentary, toggle, latch or repeat behaviour, happens in the mapping editor, which is a Pro feature.

Is a gamepad bridge better than a dedicated accessible instrument?

Not better, different. A dedicated accessible instrument such as those documented by Drake Music or funded through the OHMI Trust is designed around a specific musical purpose and will usually beat a repurposed gamepad at that purpose. A gamepad bridge wins on cost, availability and the fact that many people already own and already know the device. Both are valid and many players use both.

Try it with the switches you already have

If any device in the room registers as a gamepad, you can test this today. Download Universal Controller MIDI, confirm your switches reach the computer on the free tier, then design the layout around the input count rather than around a keyboard.