What is MPE?

MPE stands for MIDI Polyphonic Expression. It is an extension of MIDI 1.0, formally adopted by the MIDI Association in 2018, that lets every note in a chord bend, slide and change pressure independently. Ordinary MIDI applies pitch bend to a whole channel at once, so bending one note bends the entire chord. MPE gives each note its own channel, and with it its own pitch, its own timbre and its own pressure.

In practice that is the difference between a keyboard and an instrument. A guitarist bends the third of a chord and leaves the root alone. A singer scoops into one note in a harmony. A violinist adds vibrato to the top voice only. None of that is expressible in plain MIDI. All of it is ordinary in MPE.

The problem MPE solves

MIDI 1.0 was designed in 1983 around a simple idea: one channel is one instrument. Pitch bend, modulation and aftertouch are channel messages, and they affect every note sounding on that channel.

Hold a C major triad and push the pitch wheel. All three notes move together. There is no message in standard MIDI that says "bend only the E." For four decades that was simply a limit everyone worked around, usually by splitting parts onto separate tracks and playing them one at a time.

How MPE gets around it

MPE does not add new message types. It changes how the existing sixteen channels are used:

• One channel per note. Each note you play is assigned its own MIDI channel from a pool. Channel messages sent on that channel now affect exactly one note.
• A manager channel. One channel is reserved for messages meant for the whole instrument: sustain pedal, global pitch wheel, program changes. In the common "lower zone" layout that is channel 1, with channels 2 to 15 carrying voices.
• Three dimensions per note. The specification names them by axis: pitch bend (X, sliding left and right), timbre on CC74 (Y, moving forward and back), and pressure (Z, pressing harder).
• A wider default bend range. MPE member channels default to plus or minus 48 semitones, four octaves, so a note can travel a long way without running out of resolution.

Because it is built entirely from MIDI 1.0 messages, MPE travels down any cable, port or plugin path that already carries MIDI. Nothing about the transport had to change.

Do you need a special controller?

This is where most explanations stop, and it is the reason a lot of people read about MPE and quietly give up.

MPE grew up alongside a generation of expressive hardware: the ROLI Seaboard, the LinnStrument, the Haken Continuum, the Expressive E Osmose, Ableton Push 3, the Erae Touch. They are wonderful and they are expensive, and every article about MPE seems to assume you own one.

Drawing has advantages a controller does not. A performance is a take: if the vibrato drifts on bar three you play it again. A drawn curve is a document. You move one point, adjust one bezier handle, and the rest stays exactly as it was. And there is no ceiling on what your hands can do, so a bend can be slower, wider or more precise than anyone could play.

If you do own an MPE controller, drawing and playing are not exclusive. MIDISketch records what you play as editable curves, so a take becomes something you can fix a single point of rather than re-perform.

What plays MPE?

Synths

Most modern soft synths support MPE, usually behind a switch you have to find: Serum 2, Vital, Pigments, Phase Plant, Massive X, Omnisphere, Diva, Surge XT, Equator, Bitwig's built-in instruments and many more. The catch is almost always the same: the synth also needs its pitch bend range set to match, or your carefully drawn four-octave dive comes out as a wobble.

Synths without MPE are not excluded. MIDISketch falls back to a single global bend on channel 1, drawn exactly the same way, so a monophonic line still moves the way you drew it.

DAWs

Bitwig Studio and Ableton Live 11+ have native MPE editing. Cubase and Logic Pro support it. Reaper supports it with configuration.

FL Studio has no native MPE support at all, which is worth stating plainly because it is the most-used DAW in a lot of the world. Hosting a plugin that generates MPE internally and plays a synth loaded inside it is the way FL users get there. See the compatibility page for what to set where.

MPE and MIDI 2.0

MPE is often described as a stopgap. It is more accurate to call it the proof.

When MPE was ratified in 2018, per-note expression was a thing a handful of instrument builders wanted and the standard could not express. MPE got it working inside MIDI 1.0's existing structure, by spending channels. It worked, the hardware sold, the synths added support, and per-note expression stopped being exotic.

MIDI 2.0, adopted in 2020, makes it native. Per-note controllers are first-class messages rather than a channel trick. Resolution goes from 7-bit, the 128 steps that have quantised every MIDI controller since 1983, to 32-bit. Devices negotiate capabilities with each other in both directions instead of guessing. The channel-per-note workaround becomes unnecessary because the protocol finally says what MPE was working around.

The transition will take years. MIDI 2.0 needs operating system support, host support, device support and a lot of updated software, and MIDI 1.0 is not going anywhere in the meantime. MPE is what works today, on hardware you own, in software you already run.

Where MIDISketch sits

MIDISketch stores a note's pitch as a continuous curve at full floating-point precision, not as MIDI values. The 14-bit pitch wheel messages your synth receives are generated from that curve at the moment of output. The curve is the truth; MPE is one way of expressing it.

That matters for what comes next. Emitting MIDI 2.0's higher-resolution per-note controllers is a new output stage reading the same curves, not a rebuild of how the instrument thinks. The data model was already more precise than MIDI 1.0 could carry.

Common questions

Is MPE the same as MIDI 2.0?

No. MPE is an extension of MIDI 1.0 from 2018 that achieves per-note expression using existing channel messages. MIDI 2.0 is a new protocol from 2020 with native per-note controllers and far higher resolution. MPE works today on everything; MIDI 2.0 is still arriving.

Does MPE work in FL Studio?

Not natively. FL Studio has no MPE support in its piano roll. The practical route is a plugin that generates MPE internally and hosts your synth inside it, which is what MIDISketch does.

Do I need an MPE keyboard to use MPE?

No. MPE is a data format. You can draw it, record it from a controller, or extract it from audio. A controller is one way in, not the only one.

Why does my MPE bend sound too small?

Almost always a pitch bend range mismatch. MPE member channels default to plus or minus 48 semitones, and if your synth is set to the traditional plus or minus 2, a four-octave drawn dive arrives as a whole tone. Set both to the same number.

Can I use MPE with a synth that does not support it?

Yes, with a limit. Without MPE the synth cannot bend notes independently, but a single global pitch bend on channel 1 still works, so monophonic lines and single-voice parts behave correctly.

Does MPE work over a normal MIDI cable?

Yes. MPE is built from ordinary MIDI 1.0 messages, so it travels over standard 5-pin DIN, USB MIDI, virtual ports and plugin routing without any special hardware.

Draw it, do not perform it

MPE is how the sound gets to your synth. The question worth asking is how the expression gets into the MIDI in the first place, and the usual answer, buy a thousand-dollar controller and learn to play it, is not the only one.

MIDISketch is a piano roll where the pitch of a note is a line you draw. Bend, slide and vibrato go straight onto the note itself, with points and bezier handles you can move afterwards, sent to any synth as MPE or as plain MIDI 1.0, standalone or as a plugin.