A MIDI controller is a machine for faithfully transmitting everything your hands do. That is exactly what you want until some of what your hands do is involuntary. If you live with tremor — essential tremor, tremor associated with a neurological condition, medication-related tremor, or anything else — the default behaviour of a gamepad is to amplify it: double triggers, flickering on-off states, and automation lanes full of oscillation that nobody played.
The configuration goal is simple to state and fiddly to achieve: let deliberate movement through and stop involuntary movement reaching the music. This guide covers the specific settings that do that — re-trigger lockouts, deadzones, hysteresis, output smoothing and quantisation — how to tune each without over-applying it, and where the approach genuinely does not work.
One thing up front, because it matters more than anything else here. This is not clinical advice and this is not a medical device. It is a guide to software settings. Anything about tremor itself belongs with a clinician, and anything about adapting equipment for a specific person is best taken to an assessment service or an occupational therapist.
- Buttons: re-trigger lockout of 150 to 300 ms stops unintended repeats.
- Axes: wider deadzone plus hysteresis stops flickering at thresholds.
- Continuous controls: low-pass smoothing removes oscillation at the cost of some lag.
- Timing: quantise in the DAW rather than asking the hands to supply precision.
- Design: prefer absolute targets and stepped values over fine relative hunting.
- Free vs Pro: nearly all of this lives in the Pro mapping editor. Free gets one controller, one port, three presets on the stock mapping.
- Not medical advice. Not a treatment. Not a medical device.
What does tremor look like in a MIDI stream?
As unwanted repeated note-on events on buttons, and as rapid oscillating control change values on sticks and triggers.
Before tuning anything, look at the actual data. Open a MIDI monitor — most DAWs have one, and standalone utilities exist on every platform — and play normally for a minute. You are looking for three patterns:
| Pattern | What it looks like | Primary fix |
|---|---|---|
| Repeated triggers | Two or three note-ons from one intended press, typically 60 to 250 ms apart | Re-trigger lockout on that input |
| Threshold flicker | A rapid on-off-on-off burst when a value sits near a boundary | Hysteresis |
| Axis oscillation | A control change value wobbling continuously by a few units | Deadzone plus output smoothing |
Note the measured gap between unwanted repeats, because that number tells you where to set the lockout. Guessing produces either a setting that does not help or one that cripples the input. Thirty seconds with a MIDI monitor saves an hour of frustration.
How do you stop a button firing more than once?
With a re-trigger lockout — a window after each trigger during which that input cannot fire again.
This is the same mechanism as debounce, but tuned for a different cause. Mechanical contact bounce happens in a handful of milliseconds; involuntary repeat presses happen over a much longer window, often 80 to 250 ms. A 60 ms debounce designed for contact bounce does nothing about the second.
| Lockout | Stops | Caps that input at |
|---|---|---|
| 60 ms | Mechanical chatter only | About 16 presses per second |
| 150 ms | Most involuntary repeats | About 6 presses per second |
| 250 ms | Nearly all involuntary repeats | 4 presses per second |
| 400 ms | Everything, including intended fast playing | 2.5 presses per second — usually too slow for rhythm |
Set the lockout per input rather than globally. A button used for scene launch can take a 300 ms lockout with no musical cost at all. A button used for sixteenth-note hi-hats at 120 BPM needs to fire eight times a second, so a 150 ms lockout would break it — and in that case the honest answer is that the hi-hat part should be sequenced rather than played live, or moved to a different, more reliable input.
Lockouts live in the mapping editor, which is a Pro feature. The switch-accessible triggering guide covers the related debounce settings for switch hardware.
What stops a control flickering on and off at a threshold?
Hysteresis — setting a higher threshold to turn something on than to turn it off.
Any time a continuous value is converted into an on-off state, there is a boundary. If the value sits near that boundary and wobbles, the state flips rapidly and produces a burst of unwanted MIDI. This is one of the most disruptive tremor-related artefacts because it can produce dozens of events per second.
Hysteresis separates the two thresholds. For example, a trigger that turns a note on at 60 percent travel might only turn it off below 40 percent. A wobble of a few percent at the boundary now does nothing at all, because the value has to travel a meaningful distance to cross back.
Tuning guidance:
- Start with a gap of about 15 to 20 percent between the on and off thresholds.
- Widen it until the flicker stops. The cost is that the control feels less immediate near the boundary.
- Do not use hysteresis on anything where the exact transition point is musically important, because by definition it makes that point ambiguous.
The same idea applies to stick-to-note mappings and to any threshold-based trigger. Our stick drift guide covers the related deadzone calibration, which solves an adjacent problem — unwanted output when nothing is being touched at all.
How much smoothing should a continuous control have?
The least that removes the visible oscillation, because every increment of smoothing adds lag.
Smoothing is a low-pass filter on the output: instead of sending the current value, the bridge sends an average of recent values. Rapid oscillation averages out to a steady line. Slow deliberate movement passes through largely intact.
The cost is latency, and it is unavoidable — averaging over a window means the output lags the input by roughly that window. Light smoothing is imperceptible. Heavy smoothing feels like the control is made of treacle.
A practical approach:
- Apply smoothing only to the axes that need it, never globally.
- Raise it in small steps while watching the automation lane in the DAW, not while listening — the eye catches residual oscillation faster than the ear.
- Stop as soon as the lane looks clean.
- Then play a fast gesture and check the lag is acceptable. If it is not, you have over-smoothed and should fix the problem a different way.
For parameters where a small wobble genuinely does not matter — reverb size, a slow filter sweep, a pad's cutoff — you can be generous. For anything percussive or rhythmic, smoothing is the wrong tool entirely and the function should move to a button with a lockout instead.
Can quantise do some of this work?
Yes, and it is the single most effective change for timing, because it removes precision from the list of things the hands have to supply.
Input quantise snaps recorded notes to a grid as they arrive. A note played 40 ms late lands on the beat anyway. This is standard DAW functionality, it is not an accessibility feature, and it has been used by producers of every kind for decades — but on a tremor-affected rig it does a specific and important job: it means timing accuracy is no longer a physical requirement.
Settings worth understanding:
- Grid resolution. A coarser grid — eighth notes rather than sixteenths — gives more tolerance. Choose the coarsest grid the music can live with.
- Quantise strength. Most DAWs let you quantise partially, moving notes a percentage of the way to the grid. 100 percent is machine-tight; 70 to 85 percent keeps some human feel while still fixing large errors.
- Record quantise versus post quantise. Quantising on input means what you hear back is corrected immediately, which is better feedback. Quantising afterwards keeps the original performance recoverable.
Our live drum machine guide and the finger drumming walkthrough both cover quantise settings in a general production context, and the settings are the same ones here.
How should the mapping itself be designed differently?
Favour absolute targets and stepped values over fine relative adjustment, because hunting for an exact position is the hardest thing to do with tremor.
This is a design point rather than a settings point, and it probably matters more than any filter. Consider two ways to set a mix level to a specific value:
- Relative hunting. Move a stick, watch the number, adjust, overshoot, adjust back. Every adjustment requires fine motor precision and the overshoot cycle can go on for a while.
- Stepped increments. Two buttons, each moving the value by a fixed amount. Each press is a coarse action with a guaranteed result. Reaching an exact value takes several presses but no precision at all.
Stepped increments are far easier on a tremor-affected rig, and the same logic applies throughout: snap-to-value mappings, discrete scene launch rather than scrolling, preset recall rather than dialling in a sound. Design so that the required precision lives in the software rather than in the hand.
The touchpad deserves a mention as a counterexample. It requires no force, which is good, but it demands quite fine positional control, which may not be. Test it rather than assuming either way. The touchpad XY guide covers the mapping if it does suit.
Where does this approach stop working?
Where tremor amplitude exceeds what filtering can separate from intended movement, and where the musical goal requires fine real-time precision.
Honest limits, because overselling this helps nobody:
- Filtering cannot distinguish intent. Smoothing removes fast movement. If deliberate movement is also fast, it gets removed too. Where tremor and intended gesture occupy the same speed range, no filter will separate them cleanly.
- Real-time expressive control may not be viable. A live filter sweep that needs to land exactly on a beat is a precision task. Automating it, or triggering a pre-drawn automation shape from a button, achieves the same musical result without the precision requirement — and that is a perfectly legitimate way to make music.
- Some rigs need different hardware entirely. A larger input surface, a switch with a guard around it, a mounted controller that cannot be knocked, or an entirely different interface may be the answer. SpecialEffect and AbleGamers both work on exactly these problems and will have seen variations of them many times.
- Fatigue changes everything. Settings tuned at the start of a session may be wrong by the end. Having two presets — one for fresh, one for tired — is a reasonable adaptation, and the free tier's three preset slots accommodate it.
Where a gamepad is not the right instrument, the accessible instrument field has better options. Drake Music's accessible musical instrument collection documents purpose-built designs, Soundbeam has built movement-triggered instruments for decades, and the OHMI Trust funds adaptations of conventional instruments. None of these organisations is affiliated with this product.
What about recording and editing rather than playing?
Shifting work from real-time performance to editing removes the precision requirement entirely, and there is nothing second-rate about it.
A substantial amount of recorded music has always been assembled rather than performed. Step sequencing, drawing automation, editing a take, comping across several passes — these are standard professional techniques, not workarounds. If real-time playing is the hard part, moving more of the work into the editing stage is a legitimate and often better answer than fighting the input.
A gamepad is still useful in that mode. Transport control, undo, scrolling through a take, nudging a selected note by a grid unit, auditioning a loop — all of those are discrete button presses with no precision requirement, and having them under the fingers rather than on a mouse is a real ergonomic win. Our Reaper setup guide and the Logic Pro control surface guide both cover binding editing commands to pad inputs.
What should a starting configuration look like?
Conservative on the inputs that are causing trouble and untouched everywhere else, so you can tell what each change is doing.
A reasonable first pass, assuming a standard DualSense or Xbox pad and a general production workflow:
| Input | Setting | Value to start at | Why |
|---|---|---|---|
| Face buttons used for notes | Re-trigger lockout | 150 ms | Stops most involuntary repeats, still allows roughly six presses per second |
| Transport and navigation buttons | Re-trigger lockout | 300 ms | These are never played rhythmically, so a long lockout costs nothing |
| All sticks | Centre deadzone | Wider than the default | Stops involuntary movement reaching the output when the stick is held |
| Sticks driving continuous controls | Output smoothing | Light, then raise | Cleans automation lanes without obvious lag |
| Triggers used as on-off | Hysteresis | 15 to 20 percent gap | Stops on-off flicker at the threshold |
| Panic input | Re-trigger lockout | Minimal | This one should always respond immediately |
| DAW | Input quantise | Eighth notes, 85 percent strength | Removes timing precision from the physical requirements |
Save that as a named preset before touching anything else, so there is a known-good state to return to. The free tier's three preset slots are enough to hold a baseline, a current configuration and one experiment, which is exactly the rotation you want while tuning.
Then change one value at a time, play a real part, and keep or revert. This is slower than adjusting five things at once and it is the only approach that tells you which setting was responsible. Our mapping templates guide covers naming and exporting presets so a configuration you have spent an evening tuning is not lost to an accidental overwrite.
FAQ
Can a gamepad be set up to work with hand tremor?
Often yes, using a combination of re-trigger lockouts on buttons, wider deadzones and hysteresis on axes, output smoothing on continuous controls, and timing quantisation in the DAW. None of these remove tremor; they stop it reaching the musical output. How well it works varies enormously between individuals and between types of tremor, so the honest answer is that it is worth trying and not guaranteed.
What is hysteresis and why does it help?
Hysteresis means the threshold to switch something on is different from the threshold to switch it off. Without it, a value hovering exactly at a boundary flickers rapidly between states. With it, the value has to move meaningfully past the boundary in each direction before anything changes, which converts a stream of unwanted on-off events into one clean transition.
Does smoothing add latency?
Yes, and that is the trade-off. Smoothing works by averaging recent values, so the output lags behind the input by roughly the length of the window being averaged. Light smoothing is usually imperceptible, heavy smoothing is clearly laggy. Apply the least that removes the jitter, and apply it only to the continuous controls that need it rather than globally.
Should the smoothing happen in the controller software or the DAW?
Both have a place. Smoothing continuous control output is best done close to the source, because once a jittery stream is recorded into an automation lane it is harder to clean. Timing quantisation is best done in the DAW, because DAWs have mature, musical quantise implementations with swing, strength and grid options that a controller bridge does not.
Do these settings need the Pro version?
Yes for most of them. Re-trigger lockouts, deadzone and hysteresis settings, axis smoothing and custom per-input behaviour all live in the mapping editor, which is a Pro feature. The free tier covers one controller, one MIDI port and three saved presets on the stock mapping, which is enough to see what the raw input looks like before deciding whether the tuning is worth buying.
Is this medical advice or a treatment for tremor?
No. This is a guide to configuring music software and nothing in it is clinical advice, a treatment, or an assessment. Universal Controller MIDI is not a medical device. Questions about tremor itself belong with a qualified clinician, and questions about adapting equipment for an individual are best taken to an assessment service such as SpecialEffect or AbleGamers, or to an occupational therapist.
Look at the data first
Open a MIDI monitor before you change a single setting — it will tell you in thirty seconds what an hour of guessing will not. Download Universal Controller MIDI, watch what the raw input actually sends, and tune the specific thing that is causing trouble.