History of MIDI
20 years ago, in January 1983, leading synthesizer makers launched the MIDI revolution and changed the way music is created, recorded, performed, taught and enjoyed. What began as a simple objective -- to play one electronic keyboard from another -- has grown into a multi-billion dollar annual business, with companies offering MIDI in everything from synthesizers and recording equipment to computers and cell phones.
Today, most film and TV scores, as well as popular recorded music is written and performed using electronic keyboards and other MIDI-equipped musical instruments. (Thanks to advances in digital sampling and synthesis technologies, the orchestra playing behind that big-screen block buster is more likely to be the product of MIDI than a real orchestra with dozens of acoustic instruments.)
Well, this was the stepping stone for the Midi standard.
Early synthesizers, accessible to the public mostly through special exhibitions and occasional movie scores, were analog devices, and, before the advent of the transistor, were large, expensive, and difficult to operate. But following the development of the transistor, Robert Moog was able to design analog synthesizers for mass market which were considerably smaller and cheaper. By the end of the 1960s Moog synthesizers were appearing in more and more popular music, and their unique sound made Wendy Carlos' Switched-On Bach one of the best-selling classical albums of all time.
The early commercial analog synths, however, were not designed with interconnectivity in mind. In fact, the control interface used in these synthesizers was itself analog. Along came midi - to digitize and standardize it all . . .
Over the two years from autumn 1981 to 1983 most of the major manufacturers collaborated in the development of the first version of the Midi standard, which was published in October 1983. There were by this time already Midi devices in the marketplace, the first of which was Sequential's Prophet 600 which first shipped in December 1982.
The Midi 1.0 standard provides:
The Midi Manufacturers Association (MMA) was set up with responsibility for maintenance and development of the standard.
The work that would eventually result in the MIDI 1.0 standard began as a conversation between three audio engineers at the June 1981 trade show of the National Association of Music Merchants (NAMM). I. Kakehashi (Roland Corporation), Tom Oberheim (Oberhiem Electronics) and Dave Smith (Sequential Circuits) were all concerned with the difficulty musicians faced in connecting synthesizers from the different manufacturers. Starting from the existing literature on computer networks, Smith worked up an initial proposal which he presented to the Audio Engineers Society in November of that year, called the "Universal Synthesizer Interface" (USI).
Smith's proposal reached the ears of Japanese synthesizer manufacturers, who had been working on their own standard. The Japanese standard was more complex than USI, which was intended mainly for note on and off events. At the next NAMM show in January 1982, the Japanese, who included such major manufacturers as Korg, Kawai, and Yamaha, joined with the American manufacturers to coordinate the two efforts. Five months later, the basics of the Musical Instrument Digital Interface were born, but not formally standardized.
In August 1983 the MIDI 1.0 Standard was formally published. The MIDI Standard was the product of commercial manufacturers, and control over it is held by two industry organizations: the MIDI Manufacturer's Association (MMA), and its Japanese counterpart, the Japan MIDI Standards Committee. Any proposed changes must be passed by both these organizations in order to be adopted into the standard.
NOTE: There is also a user-oriented advocacy group, which began as
the International MIDI User's Group (IMUG) and later evolved into the International
MIDI Association (IMA).
MIDI specifies that MIDI devices must have a MIDI IN port, a MIDI OUT port, and (optionally) a MIDI THRU port. Though beyond the scope of this discussion, all the electrical specifications of these ports are provided in the protocol, so that connection of any two MIDI devices can be accomplished with a single cable. The MIDI IN port allows the device to accept messages in the electronic language which is also part of the MIDI protocol. The MIDI OUT port allows the device to issue such messages to other devices. The optional THRU port allows the device to pass messages directly from its IN port, allowing devices to be "daisy-chained" together.
In addition to the hardware specification, MIDI provides a description of the information that is passed along the three types of ports. MIDI commands consist of one or more bytes of data, and include such things as note on and off commands, clock synchronization, etc. These commands are transmitted serially, at a specified rate of 31250 bits/sec, but most MIDI messages have a slot for a channel number. This allows simulation of parallel instrument control: MIDI devices can be configured to accept only messages corresponding to a given channel or set of channels. The possible such configurations are specified in the protocol, and are called modes.
While the technical aspects of the MIDI protocol are borrowed from the established computer-network field, the structure of MIDI is built upon two implicit models: a performance model and a musical model. Each of these models is simple and powerful, and while their ramifications were perhaps not entirely foreseen by the authors, each has had a large impact on MIDI's popularity (both positive and, as we shall see later, negative).
Before MIDI was introduced, synthesizers consisted of two components within one integrated system. The first component, the sound synthesis engine, consisted of the electronics which actually generated sound. The second component was the controller, usually a keyboard, which captured the musician's intentions and parsed them into voltages and currents that the synthesis engine could understand. This latter process is sometimes called performance gesture capture.
MIDI made the distinction between these two components explicit, and thereby essentially broke their mutual dependence. Because any controller could control any synthesis engine, musicians could build up systems from modular components, picking and choosing from a wide range of commercially available products. In addition, purchases could be made more freely because the fear of obsolescence - the bugaboo of electronics consumers everywhere - was minimized by the universal nature of the control language.
only supports a data rate of 31.25 kBaud
128 instruments to choose from
limited to 16-channels (16 instruments) - therefore it cannot be used to recreate entire symphonic scores, that include 30, 40, or more instruments.
uses a distinct, 12-tone scale - therefore "enharmonic distinctions" (e.g. F sharp vs. G flat) are not expressible in MIDI. In addition, non-Western or microtonal scales cannot be replicated
Overcoming the Limitations - New Midi Specs
The Midi Tuning Specs - in response to the limitation of a 12-tone scale - the "Just Intonation Network" constructed a proposal to extend MIDI to include many pitches in between. This was adopted by the MMA. However, virtually all sound card manufacturers simply ignored the new specs !!!
Even with these additions to the MMA's standard, the MIDI format is not sufficient, by itself, to produce a full musical score, nor as a storage mechanism for such a score which has been parsed electronically. Several authors have devised extensions to MIDI standard to cope with such concepts as enharmonic notes, key signatures, crescendi, slurs, and the like (see [Selfridge-Field, 1997]). These extensions, while compatible with the same underlying MIDI framework, have not been adopted by the MMA.
Actually, it is a good thing that these new specs that go beyond the Midi
Tuning Specs have not been adopted by the MMA. If they had been -
users everywhere would suddenly have trouble playing all the "new"
midis, since the standards were not backwards compatible.
MIDI is not without its limitations, made all the more problematic in that they are not just accidents of implementation - they are inherent in the underlying models that MIDI builds upon. These limitations become apparent when MIDI is applied to applications which, though logical extensions of its expressive power, were not MIDI's original purpose.
Still, MIDI is nearly ubiquitous in the electronic music industry. Any
future system which attempts to replace MIDI will need to concern itself with
backward compatibility with existing systems, and it seems likely that such a
system will contain the MIDI protocol as a subset of its capabilities. At
sixteen years old, MIDI will remain a part of the electronic musician's
repertoire for many years to come.