Granular Synthesis Workshop

Fourier shows that complex phenomena (the pattern of heat flow through metal objects) can be described as a combination of simple sine waves. He validates the theory through measurement, comparing mathematical predictions of temperature change to real thermometer readings along heated rods.

This leads to the insight that any signal which changes over time could be expressed as a sum of simpler oscillations.

Fourier on Wikipedia

The first experimental validations of Fourier’s theory in acoustics. Using specially designed spherical resonators tuned to specific frequencies, Helmholtz could isolate individual harmonics within complex sounds by ear.

By using multiple tuned resonators in parallel, Helmholtz identified the spectral makeup of different instruments and confirmed that complex tones are composed of simpler sinusoidal components — demonstrating the real-world applicability of Fourier analysis to sound perception.

(Everybody in this room has used a Helmholtz resonator by blowing over the mouth of a bottle.)

Helmholtz on Wikipedia · On the Sensation of Tone

The Fourier transform gave us a way to analyze frequencies in a sound overall, but had no time-based analysis. The Gabor transform adds a time-based window, allowing us to analyze frequencies at specific moments or in series.

Gabor’s work inspired thinking about micro-time events: sounds in the 1–100 ms range. He was not a composer, but this way of thinking about sound inspired others to consider micro-sound as a compositional material.

Dennis Gabor on Wikipedia · Gabor transform

Anticipated by Igor Stravinsky, Henry Cowell, and others, Pierre Schaeffer gave us the name Musique concrète.

Traditionally, recording technology had been used to capture and recreate (as faithfully as possible) the sounds of the source material, typically traditional instrumentation and arrangements. This can be considered a first-order transformation (original sound → derived recording).

Schaeffer’s innovation was to treat recorded sound itself as raw musical material and open the gates to any kind of sound. By cutting, looping, reversing, and layering tape, he introduced a second-order transformation: composing with sounds directly, expanding the definition of what could be considered musically meaningful.

Schaeffer’s notion of the “sound object” emphasized perception over source, foreshadowing later ideas of sound as discrete, manipulable particles.

Pierre Schaeffer on Wikipedia · Musique concrète · Listen on YouTube

Le Caine was a garage inventor as well as an occasional composer. His most well-known work is Dripsody, formed entirely from a single sample of water dripping into a pail.

Le Caine built several generations of the “Multi-track”, which he used to alter pitch and create loops and layers. Sequencing was done manually by cutting and splicing.

Dripsody emerges from the musique concrète tradition and anticipates many granular and sample-based techniques to come.

Hugh Le Caine on Wikipedia · Listen on YouTube

Iannis Xenakis is considered the creator of what we now think of as granular synthesis. Concret PH can be thought of as the first branching-off point from musique concrète to a granular form of composition.

Concret PH is built from one-second fragments of recordings of wood burning. The fragments are layered and organized to create the impression of a more unified sonic texture. There is change throughout the piece, but not the big transitions we heard in the earlier Schaeffer work.

Listen on YouTube

Xenakis built on the ideas in Concret PH by connecting the aesthetic ideas to the work of Dennis Gabor, giving us the term “grains of sound” along the way.

Analogique A-B has two elements. The A part is played on traditional stringed instruments. B was produced by recording sine waves from tone generators. Both use a stochastic (statistically derived) means of composition.

The grains in Analogique B were distributed across the time horizon of the score by dividing it into a series of “frames” and building a statistical and logical framework to govern the generation of subsequent frames and the amount of change they would introduce. The version we hear unites both A and B.

“All sound, even continuous musical variation, is conceived as an assemblage of a large number of elementary sounds adequately disposed in time. In the attack, body, and decline of a complex sound, thousands of pure sounds appear in a more or less short interval of time delta.”

Listen on YouTube

Composed of sine tones, filtered noise, and impulses (short, percussive clicks), Kontakte was built with entirely synthetic source material.

Stockhausen pointed out that rhythm and pitch exist on the same continuum:

• If you speed up rhythmic pulses enough, they become pitch.
• If you slow down pitch, it becomes rhythm.

Kontakte demonstrates this repeatedly as pitches are decomposed and recomposed in various configurations.

Stockhausen on Wikipedia · Kontakte · Listen on YouTube

Curtis Roads studied with Xenakis in the late 1960s and became interested in granular synthesis.

He recorded the first digitally constructed granular compositions in the mid-1970s. Roads continued to use granular elements in his compositions throughout the later 70s and 80s, but these always required asynchronous processing: the composer would write a program to generate sound material, and it would be executed by the computer.

Listen on YouTube

Working at Simon Fraser University in British Columbia, Barry Truax developed the first real-time granular synthesizer, called GSX, using the DMX-1000 DSP.

GSX allowed the user to adjust key parameters: synthetic sound source, grain envelope, frequency, density, and panning. It supported frequency modulation and used multi-stage ramps over different parameters to modulate the characteristics of each attribute over time.

Truax’s most recognized recording from this period is Riverrun.

The SignalFunctionSet GSX module is a free VCV Rack implementation of this system.

Listen to Riverrun on YouTube

The limitation of GSX was its reliance on synthetic source material. As mentioned, we can divide granular synthesis into two camps: synthetic grains (built from simple waveforms, often sine waves) and granulated grains (extracted from any arbitrary sound material).

Truax’s next innovation was to build a real-time granular processor of arbitrary sound material. This is a form of delayed sampling: the grains are extracted from any sound material, which radically alters the available timbre of the source material.

In the second piece on his album Pacific, we hear what may be voice recordings granularized and modulated. If you’re like me, there’s a different kind of engagement here: I feel as if I’m always at the edge of understanding, and my brain is trying to reconstruct the samples.

Listen on YouTube

Cloud Generator was a granular app for Mac OS 9 that supported both synthetic and granulated synthesis. Over the following decades, Roads refined the ideas through several desktop applications, the most recent of which is EmissionControl2.

I’ll end this history portion with the reason many of us became interested in granular synthesis: the Clouds module by Mutable Instruments.

Clouds was not the first hardware granulator (and was preceded by a few Eurorack modules), but it was the first to take granular synthesis mainstream in Eurorack. Since Clouds, we’ve seen tremendous growth in granular modules in both hardware and software.

Émilie Gillet on Wikipedia · Watch on YouTube