Historical View of Synthesizer Development

Figure 1: Example Music V unit generator diagram. The left input of each oscillator is the amplitude control, and the right input is the frequency control. In this example, control parameter P5 controls overall amplitude, P6 sets the note duration, and P6 controls the note frequencies. The upper oscillator serves as an amplitude envelope generator for the lower oscillator. Function F1 is trapezoidal in shape (the amplitude envelope), and F2 is the desired oscillator waveform and sets the timbre of the note.

With the Music III program, Max Mathews and Joan Miller introduced the concept of the unit generator for sound synthesis, a technique extended in the languages Music IV and Music V [8]. A unit generator is a fundamental building block, or module, used to build a variety of sound-generating algorithms. Each unit generator accepts numeric parameters and/or audio signal(s) as input, and produces an output signal.

The unit generators of Music V included an oscillator, filter, adder, multiplier, random number generator, and envelope generator. Basic signal processing and synthesis modules could be interconnected to create interesting synthetic sounds. The techniques of additive, subtractive, and nonlinear synthesis (such as frequency modulation, or FM) could be implemented naturally with these modules. They were similar in function to the modules used in analog synthesizers at the time, such as voltage-controlled oscillators (VCOs), amplifiers (VCAs), and filters (VCFs). Analog synthesis, in turn, utilized modules from earlier audio electronics.

Composers wrote instrument definitions in Music V as networks of interconnected unit-generators. An instrument invocation was essentially a subroutine call with arguments, called pfields (parameter fields) supplied to the the instrument. A Music V score was essentially a time-ordered sequence of instrument calls. Since the instrument and score definitions in Music V completely specified the music computation in a procedural text format, Music V gave us a kind of musical counterpart of Postscript for 2D graphics (the standard marking language used first for laser printers and more recently for computer displays). Apparently, Music V was born at least three decades too soon to be accepted as the PostScript of the music world. Instead, we got MIDI (Musical Instrument Digital Interface).

In the years following the availability of Music V, a number of research centers with access to large, mainframe computers and sound converters extended the music compiler in various ways. At Stanford University's Center for Research in Music and Acoustics (CCRMA), for example, Music V descendants such as Mus10, introduced named variables, an Algol-style language for instrument definition, more built-in unit generators, piecewise-linear functions for use as envelope parameters, and an instrument compiler. Descendants of Music V appeared at numerous universities and research centers around the world. Computer music blossomed in the seventies, with many software and hardware systems appearing. It would not be feasible to adequately survey parallel developments throughout the world in this short essay, so the remainder of this historical sketch describes developments from CCRMA's point of view. The CCRMA story is applicable to other computer music laboratories that have invested significantly in digital synthesis hardware.