MIDI’s Initial Specification: Exchanging Performance Instructions between Instruments

By 1983, microprocessors in electronic keyboards could record key presses, select sounds and drive sequencers, but manufacturers’ interfaces often could not connect directly. Compatibility required more than matching plugs: earlier analog equipment differed in control voltages and trigger polarity, while digital equipment used different data formats. Controlling one synthesizer from another keyboard often required special conversion circuits.

Dave Smith and Chet Wood of Sequential Circuits proposed the Universal Synthesizer Interface, USI, in 1981. It sent messages when events such as key presses occurred rather than transmitting the whole keyboard state. After industry discussions, Roland contacted Yamaha, Korg and Kawai and continued exchanging drafts with Sequential. The Japanese proposals added richer controls and a distinction between status and data bytes. Cooperation gradually produced the shared specification.

References: [1] [2]

MIDI stands for Musical Instrument Digital Interface. At the winter NAMM show in Anaheim in January 1983, Sequential’s Prophet-600 and Roland’s Jupiter-6 demonstrated an operating connection between brands. Work on detailed rules continued afterward.

Bob Moog’s July article in Keyboard already described exchanged drafts, texts made available to the industry and early products. The initial document under the International MIDI Association is dated 5 August 1983. Circulating drafts, that dated document and the October version mark separate stages. Electronics & Music Maker reprinted the October specification in full in May 1984. January’s demonstration and later documents trace development and dissemination during the year.

References: [2] [3] [4]

A note-on message can contain three bytes: one for type and channel, followed by key number and velocity. The table published in May 1983 assigns middle C number 60, with key and velocity values ranging from 0 to 127. Velocity is a key-performance parameter that a receiving instrument may use for dynamics, distinct from musical tempo. The receiver’s own sound circuitry generates the audio.

A sequencer can store notes and control actions and send them again, without recording the sound waves from a speaker. Program changes select a sound location on the receiver. Different oscillators, filters and envelopes mean the same instructions need not generate identical timbres. Manufacturer-specific data provides additional formats for particular equipment.

References: [1] [2]

The October specification defines asynchronous serial transmission at 31,250 bits per second. A byte with start and stop bits occupies ten bits and takes 320 microseconds. Five-pin DIN connectors are labeled MIDI IN and MIDI OUT. Optical isolation at the receiver reduces ground-loop problems. One output drives one input, while optional MIDI THRU forwards incoming data for chaining devices; longer chains require attention to accumulated signal-edge errors.

The sixteen channels are logical message numbers, separate from cable count and the number of voices an instrument can produce. A receiver may listen to all channels or selected ones; monophonic and polyphonic modes determine internal voice allocation. System clock and start/stop messages coordinate sequencers and drum machines. Early texts specify twenty-four clock pulses per quarter note, allowing machines to follow a shared tempo.

References: [1] [3]

The May 1983 table still leaves some statuses undefined and labels one controller group simply as switches. Later expansions should be read as later developments. Moog reported that some products added MIDI to existing designs and did not expose every panel function through it. Manufacturers needed to connect devices and examine their behavior.

The shared interface allowed combinations of keyboard control, sequencing and sound modules from different brands through protocol and implementation together. Early documents established basic messages, wiring and responses. Full functional equality, standardized timbres and later file-exchange formats were separate issues from basic connectivity.

References: [1] [2] [3]