Full text

In the late 1950s, engineers were struggling to digitize telephone speech, running streams of acoustic data through mainframes. Max Mathews realized a paradoxical truth: if a machine could compute a human voice for a telephone line, it could generate any conceivable waveform. Thus, telecommunications engineering gave birth to the first software for direct sound synthesis.

The music creation process demanded humility, as real-time synthesis did not exist. A composer punched code onto stiff punch cards and handed them over to the operators of the vacuum-tube giant IBM 704. Mathews devised an elegant architecture, dividing the system into two parts. The "orchestra" consisted of virtual building blocks (Unit Generators, or UGens)—mathematical filters and oscillators. The "score" dictated timing and pitch. To save memory, wavetable synthesis was employed: the computer took a digital snapshot of a single wave cycle and looped it at the required speed. The machine chewed through these equations in batch mode overnight. In the morning, the numbers were run through a DAC onto magnetic tape. To hear 17 seconds of the dry, mechanical beeping of the piece "The Silver Scale," the composer waited for long hours, listening to the hum of the cooling fans.

Despite the primitive sound of early experiments, the architecture of MUSIC N defined the development of digital audio. In 1968, Mathews rewrote the code in FORTRAN (as MUSIC V), allowing the program to run on mainframes from other manufacturers. Distributing the source code punch cards to universities stimulated independent research: Jean-Claude Risset used this environment for the acoustic modeling of a trumpet's timbre, and John Chowning at Stanford discovered the method of FM synthesis. Mathews' concept of Unit Generators (independent software blocks for signal generation and processing) became the foundation for modern visual programming environments (Max/MSP, Pure Data) and the entire architecture of VST plugins.