If you have spent any time around electronic music in the last decade, you have probably seen the letters MPE written on the side of a controller and not thought too much about them. The acronym sounds like a feature bullet. It is not. It is a quiet but fundamental reframing of what an electronic instrument can do, and once you have spent serious time playing one, going back to a fixed-velocity keyboard feels like trading a touch screen for a number pad.
This post is the explanation I wish someone had given me when I first encountered MPE. It is the technical version, the musical version, and a working set of opinions about where it is genuinely worth the cost and where it is overkill.
What MPE actually is
MIDI Polyphonic Expression, abbreviated MPE, is a way of using MIDI in which each note that is sounding has its own continuous control over pitch, pressure, and timbre, independently of every other note. That single sentence contains the whole idea. The rest of the post is unpacking what it actually means in practice.
Standard MIDI, as most synths have implemented it for forty years, works on a per-channel basis. Pitch bend applies to all notes on the channel. Aftertouch applies to all notes on the channel. Continuous controllers apply to all notes on the channel. If you press one note hard and another softly, the synth knows about the per-note velocity, but everything that happens after the note starts is shared. You cannot bend one note up while letting another sustain. You cannot apply vibrato to the top voice of a chord without applying it to the bottom too. The instrument is, in a real sense, monophonically expressive.
MPE solves this by spreading notes across multiple MIDI channels. The convention, formalised in the MMA’s MPE specification, is that one channel acts as a “master” carrying global messages and the other channels each carry one note at a time, each with its own pitch bend, pressure, and timbre control. The synth on the receiving end allocates each incoming MIDI channel to a voice, and now per-voice expression is a first-class citizen.
The effect, when you encounter it for the first time, is not subtle. You can play a chord and bend one note. You can hold a long note and have it gain pressure while the next note plays cleanly underneath. You can shape a phrase with the kind of nuance that a string player or a saxophonist takes for granted, and that electronic instruments have spent decades being unable to deliver.
Why this is bigger than it sounds
The word “expression” is a marketing favourite, which is part of the problem. It tends to slide off the brain. It is worth saying explicitly what is at stake.
Acoustic instruments are continuously expressive at the per-note level. A violinist is not making a discrete decision about a note’s loudness at attack and then letting it ring at that loudness. They are continuously shaping pitch, dynamics, and timbre across the duration of every note, independently for every string under their fingers. A pianist has fewer dimensions of control - velocity at attack, sustain pedal, una corda - but the basic unit of expression is still the individual note, and a chord is an event in which each voice has its own life.
Electronic keyboards have, until very recently, been built around the assumption that this is too hard to capture, so the per-note dimensions get collapsed into per-channel ones. The result is that synth performances tend to sound static in a specific way. The notes are perfectly in tune, perfectly aligned, perfectly identical in attack and release. The interesting variation has to come from outside the playing - from automation, from layered effects, from the synth’s own modulation engine - rather than from the player.
MPE puts the variation back where it belongs: in the player’s hands. It is not just adding control. It is restoring a kind of control that was always present in acoustic instruments and was lost in the transition to MIDI.
The hardware that makes it work
You cannot really evaluate MPE in the abstract. The experience is a function of the physical interface. There are roughly four families of MPE controller in 2026, and they feel meaningfully different. I keep an updated list of specific instruments in MPE controllers.
Soft-surface controllers like the ROLI Seaboard replace the keyboard with a continuous silicone surface. There are no discrete keys. You press into the surface, slide along it, lift off it. Pitch bend is the X axis, pressure is the Z axis, timbre is the Y axis. This is the most acoustic-feeling MPE controller in the sense that the playing technique is closest to a fretless instrument, and it is also the most disorienting if you come from a piano background.
Grid controllers like the LinnStrument lay out notes on a grid of pads, with each pad sensing X, Y, and Z independently. The grid is typically tuned in fourths, which gives you the same fingering across the whole instrument and lets you build chords and runs in ways a piano cannot. The pads are firm rather than squishy, and the response feels more like a percussion instrument than a string one.
Touch screens with haptic enhancement like the Sensel Morph and successors take a software-defined approach. The surface is a continuous pressure-sensing pad with overlays that define what you are playing. They are extremely flexible and tend to feel less committed than a dedicated MPE controller, which is sometimes good and sometimes the source of an uncanny-valley experience.
Hybrid keyboards like the Osmose and the more recent generation of MPE-aware traditional keyboards layer continuous control on top of a familiar weighted-key feel. You press a key and you get a normal note. You press harder, slide, or rock the key and you get continuous expression. These are the easiest to adopt for someone with a piano background and also the most compromised, in the sense that the per-note expression is constrained by what can be sensed through a key mechanism. If you are deeper in the traditional controller world, my comparison of the Arturia KeyLab and Native Instruments Kontrol covers the non-MPE side of that decision.
The right choice depends on what you already play. A pianist who wants more expression should probably look at a hybrid keyboard first. A guitarist or a string player will feel at home on a soft-surface controller. Someone coming from drum pads or finger drumming will find a grid controller most natural.
The synth side has to participate
An MPE controller is only half the picture. The synth on the receiving end has to actually use the per-note control data, or you are sending expression into a void.
In 2026, MPE support is no longer rare. Most major soft synths support it: u-he, Native Instruments’ Reaktor and Massive X, Arturia Pigments (covered in my Pigments 7 write-up), Ableton Wavetable and a growing number of stock instruments, Equator2 by ROLI which is purpose-built for MPE input. Hardware support is broader than it used to be too, with the Modal Argon8 family, ASM Hydrasynth, and the latest generation of Yamaha and Korg flagships all reading per-note expression. For a wider survey of the soft-synth landscape, see The best software synths of 2026.
The catch is that “supports MPE” covers a wide range of actual behaviour. The simplest implementation reads per-note pitch bend and pressure and routes them to the synth’s existing modulation system. A richer implementation lets you configure independent envelopes, filter responses, and modulation routings for each MPE dimension, so the playing nuance can drive sound design that simply was not possible on a per-channel synth.
The way you can tell the difference is to play a chord on the controller and hold pressure on one note. On a thin implementation, the chord swells but loses its internal balance. On a real implementation, only the held note responds, and the others sustain cleanly. That single test sorts most synths into the “really MPE” or “MPE-aware” buckets.
Where it is genuinely worth the cost
MPE is not a universal upgrade. There are styles of music for which it is overkill, and there are players for whom the learning curve is steeper than the reward.
It is genuinely worth the cost in:
- Lead playing on synth voices. This is where MPE shines unambiguously. A solo line over a track gains dimensions of expression that put it on a par with a saxophonist’s vibrato or a guitarist’s bends, with an instrument that does not require thirty years of practice to control.
- Pad and texture work. Long held notes that breathe, evolve, and shift timbre under pressure are an obvious MPE win. The ambient and cinematic communities have been the early adopters for exactly this reason.
- Acoustic emulation. Strings, winds, and bowed instruments have always sounded bad on standard MIDI keyboards because the per-note articulation is impossible to reproduce. MPE makes the simulation passable for the first time, and the same is true of the next-generation physical modeling synths where MPE input is the natural way to drive the model.
- Improvised and live performance. The amount of variation an MPE player can introduce within a single phrase is enormous, and live performance is where that variation matters most.
It is overkill for:
- Beat programming and tracker-style production. If you are programming notes into a piano roll one at a time, you do not need per-note expression because there is no playing happening. Velocity is enough.
- Bass lines that want to sit still. A bass line that is supposed to be metronomic and unmoving is actively harmed by adding expression dimensions you might accidentally activate.
- Tracks built around drum machines and groove boxes. The MPE benefit is in the playing. If your music does not have much playing in it, the benefit is small.
The honest answer is that MPE is a player’s feature. If your relationship to music is mostly through programming and production, it will feel like a curiosity. If your relationship to music is through playing, it is one of the most consequential improvements the industry has made to electronic instruments in twenty years.
What I would tell someone starting out
Buy the cheapest MPE controller you are willing to live with for six months. Borrow one if you can. The reason is that MPE is a thing you understand by playing, not by reading about. You will know within a few weeks whether it has changed your relationship with electronic instruments or whether it is a feature you can leave behind.
If it has changed your relationship, you will probably want to upgrade to something that fits your playing style more deliberately. If it has not, you have learned something useful about yourself as a musician, which is also worth the cost of admission.
The thing to avoid is the trap of treating MPE as a spec to compare. The data is in the playing. Spec sheets are useful for picking which controller to evaluate. Only the hours under your fingers will tell you whether MPE is for you.