Why the first computer music is hard to pin down
Several groups were experimenting with sound on computers at around the same time, so naming a single “first” would be misleading. In the early 1950s, machines in Australia and Britain were programmed to play simple tunes through a loudspeaker. These early efforts are often described as the earliest known examples of computers producing music, though the record is patchy and accounts differ in the details.
The approach was crude by later standards. Early machines had no sound hardware. Programmers worked out that a program could repeat a short loop of instructions at a chosen rate, and that the clicks this produced from a speaker attached to the machine would be heard as a pitch. Change the loop timing, and you change the note. The results were melodies of single notes, mostly familiar songs.
From playing tunes to generating sound
The next step was to stop treating the computer as a clever music box and use it to build sounds from numbers. In the mid-1950s, researchers at Bell Telephone Laboratories, who were already studying how to transmit and analyze speech, began to explore converting computed numbers into audio. An engineer named Max Mathews is widely credited with developing a series of programs, usually called MUSIC, starting in the late 1950s. Their central idea was simple and lasting: describe sound as a stream of numbers, sample by sample, and let a program calculate it.
This was an expensive and slow way to work. Machines of that era were large mainframe computers, and a few seconds of audio could take far longer than that to calculate. Composers often submitted instructions, waited, and then heard the result later, after the numbers had been converted to sound by separate equipment. Still, the method was flexible. Almost any sound that could be described mathematically could, in principle, be produced. Descendants of the MUSIC programs spread to universities and research centers, and many later music programming languages trace their ideas back to them.
Composition as well as sound
Computers were also used to decide which notes to play, not just how they sounded. In the late 1950s, two researchers at the University of Illinois produced a piece for string quartet that used a computer to generate and test musical material under rules. Their work, usually called the Illiac Suite, is one of the best-known early examples. It belongs to a separate thread of history, rule-based and chance-based writing, which is discussed on its own under algorithmic composition. What matters here is that from the start, two questions ran side by side: can a computer make sounds, and can it help decide what to play?
Research centers and new instruments
Through the 1960s and 1970s, a handful of institutions became hubs for the field. Bell Labs, several American universities and, later, a major research institute in Paris were among the places where composers and engineers worked together. Composers such as Jean-Claude Risset, working with Mathews, explored how real instruments are made of many changing components, and used computers to imitate and then reshape them.
A well-known contribution from this period was frequency modulation synthesis, developed by John Chowning at Stanford University in the late 1960s and early 1970s. It produced rich, changing tones from relatively simple calculations, which later made it attractive for commercial instruments. The method was licensed to a manufacturer and appeared in widely used keyboards in the 1980s.
Meanwhile, some work aimed at live performance. Hybrid systems paired a computer with analog synthesizers, using the computer to control voltages and sequences. Real-time digital sound was difficult before the late 1970s because general-purpose machines were simply not fast enough, so special-purpose digital synthesizers began to appear.
Hardware shrinks, studios change
As integrated circuits became cheaper, digital sound moved from research centers to musicians. By the late 1970s and early 1980s, commercial digital synthesizers and samplers appeared. A sampler records a real sound and lets a player trigger it from a keyboard, which is a very different idea from building sound entirely from equations.
In 1983, manufacturers agreed on MIDI, a standard way for instruments and computers to exchange messages such as which note was pressed and how hard. MIDI does not carry audio itself, only instructions, so files are tiny and easily edited. It let a computer act as a sequencer, recording and replaying a performance on any compatible instrument. For many musicians, this was the moment computers became practical studio tools.
Home computers and game sound
At the same time, personal computers brought sound to ordinary households. Many home machines of the late 1970s and 1980s had built-in sound hardware, from a simple beeper to dedicated chips capable of several voices at once. The story of those chips belongs largely to sound chips and game music, but their influence on popular culture was large: a generation heard computer-made music mostly through early video games and the demo scene.
By the late 1980s, personal computers were fast enough to record and edit digital audio directly. Storage was a major limit at first, because sound takes a lot of space, and the arrival of cheaper optical discs and larger hard drives helped. Over the 1990s, software that combined sequencing, recording and effects became ordinary desktop programs.
Limits and open questions
The sequence above is simplified. Many of these developments overlapped, happened in different countries, and were shaped as much by funding and available hardware as by ideas.
It is easy to tell this as steady progress, but several caveats apply. Early computer music was a small, specialized field, and many pieces were heard by few people. Credit is shared among many people, including engineers and composers whose names are less remembered. And the aesthetic question never fully went away: listeners disagree about whether music made with heavy computer involvement is the composer’s work, the programmer’s, or something new.
What is clear is that the core idea, treating sound as numbers that can be calculated, stored and edited, proved durable. Nearly all recorded and streamed music today passes through that digital representation at some point.