Computer Sound Synthesis

Sound Waves. Sounds travels in the form of air pressure waves. Microphones transform them into analog variations of electric current. These variations travel through the wires to speakers where they make the cone vibrate and set the air in motion, producing again air pressure waves. The range of the human aural perception is roughly 20 - 20,000 Hz (cycles /sec or such variations per second).

For all this and the following topics consult

John Pierce and Max Mathews - The Science of Musical Sound

on reserve

Sampling. The variations of the electric current are in the form of a wave which can be measured from time to time. These measurements or samples are done with a device called Analog to Digital C onverter (ADC). In order to obtain adequate measurements which will catch all present frequencies, the number of samples per second should be at least twice the number representing the highest frequency. The recording standard is 44,100 samples/sec and the broadcasting standard is 48,000 samples/sec. These samples (numbers) can be stored in a computer's memory and used to produce a CD. When you listen to your CD, such samples (numbers) are read by a Digital to Analog Converter (DAC). A composer can either process further these numbers (samples) or create them from scratch with the aid of known algorithms.

Sound Synthesis Methods.
In class we briefly mentioned a few methods or systems used in digital sound synthesis:

The MusicN paradigm. Proposed by Max Mathews in the 1950s, it is modeled after both the 19th. century orchestra and the early electronic music equipment. It consists of an "orchestra" of digital "instruments" which include oscillators, mixers, etc. and a "score" which contains instructions for these instruments on how and when to produce each sound: start time, duration, frequency, amplitude, etc. A MusicN type of program (Music V, Music 360, 4BF, M4C, cmusic, etc.) reads such instructions, performs the computations, and generates an audio file (binary). Example: John Melby - "Chor der Waisen".
FM Synthesis. Authored by John Chowning in the early 1970s, this method uses the same principle as the broadcasting technique with the same name. By carefully controlling the ratio between a carrier frequency and a modulating frequency the user can obtain "side bands" and generate complex sounds. At the time it representing a big step by allowing the creation of rich sounds with relatively little computing power. Even today, most synthesizers use Chowning's FM technique to deliver a very large array of timbres. Example: John Chowning - "Turenas" .
L.P.C. technique (Linear Predictive Coding). In very simple terms, this technique works somewhat like a word processor, allowing the user to modify certain attributes of the sound while keeping others constant. The example presented in class, Charles Dodge - "Any Resemblance Is Purely Coincidental", takes an 78 rpm recording of Enrico Caruso and renders sometimes the original recording, sometimes only Caruso's voice accompanied by a pianist on stage, sometimes making the tenor sing completely out of his natural range or as a choir.
Subtractive synthesis. Using complex, rich pre-existent sounds, the user modifies them by changing their carachteristics through further processing. Similar to Pierre Schaeffer's musique concrete this method is very popular today. Example: Jean-Claude Risset - "Sud"
Granular synthesis. First proposed by Iannis Xenakis in the late 1950s, this method did not become available until two decades later due to the lack of computer power at the time. Based on the theories of Gabor, the sound is regrded as a collection of extremely brief sound "quanta" or "grains" clustered in a "cloud" which approximates the profile of the desired sound. A very flexible tool, granular synthesis is used extensively today when computers have become so much more powerful. Example: Barry Truax - "Riverrun"
SAWDUST A system created by Herbert Brun at the University of Illinois, it distinguishes itself through the elegance and relative simplicity of its design. Herbert Brun composed with this system a number of pieces reflecting an experimental attitude as well as his strong beliefs about th role of art in the society. Example: Herbert Brun - "I toLD You so !"

GACSS (Genetic Algorithms in Composition and Sound Synthesis) was devised by Ben Grosser at the University of Illinois. It uses GA techniques, or search procedures patterned after the workings of natural genetics (survival of the fittest), as a control structure for sound synthesis and compositional parameters. Example: Zack Browning - "Breakpoint Screamer"
Additive Synthesis builds a complex wave (sound) by adding simple waves - usually sine waves - whose frequency, amplitude, and phase can be controlled in a very precise way.