The View from 2005: This stuff is absolute bedrock. MIDI may be a little more sophisticated now, and we know a LOT more about multitrack production and overdubbing, but these are still timeless verities. I think this may be one of the most important articles I’ve written.

Bob Moog

The really important thing that Bob Moog did, back when he developed the analog voltage-controlled synthesizer in the 60s, was to create a control language that was simple, robust and effective. There was nothing particularly new about his oscillators, filters and amplifiers at the time, and, hey, the keyboard has been around since the 14th Century. But if you were to look at the synthesizers that came just before Moog’s, particularly the ones built by RCA in the 50s that came to be housed in the Columbia-Princeton Electronic Music Labs, what you would notice is that they were really clunky to use. Sounds were essentially treated as static objects, and you created them one at a time, using punched holes on a paper tape. Given the editing limitations of the time, this was not a particularly musical way to create music.

Moog controlled the behavior of his audio devices via changing DC voltage, so that the frequency of an oscillator or filter and the gain of an amplifier was varied by changing a DC control voltage applied to the device. He unified the behavior of these devices by creating a remarkably compact language that linearised the exponential behavior of music, defining the ratio of 2:1 (which in music is equivalent to an octave, and in audio to 6 dB) as a change of 1 volt. If you wanted to go up an octave, you increased the control voltage by 1 volt; ditto if you wanted to go up 6 dB. This was really simple and elegant in application. It made the musical use of voltage-controlled oscillators, filters and amplifiers comparatively easily and intuitive. Once you got a handle on the system, it was really pretty easy to get around. And given the right devices for generating control voltages (keyboards, envelope generators, LFOs, etc.), you could really have a pretty good time playing.

As time passed, however, and multiple manufacturers got into the act, it got a little frustrating, because devices didn't all track control voltages exactly the same, and some manufacturers used slightly different voltage conventions (like 1.2 volts per octave), and naturally, all us studio slobs wanted to hook everything together.

MIDI came as a response to these expressed needs - a digital control language that defines notes, events, and a variety of control parameters, as well as patches, that was quickly and broadly embraced by all synthesizer manufacturers. It expanded on the analog control system and raised it to a higher musical level. By adding multiplexing (the addition of an address to each word of data), it permitted polyphonic and multi-timbral control of quite complex systems. Because it was digital, it opened the way for elaborate sequencing possibilities that eclipsed the fixed-stage analog sequencers that used to just generate parallel strings of control voltages that you would use to control oscillators, filters, amplifiers and to trigger envelope generators.

The Musical Assumptions Underlying Synthesis and Synthesis Control Languages

This little bit of history is interesting for the common-sense and intuitive way it defines some basic elements of music. Probably without such an approach, synthesis would never have worked (like, suppose we had started by trying to do Fourier Transforms of every millisecond of sound – I really don't think it would have caught on). The premise underlying conventional music synthesis is that the synthesizer note is the smallest complete musical element. It has a pitch (frequency), timbre (waveform), loudness (amplitude), envelopes (shapes over time of amplitude and spectrum) and it is oriented in time relative to other notes. Just like on an orchestral score, only electronical. Cool! What could be simpler and more straightforward?

It helps that the development of multitrack recording/overdubbing occurred in parallel to this. In the multitrack world, music is thought of as parallel strings of signals from individual instruments that are "mixed" into a virtual stereophonic ensemble. Such world-views of music are comfortable ones, and they are directly related, in a conceptual way, to the orchestral score of notated music.

Traditional Musical Notation, Yet Another Control System

An orchestral score, or a lead sheet, or any piece of notated music for that matter, can be thought of as set of fairly detailed instructions to a performer or performers, including note choice, timing, loudness, etc., in fact all of the values that are included in a MIDI data stream. In most cases, the instructions are lined up in parallel for multiple performers, so the conductor and/or composer can see at a glance the vertical, or harmonic, relationships of the various simultaneous combinations of sounds. Timing, since Beethoven, has been spelled out in beats per minute, sometimes with Italian adjectives like molto Vivace or Largo.

This convention of music notation has been around for so long and is so deeply ingrained in our culture that it is thought of as “the music” even though it’s just dots on a page. Composing music is generally thought of as the act of making up and writing down such dots on a page, dots that performers will subsequently convert into music. Even musicians who don’t read music often think this, and many use MIDI for the express purpose of converting a perfectly fine performance into dots on a page! Anyway, the idea behind this view is that if the dots are any good, the music the performers play from the score will be way cool.