Computer sound synthesis
The production of electronic sounds by digital
techniques is rapidly replacing the use of oscillators,
synthesizers, and other audio components (now commonly
called analogue hardware) that have been the standard
resources of the composer of electronic music. Not only
is digital circuitry and digital programming much more
versatile and accurate, but it is also much cheaper. The
advantages of digital processing are manifest even to
the commercial recording industry, where digital
recording is replacing long-established audio
technology.
The three basic techniques for producing sounds with
a computer are sign-bit extraction,
digital-to-analogue conversion, and the use of hybrid
digital-analogue systems. Of these, however, only the
second process is of more than historical interest.
Sign-bit extraction was occasionally used for
compositions of serious musical intent--for example, in Computer
Cantata (1963), by Hiller and Robert Baker, and in Sonoriferous
Loops (1965), by Herbert Brün. Some interest
persists in building hybrid digitalanalogue facilities,
perhaps because some types of signal processing, such as
reverberation and filtering, are time-consuming even in
the fastest of computers.
Digital-to-analogue conversion has become the
standard technique for computer sound synthesis.
This process was originally developed in the United
States by Max Mathews and his colleagues at Bell
Telephone Laboratories in the early 1960s. The best
known version of the programming that activated the
process was called Music 5.
Digital-to-analogue conversion (and the reverse
process, analogue-to-digital conversion, which is used
to put sounds into a computer rather than getting
them out) depends on the sampling theorem. This states
that a wave form should be sampled at a rate twice the
bandwidth of the system if the samples are to be free of
quantizing noise (a high-pitched whine to the ear).
Because the auditory bandwidth is 20-20,000 hertz (Hz),
this specifies a sampling rate of 40,000 samples per
second though, practically, 30,000 is sufficient,
because tape recorders seldom record anything
significant above 15,000 Hz. Also, instantaneous
amplitudes must be specified to at least 12 bits so that
the jumps from one amplitude to the next are low enough
for the signal-to-noise ratio to exceed commercial
standards (55 to 70 decibels).
Music 5 was more than simply a software system,
because it embodied an "orchestration" program
that simulated many of the processes employed in the
classical electronic music studio. It specified unit
generators for the standard wave forms, adders,
modulators, filters, reverberators, and so on. It was
sufficiently generalized that a user could freely define
his own generators. Music 5 became the software
prototype for installations the world over.
One of the best of these was designed by Barry Vercoe
at the Massachusetts Institute of Technology during the
1970s. This program, called Music 11, runs on a PDP-11 computer
and is a tightly designed system that incorporates many
new features, including graphic score input and output.
Vercoe's instructional program has trained virtually a
whole generation of young composers in computer
sound manipulation. Another important advance,
discovered by John Chowning of Stanford University in
1973, was the use of digital FM (frequency modulation)
as a source of musical timbre. The use of graphical
input and output, even of musical notation, has been
considerably developed, notably by Mathews at Bell
Telephone Laboratories, by Leland Smith at Stanford
University, and by William Buxton at the University of
Toronto.
There are also other approaches to digital sound
manipulation. For example, there is a growing interest
in analogue-to-digital conversion as a compositional
tool. This technique allows concrete and recorded sounds
to be subjected to digital processing, and this, of
course, includes the human voice. Charles Dodge, a
composer at Brooklyn College, has composed a number of
scores that incorporate vocal sounds, including Cascando
(1978), based on the radio play of Samuel Beckett, and Any
Resemblance Is Purely Coincidental (1980), for computer-altered
voice and tape. The classic musique concrète
studio founded by Pierre Schaeffer has become a digital
installation, under François Bayle. Its main emphasis
is still on the manipulation of concrete sounds. Mention
also should be made of an entirely different model for
sound synthesis first investigated in 1971 by Hiller and
Pierre Ruiz; they programmed differential equations that
define vibrating objects such as strings, plates,
membranes, and tubes. This technique, though forbidding
mathematically and time-consuming in the computer,
nevertheless is potentially attractive because it
depends neither upon concepts reminiscent of analogue
hardware nor upon acoustical research data.
Another important development is the production of
specialized digital machines for use in live
performance. All such instruments depend on newer types
of microprocessors and often on some specialized
circuitry. Because these instruments require real-time
computation and conversion, however, they are restricted
in versatility and variety of timbres. Without question,
though, these instruments will be rapidly improved
because there is a commercial market for them, including
popular music and music education, that far exceeds the
small world of avant-garde composers.
Some of these performance instruments are specialized
in design to meet the needs of a particular composer--an
example being Salvatore Martirano's Sal-Mar
Construction (1970). Most of them, however, are
intended to replace analogue synthesizers and therefore
are equipped with conventional keyboards. One of the
earliest of such instruments was the "Egg"
synthesizer built by Michael Manthey at the University
of Århus in Denmark. The Synclavier later was put on
the market as a commercially produced instrument that
uses digital hardware and logic. It represents for the
1980s the digital equivalent of the Moog synthesizer of
the 1960s.
The most advanced digital sound synthesis, however,
is still done in large institutional installations. Most
of these are in U.S. universities, but European
facilities are being built in increasing numbers. The
Instituut voor Sonologie in Utrecht and LIMB (Laboratorio
Permanente per l'Informatica Musicale) at the University
of Padua in Italy resemble U.S. facilities because of
their academic affiliation. Rather different, however,
is IRCAM (Institut de Recherche et de Coordination
Acoustique/Musique), part of the Centre Georges Pompidou
in Paris. IRCAM, headed by Pierre Boulez, is an
elaborate facility for research in and the performance
of music. Increasingly, attention there has been given
to all aspects of computer processing of music,
including composition, sound analysis and synthesis,
graphics, and the design of new electronic instruments
for performance and pedagogy. It is a spectacular
demonstration that electronic and computer music
has come of age and has entered the mainstream of music
history.
In conclusion, science has brought about a tremendous
expansion of musical resources by making available to
the composer a spectrum of sounds ranging from pure
tones at one extreme to random noise at the other. It
has made possible the rhythmic organization of music to
a degree of subtlety and complexity hitherto
unattainable. It has brought about the acceptance of the
definition of music as "organized sound." It
has permitted the composer, if he chooses, to have
complete control over his own work. It permits him, if
he desires, to eliminate the performer as an
intermediary between himself and his audience. It has
placed the critic in a problematic situation, because
his analysis of what he hears must frequently be carried
out solely by ear, unaided by any written score.
|