You are here

The ABC Of MPE

An Introduction To MIDI Polyphonic Expression By Rory Dow
Published June 2021

The ABC OF MPE

You’ve probably heard of MPE, but what exactly is it and what can it do for you?

You’d be forgiven for thinking that MIDI had largely remained unchanged since its inception and widespread adoption back in the mid‑1980s. Indeed, the core functions of MIDI have remained steadfast, meaning that a synth made back in 1985 can still be played by a DAW released last month. MPE, or MIDI Polyphonic Expression, became part of the official MIDI specification back in 2018 and has been a success, with both hardware and software manufacturers taking full advantage of the additional features it brings. Now that MPE has had time to mature, it seems like a good point to take a look at the state of the market, what’s on offer and what it can do for you.

What Problem Does MPE Solve?

MIDI Polyphonic Expression uses standard MIDI messages to enable features that standard MIDI doesn’t support. Specifically it allows for ‘polyphonic expression’ — per note automation of synthesis parameters.

Every keyboard player knows that when you use the pitch wheel or the modulation wheel, it affects all notes playing at once. That’s because, within the MIDI language, expression information such as pitch‑bend or modulation is not tied to individual notes but rather to a ‘channel’ or instrument. So, when you apply pitch‑bend, it affects the whole instrument. Emulating stringed instruments like guitar or violin, where you can apply vibrato to an individual string, is impossible. Enter MPE...

If we think of a standard piano keyboard, most MPE controllers add three new forms of polyphonic expression. The first is pressure, a force exerted onto the key much like aftertouch. This isn’t a totally new concept as Polyphonic Aftertouch is in the original MIDI specification, but it was not widely adopted. Most MPE controllers use ‘continuous pressure’ which is always active, unlike aftertouch which is only activated by further pressing into a key (see ‘Continuous Pressure vs Aftertouch’ box).

The second new expression gesture is horizontal movement, mostly used to generate polyphonic pitch‑bend for vibrato. Some MPE controllers also allow you to slide along the entire length of the keyboard. There is something wonderful about sliding from one chord shape to another without lifting your fingers from the surface. This is one of the first techniques I demonstrate to people who are new to MPE and it never fails to impress (I refer to the possibilities, not my playing technique!).

The third new expression is vertical movement, sliding a finger up and down a key, which can be used for a wide variety of timbral changes to a sound. It might, for example, be used to crossfade between a string and a brass ensemble, or to send individual notes to a reverb or delay. The potential for sound design is as varied as the synthesis engine allows.

All this additional per‑note expression can, when implemented and played well, lead to astonishing levels of expressivity which are simply not possible on a traditional MIDI keyboard. They do take time to master even for an experienced piano keyboard player.

How Does MPE Achieve This?

To understand how this works we must understand MIDI channels. They were put into the original MIDI spec to allow one MIDI cable to carry information meant for several different instruments. Every MIDI port or cable can carry 16 channels and every MIDI message is given a channel number. Using this method, MIDI allows you to use several synthesizers connected to a single MIDI port. Each synth receives the same MIDI information, but only responds to information on the channel to which the synth is set.

MPE repurposes MIDI channels as notes instead of instruments. This means you can only have a single instrument on each port or cable and the instrument is limited to a maximum of 16‑note polyphony (one note on each of the 16 channels). Each note can have its own pitch‑bend, mod wheel and other expression messages, meaning you can now emulate the single‑string pitch vibrato of a guitar or violin without having the vibrato applied across the whole instrument. It’s not restricted to emulating real acoustic instruments of course. The possibilities for expressive sounds of any kind are vast.

MPE therefore is a tradeoff. It trades multitimbrality — being able to use different sounds or instruments in the same MIDI stream — for polyphonic expression, because each instrument effectively becomes a single note. In a DAW environment, this tradeoff isn’t a problem. DAWs allow you to record as many different MIDI streams and use as many virtual instruments as you need (CPU allowing). In the hardware world, it may be more of a compromise as you might need more physical MIDI ports depending on how many physical instruments you have.

Some instruments like the Deckard’s Dream from Black Corporation have supported MPE from the word go...Some instruments like the Deckard’s Dream from Black Corporation have supported MPE from the word go...

Will MPE Work With My DAW?

Generally yes, although there are caveats. Thanks to the invested hard work of companies like ROLI, Roger Linn, Apple, Moog Music, Yamaha and others, the MPE specification was not just a theoretical improvement to MIDI, but a necessary one propelled by the design and manufacture of real products. So when MPE became an official thing, there were immediately hardware controllers and software synthesizers able to take advantage of it.

DAWs took a little more time to catch up. Some were already able to record and play back multichannel MIDI streams without a problem, which means recording and playback of MPE, but some have taken much longer to offer this and a few still don’t without seriously clunky workarounds (Pro Tools, I’m looking your way).

Recording and playback of MPE, once implemented, tends to work well. However editing can be more problematic. The reason for this lies with the way that MPE controllers use MIDI channels for each note. Channels are dynamically allocated by the controller when you play it — if you add a note to a chord, it will uses a spare, currently unused channel for that note. Most DAW piano‑roll editors don’t have this dynamic capability to calculate which channels are in use at any particular time and so something as simple as adding an extra note to a previously recorded performance can break the playback.

Another troublesome area is quantisation, or just moving notes. If you record an MPE performance unquantised, then decide you want to shift some notes forwards or backwards in time, the DAW has to be aware of the polyphonic information that belongs to that note, and shift it alongside. If it doesn’t, your carefully played vibrato will be out of time with the note.

Different DAWs offer different levels of MPE editing functions. The clever DAWs will allow you to move a note and have all the polyphonic channel info move with it. You may also be able to edit the polyphonic MIDI, delete notes, and have any associated polyphonic expression deleted as well. But if you start adding notes, copy, paste or move stuff more dramatically, it is very easy to end up with glitchy, broken playback.

It is best therefore to approach MPE very much as a performance tool. If you don’t nail your take on the first recording, re‑record it until you do. This will save you from editing headaches later on.

Others, like the DSI OB‑6, have been updated to include MPE compatibility.Others, like the DSI OB‑6, have been updated to include MPE compatibility.

Controllers

MPE is nothing without a suitable controller, which are available in many shapes and sizes, all with their respective merits. Here are a few to consider.

Eigenlabs Eigenharp

The Eigenharp was one of the first alternative MIDI controllers to embrace MPE. Its breath control, combined with its unusual hand‑crafted design, makes it unique. There are several different versions suited to different budgets.

Expressive E Osmose

Although still in development, the Osmose looks like being the first fully MPE compatible, mechanical synthesizer. The keys look and respond like piano keys, but are also capable of vibrato, horizontal expression and continuous pressure.

Haken Continuum Fingerboard

The Continuum is another single‑piece expressive surface. Although the Continuum can output MPE MIDI, it uses its own DSP sound engine and communication protocols to achieve a very fast, high‑resolution response.

Zivix Jamstik Studio MIDI Guitar

The Zivix Jamstick Studio MIDI Guitar outputs MPE data.The Zivix Jamstick Studio MIDI Guitar outputs MPE data.If you thought MIDI guitars were a thing of the past, think again. The Jamstik Studio is a hybrid guitar and MIDI controller that uses MPE to convert guitar performances into fully polyphonic MIDI data.

Joué Music Joué Play

Joué Play allows you to put different silicone overlays onto a rectangular MPE surface. This means it can be a drum pad one minute, keys or fingerboard the next.

Keith McMillen K‑Board Pro

The K‑Board is possibly the most traditional piano‑like controller on the market, with each key being made up of a separate silicone piece, unlike the ROLI Seaboard, which is one large piece.

Madrona Labs Soundplane

The Soundplane is a wooden grid‑based controller similar to the LinnStrument. It’s boutique, made in small batches, and rather beautiful.

ROLI Seaboard

ROLI Seaboard RiseROLI Seaboard Rise.

Comes in a variety of sizes and prices. It keeps the traditional chromatic piano layout so is popular with keyboard players. The single‑piece silicone rubber surface allows for anything from subtle vibrato to multi‑octave pitch sweeps, as well as continuous pressure.

Roger Linn Design LinnStrument

While the ROLI Seaboard is perhaps what many of us think of as an MPE controller, the reality is becoming ever more varied, for example this Roger Linn LinnStrument...While the ROLI Seaboard is perhaps what many of us think of as an MPE controller, the reality is becoming ever more varied, for example this Roger Linn LinnStrument...

The LinnStrument does away with traditional piano keyboard layout in favour of a grid system called ‘fourths string layout’, where a single note will be five semitones (a fourth) above the note directly below it (it can be reconfigured). The consistent layout means that, unlike a piano, interval distances and chord shapes are always the same.

Sensel Morph

Like the Joué Play, the Morph is a chameleon amongst MPE controllers. It is made up of a single rectangular MPE surface upon which you can attach different silicone overlays. There are overlays for pianos, drum pads, the Buchla Thunder, and even for gaming and video editing.

If you haven’t had a chance to play with an MPE controller, I highly recommend it. When it works right, it feels like a real musical instrument that responds to your touch in an organic way with no two performances sounding quite the same.

Summary

If you haven’t had a chance to play with an MPE controller, I highly recommend it. When it works right, it feels like a real musical instrument that responds to your touch in an organic way with no two performances sounding quite the same. After spending some time with an MPE instrument, returning to a standard keyboard can feel rigid and limited. On the other hand, MPE sounds can feel overwhelming. Every tiny movement of your fingers can cause sounds to slip and slide and that isn’t always the playing experience you want. Learning how MPE works can help you with sound design in order to add or remove gesture expression from your favourite sounds.

Not all MPE controllers have exactly the same features and not all instruments will respond to MPE in exactly the same way. Choosing an MPE controller is a personal thing. You may enjoy the squidgy silicone surface of the ROLI Seaboard, or perhaps the non‑traditional grid layout of the LinnStrument is more your thing. Either way, MPE can be an excellent way to add expressivity and nuance to your MIDI performances.

One common misconception is that MPE is only good for emulating ‘real’ instruments. It can equally be used to control a granular synthesizer, a modular system or even a lighting rig. Sound design can take on whole new dimensions, literally, and performing on a keyboard‑like MPE interface gets you closer to certain acoustic and electric sounds than ever before. In short, there are many reasons to try out MPE and there’s never been a better time.  

Continuous Pressure vs Aftertouch

There are aspects of playing an MPE device that can take even the best piano or keyboard players by surprise. One is the concept of ‘continuous pressure’, which refers to the instant pressure response which happens when you trigger a note. Instead of needing to push down further into the key to activate aftertouch, as you would on a standard MIDI controller, continuous pressure is always active even with the lightest touch.

Clearly, this instant response will give a much more immediate and natural interaction with the MPE surface, but it can take some getting used to. It also represents a challenge for sound designers, who must navigate a fine line between continuous changes in sound and playability. It is also one of the reasons why sounds designed for standard MIDI controllers will not always play well on an MPE controller.

MPE vs MIDI 2.0

MIDI 2.0 was officially introduced at the 2020 Winter NAMM show, and although there are very few hardware or software manufacturers adopting it yet, we do know what it will be capable of. MIDI 2.0 is an extension of the MIDI protocol, which includes MPE as a subset as well as many other enhancements. MIDI 2.0 will allow much greater communication between two devices. For example, it will allow a synthesizer to upload a complete list of preset names to a controller as well as information about every parameter. This will mean that a suitable MIDI 2.0 controller will be able to display preset names, parameter names and other information specific to that instrument.

In general, MIDI 2.0 is much higher resolution. For example, MIDI velocity is traditionally limited to 128 stepped values. MIDI 2.0 increases this to 65,536 possible values. It is only speculation at this point, but this increase in resolution across all MIDI messages should benefit any MPE‑capable controller or instrument that embraces it.

So whilst MPE and MIDI 2.0 are both enhancements to the original MIDI specification, MIDI 2.0 is far wider reaching and encompasses MPE, which is a very specific solution to a specific problem. For a full rundown of MIDI 2.0 see Nick Rothwell’s ‘Introducing MIDI 2.0’ article from the March 2020 issue.

The Master Channel

Part of the MPE spec includes provisions for a Master Channel. This is a reserved MIDI channel which is designated for messages which are common to all notes. For example, you might want to make use of a sustain pedal or a master volume control. Rather than sending out these messages across 16 different channels, it makes more sense to reserve one channel which is applied globally. Normally this is channel 1, with channels 2‑16 remaining for individual notes. This will of course reduce the total polyphony down to 15 voices or less.

Not all instruments use the master channel, and many MPE controllers will come with setup software in which you can reconfigure the master channel to suit instruments which may require a slightly different configuration.

Modular Adventures

MPE controllers are a great fit for the endless possibilities of modular synthesis. There are a few Eurorack modules from different manufacturers that can convert MPE MIDI streams into Gates and CVs, such as the Endorphin.es Shuttle Control, the Expert Sleepers FH‑2 or the Polyend Poly 2.

If software modular is more your thing, Softube’s Modular environment has a Rise/Grand modular designed to work directly with ROLI Seaboards (but can work with other controllers). The fantastic VCV Rack Eurorack simulator can also support MPE to CV, via the free moDllz MIDIpolyMPE plug‑in.