While music made with software feels ubiquitous today, due in part to electronic music and digital audio workstations, it wasn't long ago that the thought of making music on a computer was incredibly impractical. Now, many songs can be produced through entirely digital means, and artificial intelligence programs can even generate complete songs by mimicking styles recognized in their neural webs. Computers in the early days often didn’t even have monitors, and in fact, most institutions had to rent them from the manufacturer because they were so expensive to own.

While the Department of Computer Science at Indiana University wouldn't take shape until 1971, many researchers were already experimenting with computers, even at the Jacob's School of Music. In 1967, a theory course in computer-based music would be offered by a composer gaining recognition in Europe for his mathematical approach to music composition. Iannis Xenakis came to IU in the fall of 1967, setting up camp in Jacobs’ music addition building, with the mission to install a computer system for making digital music without having to know how to read or write in musical notation.

The device would simplify his personalized music composition process. Xenakis had been using computers to create what he calls stochastic music using punch-card computer programming methods in the Fortran II language. Xenakis took inspiration from concepts like set theory and probability using mathematical functions to process pitch, volume, timbre, and other musical variables to write his compositions. This required punching a lot of cards to be fed into a computer. He would then painstakingly convert the punched out results into musical notation and arrange it for the orchestra.

The Polyagogic Information Unit (UPIC) would make it possible to draw music compositions on an X/Y axis. You first begin by drawing waveforms on a small electronic tablet to serve as a generated tone. Traditional synthesizers of the day used analog oscillators to do this. Oscillators can be understood as a circuit with an amplifier that enables a feedback loop, creating a certain sustained frequency. Combined with a filter, the frequency can take a variety of shapes - sine waves, square waves, and triangle waves, to name a few. The UPIC would enable a digital version of this mechanism by looping custom waveforms that otherwise would not have been available in analog synthesizers at the time.

After choosing a waveform, you could then use a larger electronic tablet – one that looks more like an architect’s easel – to begin drawing the layout of your composition. The X-axis represents time, so your composition starts on the left and ends on the right, while the Y-axis represents a waveform’s pitch. If you wanted to draw a parabola for example, you would get a sound that is constantly bending its pitch in a given direction - a musical tool known as a glissando, one that is very characteristic of Xenakis’ style. You could also just scribble all over and get some very interesting, albeit chaotic sounds. In addition, you could apply different algorithms to the composition as a whole, reversing it, inverting it, and so on.

From 1967 to 1972, efforts made to construct this computer synthesizer at Indiana University were delayed and eventually abandoned mainly due to costs of the necessary digital-to-analog conversion circuits – the main ingredient for converting the digital signals into audible sound. It was no simple task as digital audio was still experimental at the time.

The challenge had to do with the nature of computer architecture. Everything that makes up the digital environment of a computer - every file or instruction - must be represented by a unique string of ones and zeros. To get the computer to understand sounds from the real world, it has to be broken down in the same manner. Pulse Code Modulation (PCM) solves this problem by taking thousands of samples of the sound per second to create a binary representation of the analog waveform.

Only a few institutions had working audio synthesis and analog conversion, including Bell Labs and Princeton, utilizing the PDP-1. While Indiana University had the computing power necessary to run the synthesizer’s program, converting the digital audio output into amplified audio proved to be difficult. The task required a lot of customized circuitry that had to be made from scratch.

Though the UPIC would not be completed at Indiana University, it did eventually take shape in 1977 in Issy les Moulineaux, Paris, France. But that wasn’t the end of the story for IU’s computer music program - Xenakis’ Center for Automated and Computerized Music at IU would continue and eventually evolve into the Center for Electronic and Computer Music that we know today, comprising two recording studios and state of the art technology in the Music Arts Center, just right next door to where it all started.

In the video below, a composition made entirely on the UPIC:

For more about the science of sound, computers, and synthesizers, check out some of our other episodes:

• Your Thoughts Could Be Music: The Encephalophone
• Can tooth fillings pick up radio signals?
• How does Autotune work?
• FM vs AM Radio: Driving through a Tunnel
• The Paper Computer Chip

Sources

Turner, C. (2014) Xenakis in America. Tappan, NY: One Block Avenue.

https://120years.net/upic-system-iannis-xenakis-france-1977/ 

Learn more:

Iannis Xenakis was also interested in granular synthesis, or breaking down sound into its smallest parts - first discussed as a topic of quantum physics in 1940. Read about how it works and play with a granular synthesizer in this interactive Introduction to Computer Music textbook from the Center for Electronic and Computer Music website.

There are several open source programs that imitate the UPIC, incluing Iannix, that you can download for free! 

An introduction to Early Computing by PBS studios Crash Course

https://www.youtube.com/watch?v=O5nskjZ_GoI