(compMus1) Granular Synthesis

Overview

• Any sound can be thought of as containing discrete particles. (Mosaic or Pointillism)
• A grain of sound typically lasts from 1 to 100 ms.
• Each grain shaped by an amplitude envelope.
• Sound parameters (pitch, panning, duration, envelope, location in sound file, wave type, etc) change on a grain-by-grain basis. Parameters are static within each grain.

High-Level Organization

Because of the number of grains that can be in play at one time, control of grain parameters is usually organized by functions of parameter of changes over time. Overall organization can be categorized according to how a change in one parameter affects changes in other parameters.

• Pitch Synchronous: control of grain parameters linked to synthesizing desired pitch and formant content. Pre-analysis of sampled sound is needed. The result is sonically similar to a buzz generator fed into formant filters.
• Quasi-Synchronous Streams: grain length can be modified independently, but grains usually follow at regular intervals. Quasi-synchronous streams are usually used for time compression/expansion and transposition. This organizational technique is commonly implemented in commercial synth’s and software, such as Kontakt.
• Asynchronous Clouds: all parameters are independently controlled. Has been compared to controlling a jet spray water nozzle.

Compositional Approaches for Using Granular Synthesis

• Isolate a single sound or gesture in a soundfile, with just a little silence around it.
• Use Th0nk with the Hectic preset to generate a number of variations of the gesture.
• Isolate individual gestures in the output for further processing. (phase vocoding can work well on Hectic output.)
• Use the Flowing preset to create slowly evolving textural sounds that you can use for background layers in your composition.
• Use Cecilia and the stretcher module to time expand with more parameter control. You can randomize the grain length within boundaries, shift pitch over time, or change the panning of grains. Short grain lengths give a more tremelo effect; longer grains sound smoother.
• Use Cecilia and the warper module to time expand/compress with smoother results (generally) than stretcher. Glissando effects are very nice with warper.

Time Stretching, Phase Vocoding, and a little more on Th0nk

• Inharmonic sounds (sounds with partials that are not integer multiples) are harder to phase vocode without artifacts. You can still get interesting results, but the process usually results in noticeable artifacts.
• Since Th0nk only will stretch to 1 or 5 minute(s), add silence to the soundfile to get the desired expansion factor. (Example, a 5 second file will be expanded 12x to one minute. If you only want to expand it 3x, add 15 seconds to it. 20 seconds expands 3x.)