What Is MIDI, and How Does It Actually Work?

MIDI stands for Musical Instrument Digital Interface. It is a protocol that lets instruments, computers and studio gear send each other performance instructions, and the single most important thing about it is that it carries no sound at all. None. A MIDI cable is the most influential empty envelope in the history of recorded music, and it has been quietly running the charts since 1983.

What travels down the wire is a stream of short messages. This note started. It was hit this hard. Now it stopped. The mod wheel is at 64. Something on the receiving end reads those instructions and makes the actual noise. That split, instructions in one place and sound in another, explains almost everything people find confusing about MIDI, including the classic first-day panic where you plug in a brand new controller and hear absolutely nothing.

Data, not sound, and why that matters

The quickest way to understand MIDI is to put it next to an audio recording and look at what each one can do afterwards.

That last row is the catch, and it is worth dwelling on because it is the source of MIDI's slightly unfair reputation. If you are old enough to remember websites that played music at you, you will remember it sounding like a cheap keyboard having a bad day. The MIDI file was not the problem. The tinny sound module inside the computer was. Send those identical instructions to a decent instrument today and it sounds like whatever that instrument sounds like.

So MIDI is not a sound quality. It is not high fidelity or low fidelity. It is a set of directions, and directions are only as good as the venue you follow them to.

What is actually travelling down the cable

A handful of message types cover almost everything you will ever meet.

Note On and Note Off. Exactly what they sound like. Each carries a note number from 0 to 127, where 60 is middle C, and a velocity value, also 0 to 127, describing how hard the key was struck. Velocity is why a MIDI performance can breathe rather than sounding like a typewriter, and it is the first thing to check when a part feels mechanical.

Control Change, usually shortened to CC. The continuous stuff. CC 1 is the modulation wheel by convention, CC 64 is the sustain pedal, CC 7 is volume. There are 128 of them and most gear lets you point any knob at any number, which is how one controller ends up driving a plugin nobody designed it for.

Program Change. Switch to a different preset. One message, whole new instrument.

Pitch bend. Its own message type rather than a CC, because it needs finer resolution than seven bits allows.

Clock and transport. The messages that keep separate boxes in time with each other, plus start, stop and continue. This is the part that turns a pile of hardware into something that plays together, and we come back to it later.

All of this rides on sixteen independent channels. One cable, sixteen conversations, each addressed to a different instrument. It is also why a very large number of MIDI problems turn out to be a channel mismatch, which brings us neatly to the next section.

One number for the trainspotters: classic five-pin MIDI runs at 31,250 baud. That is slow by any modern standard, and it has been entirely adequate for over four decades, which tells you something about how carefully it was specified.

Connected, and nothing happens: five checks

This is the single most common MIDI experience in the world and almost nobody writes it down, so here it is in the order a shop actually works through it. Nine times out of ten it is the first one.

1. Your controller has no sound in it. A MIDI controller is a set of keys, pads and knobs that sends messages. That is the entire product. It has no sounds inside it, no speaker, and it is not broken. You need something to receive those messages: a plugin in your synthesizer software, or a hardware instrument. If you are still working out what that hardware would be, our plain English guide to synthesizers is the place to start.

2. Nothing is armed to receive. In your software, the track needs an instrument loaded, the track needs to be selected or record-armed, and the controller needs to be enabled as an input device. Three separate switches, and the software will cheerfully let you set two of them.

3. Wrong channel. Your controller is transmitting on channel 1, your instrument is listening on channel 2, and both are working perfectly. Set the receiver to omni, or set both ends to the same number.

4. In and Out are the wrong way round. MIDI Out of the sender goes to MIDI In of the receiver. Out to Out achieves nothing at all, quietly and without complaint. Thru is a third socket that passes on a copy of what came in, for daisy-chaining a third box.

5. Two USB devices staring at each other. Most controllers are USB devices, and so are most hardware instruments. Two devices cannot talk over USB without a host in the middle, which is normally your computer. Connect them directly with a USB cable and nothing happens, because neither one is in charge. This is what USB host boxes and old-fashioned DIN sockets are for.

There is a sixth, rarer one worth knowing. If a keyboard with its own sounds goes silent only when the computer is unplugged, look for a setting called Local Control and switch it back on. It was turned off so the keyboard would stop double-triggering, and then it was forgotten. I have made that exact phone call from a stage, twenty minutes before doors, and I would rather you did not.

Three ways MIDI gets from A to B

The five-pin DIN socket. The round one, formally DIN 41524, and the reason MIDI gear from 1985 still works with MIDI gear from this morning. It is optically isolated, which is a quiet blessing you only appreciate the first time a ground loop does not happen. Many audio interfaces still carry a pair, and if you are choosing one, our guide to picking the right audio interface covers which ones bother. A plain DIN lead such as the SSQ MIDI1 is 14 euro and will outlive several laptops.

USB. One cable for power and data, no interface required, which is why every controller under about 200 euro uses it. The catch is the host problem from check five above: brilliant with a computer, useless between two instruments.

The 3.5 mm TRS mini jack. Compact gear does not have room for a DIN socket, so manufacturers shrank it, and here is the part that catches people out. Different makers wired that mini jack differently, and the two wirings are not compatible. The MIDI Association eventually published a specification precisely because, in their own words, two devices may not have chosen the same scheme, so compatibility between devices is not known. You will see gear labelled Type A or Type B as a result.

The practical advice is short. Check the manual, use the adaptor that came in the box, and when you need a spare, buy the type your gear specifies rather than the one that physically fits. An adaptor cable like the ALM DIN to TRS lead at 23 euro solves it for 23 euro, which is a lot less than an afternoon.

MIDI 2.0, and whether you should care

Short answer: it is real, it is genuinely clever, and it is almost certainly not your problem this year.

The initiative was announced in January 2019 and the standard was introduced at the 2020 Winter NAMM show. The headline changes, per the MIDI Association's own description, are extended data resolution across channel voice messages, much better per-note control for expressive playing, and MIDI-CI, which lets two devices talk back and forth and agree on what they can both do. That last one is the structural difference. MIDI 1.0 only ever spoke in one direction and simply hoped for the best, rather like leaving voicemails.

The reason not to panic is written into the design. The MIDI Association states plainly that compatibility for translation to and from the MIDI 1.0 protocol was given high priority. Your existing gear is not being orphaned. Note also that the expressive playing most people actually want today is usually delivered by MPE, which runs perfectly well over MIDI 1.0 by using a separate channel per note. Clever, slightly cheeky, and it works.

So: buy the instrument you want to play. Do not buy a box because it has a 2.0 sticker, and do not avoid one because it does not.

Six pieces of gear, and what each one is for

All on our shelves, prices as of today, arranged by the job rather than by the price tag.

1. The sensible first one: Arturia MiniLab 3, 94 euro. Twenty-five keys, eight encoders, pads, and it fits next to a laptop. If you have read this far because you own no MIDI gear at all, this is the answer, and the amount of music made on this exact controller is faintly ridiculous.

2. The one that talks to hardware: Arturia KeyStep 37 mk2, 175 euro. The important one for this article. Alongside USB it has proper DIN MIDI and CV/gate, so it drives a computer, a modern synth and a modular rack without asking which decade they came from. If your plan involves anything with knobs on it, start here rather than at number one.

3. The studio centrepiece: Arturia KeyLab Essential 61 mk3, 223 euro. Sixty-one keys is the point where you stop transposing octaves mid-take and start playing with both hands like an adult. The obvious desk-centre choice.

4. The deep integration one: Native Instruments Kontrol S61 MK3, 693 euro. Where the controller stops being a keyboard and starts being a control surface, with screens that show you the instrument you are actually holding. Worth it if you live inside one software ecosystem, overkill if you do not.

5. The one that keeps hardware in time: Elektron Digitakt II, 868 euro. This is where MIDI clock stops being a bullet point. A box like this can sequence and synchronise the rest of your gear over MIDI, so a room full of instruments starts and stops as one thing. It is also a superb sampler in its own right, which does not hurt.

6. The far end: Expressive E Osmose 49 CE, 963 euro. A keyboard whose keys respond to pressure, wiggle and after-touch per individual note, which is precisely the expressive territory MPE and MIDI 2.0 exist to carry. Play one and the phrase "MIDI feels robotic" quietly retires. This is a serious instrument rather than a first purchase, and we would rather say so than sell you one by accident.

A controller is one of the few things in a studio you genuinely cannot judge from a photograph, because what you are buying is how the keys feel under your hands and nothing written down has ever captured that. Our Warsaw showroom on Nowogrodzka has these set up and connected, so you can play a MiniLab, a KeyLab and an Osmose one after the other and feel the difference in about ninety seconds. Or take one home on a free 30-day Test-Drive: a short contract, real sessions in your own room, then you decide. Everything carries a 36-month warranty, 60 months on PMC and ADAM Audio, there is a 30-day return window, and we ship to every EU country.

Straight answers

Forty-odd years on, a standard designed so that two rival manufacturers could stop sabotaging each other is still the thing holding your studio together. Ikutaro Kakehashi of Roland and Dave Smith of Sequential showed it working at NAMM in 1983 by connecting a Prophet 600 to a Jupiter-6, and the industry has been quietly grateful ever since, formally so in 2013 when both men collected Technical Grammys for it. So the next time your controller sits there in silence, remember that it is not sulking. It is sending perfectly good instructions to nobody in particular, which, as my partner will tell you, is also my main hobby at dinner parties.

About the author

Michał Kaniuszkiewicz is a content manager and SEO specialist at Wired Tunes. He creates and optimizes content that helps readers find useful information about music, gear, and sound. Beyond his editorial work, he has been making music for 8 years in hip-hop, pop, and electronic genres — giving him insight into the subject not just from an SEO perspective, but from the inside, as a practicing musician and producer. This experience allows him to create content that's equally useful and clear for both beginners and experienced listeners and music creators.

Back to top ↑