In 1951 in Sydney, a room-sized computer played "Colonel Bogey March" — and this was not accidental relay clicking or a side effect of calculations. Engineer Geoff Hill wrote a program that turned a diagnostic speaker into a musical instrument, thirteen years before the first Moog synthesizer and thirty-two years before MIDI. The story of CSIRAC is the story of how a byproduct of engineering control became the first digital interface for programming music.
CSIR Mk1 was born in the Radiophysics Division laboratory in Sydney under the direction of Trevor Pearcey and Maston Beard. The first program ran in November 1949 — the machine operated on 2000 vacuum tubes and occupied an entire room. Memory was built on mercury delay lines: 768 words of 20 bits, where each word traveled through a column of mercury as an ultrasonic wave, bounced off a crystal at the opposite end, and returned. While the pulse flies back and forth — the data lives. Stop the circulation — memory evaporates.
The architecture was serial: the processor handled one bit at a time, queuing operations. Programs loaded from paper tape — a narrow ribbon with punched holes, where each line encoded one instruction. The operator inserted the tape into the reader, the machine pulled it through photocells, translated holes into electrical pulses, and wrote commands into mercury memory. Paper tape was the only means of communication: no keyboard, no screen, only mechanical input and light bulbs on the panel blinking in time with calculations.
In June 1955, CSIRAC was disassembled, packed into crates, and shipped to Melbourne — to the university, where the machine operated until 1964. The move stretched for almost a year: each rack was marked, each wire documented. In Melbourne, the machine was reassembled like a construction set, and on June 14, 1956, it worked again — now under the name CSIRAC. This move saved CSIRAC: when in the 60s old first-generation computers went to scrap metal, Melbourne University preserved the machine as a museum exhibit.
In the center of CSIRAC's control panel protruded a loudspeaker — "hooter," as engineers called it. Its task was purely technical: to emit audio signals when the machine completed an operation or encountered an error. The speaker was connected to the data bus as an ordinary destination register — any write command to this address sent a pulse to the membrane. One pulse — one click. If commands come fast — clicks merge into a hum.
Geoff Hill saw musical potential in this click. Sound frequency depends on pulse frequency: the faster the write commands, the higher the tone. The problem is that CSIRAC executed instructions at variable speed — memory operations took more time than arithmetic ones. Hill wrote a program that compensated for this unevenness: short loop for high notes, long — for low. Each note turned into a carefully calculated sequence of write commands to the speaker register.
In August 1951 at the first Australian computer conference in Sydney, CSIRAC publicly performed "Colonel Bogey March" — a military march that later became the theme of the film "The Bridge on the River Kwai." The sound was harsh, mechanical — not a sine wave but a square wave, rich in harmonics. But the melody was unmistakably recognizable. The repertoire expanded slowly: each composition required weeks of programming on paper tape, because an error in one line meant punching a new tape entirely.
Around 1957, mathematician Thomas Cherry created a program that translated standard musical notation into CSIRAC machine code. Before this, programmers manually calculated loop durations for each note — Cherry automated the process. A musician could write a melody in familiar symbols: quarter note C, eighth note D, rest. The program translated these symbols into write commands of the required frequency, added pauses between notes, and optimized loops for CSIRAC's architecture.
Cherry's system had two modes. The melody could be loaded into mercury memory and performed from there — fast, but long pieces simply didn't fit. It could be read directly from paper tape: the piece could be any length, but no faster than ten notes per second — that's what the reader could handle. This way composition separated from performance: the same score lived both in tiny memory and on long tape.
But CSIRAC had no polyphony — the machine could only play one note at a time. Chords had to be broken into fast arpeggios: A-C-E instead of simultaneous harmony. Volume yielded only to coarse adjustment — Cherry expanded the machine's dynamic range, but fine gradations still died on the way to the speaker. Timbre remained one — harsh, metallic. But despite these limitations, Cherry's system turned CSIRAC into an instrument accessible to musicians, not just programmers.
While CSIRAC played marches in Sydney, at Manchester University Christopher Strachey experimented with Ferranti Mark I — a British computer that also had a diagnostic speaker. In September 1951, Strachey wrote a program that made Mark I perform "God Save the King." A recording survived too: in the fall, a BBC team that came to Manchester for the children's program "Children's Hour" immortalized the machine's singing on an acetate disc — the oldest surviving recording of computer music. Laboratory engineer Frank Cooper asked for his own copy at the time.
Ferranti Mark I worked faster than CSIRAC — 1.2 milliseconds for addition versus about two milliseconds for the Australian machine. This gave the British computer a range almost an octave higher: Mark I hit notes roughly from B in the great octave to C in the second octave, while CSIRAC — from C in the great octave to D in the first. Strachey used subroutines to generate notes of fixed duration — an approach similar to Hill's method, but with more flexible code structure.
British experiments ran parallel to Australian ones, but independently. There's no evidence that Hill and Strachey exchanged methodologies — rather, both arrived at the same solution through machine architecture. Any computer with a programmer-accessible output device could become a musical instrument. The speaker proved more convenient than light bulbs or a printer: it responded instantly, required no consumables, and gave continuous feedback.
After 1964, when CSIRAC was finally shut down, the machine was not dismantled for parts. Melbourne University transferred the computer to Museum Victoria, where it was restored and exhibited. This made CSIRAC the oldest surviving electronic digital computer of the first generation — not the first built, but the first that survived to our days in complete form.
The label "progenitor of electronic music" is incorrect. CSIRAC was one of several experiments of the early 50s, when engineers realized that a digital machine could control sound frequency. Ferranti Mark I, American machines like IBM 704 — all of them had speakers, all could emit tones. But CSIRAC is better documented than the rest: programs, algorithm descriptions, eyewitness accounts survived. The public performance in August 1951 officially recorded the event, and paper tapes with notes survived decades in archives.
The title "first musical computer" depends on criteria. If you count as first the one that played a melody publicly — CSIRAC is among the leaders. If you count as first a machine synthesizing sound by control signals — that's Bell Labs with Dudley's vocoder and the Voder speech synthesizer back in the late thirties, though it was not a general-purpose computer. If you count as first a surviving specimen — CSIRAC is beyond competition.
The CSIRAC story shows how engineering necessity turns into a creative tool. The speaker was installed for debugging — programmers heard a synthesizer in it. The data register was designed for control — musicians found a key in it. This is the mechanism of cultural evolution of technologies: function follows imagination, not specification.
Modern DAWs, VST plugins, modular synthesizers — all inherit CSIRAC's principle: sound as the result of calculations, melody as a sequence of commands. The difference is in scale: instead of 768 words of memory — gigabytes of samples, instead of one speaker — multichannel systems, instead of paper tape — MIDI controllers. But the basic idea hasn't changed: a programmable machine can be a musical instrument if you give it an output device and a frequency control algorithm.
CSIRAC played until 1964 and fell silent forever — turning it on now is dangerous, the tubes might not withstand it. But the fact that the machine survived physically makes it not just a historical artifact, but a witness to the moment when computers ceased to be only calculators. At some point between mercury tubes and paper tape, a space for art emerged — and it turned out to be wider than the creators could have imagined.