Happy birthday MIDI 1.0: Slave to the rhythm

In part one, the main focus was on MIDI’s myriad data forms and how it began chattering to synths and drum machines, but it was the sequencer that really demonstrated MIDI’s brilliance. In fact, so brilliant it was, that musicians even began to think they were brilliant too by association. Brilliant! It’s early 1988 and it’s …

[bag o' spanners]

Let's not forget Doctor T's contributions to software and hardware. Doctor T's software presaged the Ableton/Launchpad trigger system by almost three decades, with their "trigger from keyboard" page. Up to 100 sequences could be fired in a variety of modes from a simple Atari ST. Admittedly, they could have made life easier by having a simpler recording interface. Dividing 768 clocks by your chosen meter usually required a calculator for anything but basic 4/4. They also made some sophisticated midi matrices, fader controllers, and event generators that could lock up pyros and lighting desks to midi clock.

Gigging with Ataris was fun, as long as the midi cables weren't placed under strain/tripped over by drunk musos, and you remembered to bring the right box of floppies.

The struggle in a big midi studio was locking up the various flavours of proprietary din sync to the midi clock and the smpte generator, and getting the latency out of cv gate triggering, all without a visual representation of waveforms.

I don't miss the legwork, but I do miss the weird cranky machines and wibbly software we used to play with in the pre-DAW days. I sold my Stacey, with its massive 40meg hd, about ten years ago, when I finally felt comfortable about the timing on pcs.

[Chas]

Very few people remember Doctor T and yet Emile Tobenfeld was a pioneer, producing what was arguably the first editable computer sequencer, KCS (Keyboard Controller Sequencer) for the Commodore 64 in 1984. In fact, Emile had written a paper for MIT's "Computer Music Journal" in the winter of 1983 outlining outlining the basic concepts of a computer sequencer and talking about potential concepts and features that would not be implemented until KCS Omega 5 was released around 1989.

KCS was always the red-headed step-child compared to sequencers like Pro-24 and Creator, with a quirky interface and a steep learning curve, but it had many features that even today have no equal. KCS's unparalleled Open Mode allowed sequences to be triggered in real time from the keyboard. These sequences could contain other data to in turn trigger other sequences or alter some other aspect, such as playing a sequence but up a Major 4th. There were even control messages that could be inserted into a sequence that allowed for the use of flow control, stochastic and aleatoric composition techniques: I once wrote a four-section work where, although the musical structure was tightly defined, it never played the same thing twice.

This, coupled with KCS's almost contemptuous regard for the "tyranny of the bar line", made KCS the perfect choice for composers interested in complex interlocking polymetre and nonstandard time signatures.

Doctor T's also produced editors for the major synths of the time, a score-writer and a number of other music composition tools such as Fingers and Tunesmith. Emile's own MIDI-Ax, essentially Fingers with a whole slew of interactive real time controllers and triggers was deemed far too complex for general release—in fact, Emile designed it solely or his own use, but would gladly give copies away to those brave enough to take on the extraordinarily rich musical landscape it engendered.

I can remember being at NAMM in 1994 demoing KCS Omega for Doctor T's and was gratified to see the reverence he was held in by his peers in other companies. They—quite rightly, in my opinion—regarded him as a legend.

=:~)

[TheOtherHobbes]

Tobenfeld

is still gigging as a VJ. His MIDI tools were a very nice combination of creativity and convention, and he's done something similar with visuals.

The smart kids these days use some combination of Max, Processing, PD, and Supercollider. The really smart kids roll their own projects with Haskell, Clojure, OpenMusic, and the like.

[Anonymous Coward]

Nice article, but I'll have to admit that just with the first part I got lost somewhere in between. Even so; I think it's a very nice gesture considering that MIDI has got to be one of the most impressive protocols out there.

With that I'm referring to the fact that it's one of those few standards where people and organisations actually comply to and follow said standard, while still trying to come up with improvements. For example; modern MIDI controllers and synthesizers also utilize USB these days. My electronic keyboard (no synth, that's the task for my computer) as well as my MPD24 MIDI drumpad are both connected through USB. Where my MPD24 can even be used as some kind of bridge since it also provides 2 regular MIDI connectors.

So basically a 30 year old protocol which has adapted to modern times, while still making very sure not to 'abuse' the protocol itself. I think that's pretty impressive.

[Anonymous Coward]

The problem is MIDI is the success you talk about. It became so widely adopted that nobody dared improve it as they would suffer compatibility issues.

MIDI was designed for a mono-timbral world where it was envisaged that you would use up to 16 different synthesisers each playing one instrument.

Later on when synths became multi-timbral it meant that one synth could gobble up all 16 channels. So then you require a device with multiple MIDI outputs (which you should use anyway, since the busier MIDI gets the more jitter).

It's just a shame that mLAN didn't take off. But I suppose the future is ethernet/wifi using OSC or similar.

[Richard Neill]

How is 31.25 kbd fast enough?

I've never quite understood this. If a single note takes about 8 bytes (on and off, pitch, velocity, control data), then MIDI is limited to 500 notes per second. That seems reasonable, but not exactly generous, especially in a large rig,

which might have 10 instruments. However, given the beat-structure, perhaps 90% of the cycle should be silent. That only leaves a headroom of ~5. So, how does MIDI avoid a significant amount of skew between notes played by different instruments that should be on the same beat?

[jonathanb]

Re: How is 31.25 kbd fast enough?

500 notes per second is more than enough for anyone. No, seriously, it is.

Musicians measure the speed of music in beats per minute. The slowest you are likely to go is about 40 beats per minute. The fastest is somewhere just above 200 bpm.

If you have a tune being played at 210 bpm, that allows for 142 note changes per beat. I can't think of any tune that has that level of complexity, and if it did, the brain wouldn't be able to pick most of it up anyway.

Most of the time, 500 notes per minute would probably be enough.

[TheOtherHobbes]

Re: How is 31.25 kbd fast enough?

Most notes only need a note on MIDI message, which is three bytes, compressed to two with a feature called running status. You only send control messages when you need them, and all sequencers apply bandwidth limiting of control messages.

Even so, MIDI has relatively poor timing. The pros had giant interfaces so they could give each synth a separate dedicated MIDI cable. The host computer buffered all the MIDI events on a track ahead of time with a short look-ahead and sent them to the interface with timing tags. The interface then spat them out when needed.

If you tried to control more than three or so synths over a single MIDI cable you did indeed get slop and creep, because of a combination of limited MIDI bandwidth and slow synth response - the internal processors were all 8-bit slow and took some time to parse an incoming a MDI message and set up the hardware.

Mostly you could feel it wasn't right. Sometimes you could hear it clearly, especially on over-quantised tracks with multiple chords piled up at the start of a bar.

Sometimes you had to hand edit some notes back by a few clocks to avoid the log jam.

[Anonymous Coward]

Pitch bend is optional, the note on event contains the note number. Control changes are optional too.

But yes, timing does start to shift a little. Vince Clark went back to CV as a result, these days he divides his time between big modular synths using CV and software plugins.

[Frankee Llonnygog]

Great articles

Really enjoyed both parts - thanks Mr Dormon

I can't help thinking that the SynthAxe would probably be a viable DIY project these days.

[Anonymous Coward]

Re: Great articles

Yes, +1 from me too. Very nice.

[Britt Johnston]

MIDI can now do lyrics

I don't know if all MIDI editors do the same, but I've had files from one MIDI with lyrics come out into an unrelated notation programme. I saw it first quite recently, less than 10 years old.

[bag o' spanners]

correction

In my post above I alluded to hardware gizmos that Dr T created, but which were in fact created by J.L. Cooper.

I'm blaming bank holiday early morning brainfarts.....and a measure of senility.

Cooper was one of the pioneers of hardware midi merge, thru, and switch boxes, along with the Fadermaster (the original midi fader controller) The company still makes high end A/V studio gear that does exactly what it says on the tin, without breaking sweat. One of the few peripherals companies from the dawn of midi that has survived intact.

[Anonymous Coward]

Midi controllers....

...they didn't all have to be quirky....violins and chelloscould either be a little futuristic or odd looking, but others can be absolutely stunning

[J.G.Harston]

Self build

I remember building a MIDI interface for my Beeb (based on a partial circuit diagram reflected in somebody's glasses in a cover photo!) and then plugging it into my friend's Amstrad CPC MIDI interface and swapping files.