If images fail to appear, try left-clicking mouse/"load image" or try a RELOAD.
SERENDIPITY
COMPUTER PROGRAM COMPOSES BEAUTIFUL MELODIES
Greenwich © 1995
From a letter to a friend By Bob Fink

THIS MAY BE THE WORLD'S FIRST computer-composed music. It has a unique look and effect. I first announced this program on CBC radio in Saskatoon, in November 1995.

See score at bottom. For more scores in the Serendipity series, click on the score image. It will find "Serendipity" in the New "Olde-Musicke" page listings.

Since that time, the program (now there are a set of two programs) was nominated by the editorial board of Discover magazine to be entered in their 1997 Technological innovations contest, which carries a top prize of $1oo,ooo -- which I certainly do not expect to win. But one can hope, anyway.*
.

I. BACKGROUND

I found many aspects of composing to be drudgery. When I learned to write programs for my computer, I eventually wrote a program that gets the computer to do what I do when I compose. It will arrange sounds into music far faster than I (or than anyone in the past) could ever do.

It's based on the anecdotal notion that a million monkeys typing randomly on a bazillion typewriters, given enough time, would eventually produce the works of Shakespeare and everything else in the Library.

So, some time ago, I tried that experiment. After my monkeys typed for a while, I looked to see what one monkey had typed. He typed: "TO BE, OR NOT TWO THREE and uH-one, aNd Uh-two, an D UH bflitz FRigh Idit*%6#."

It wasn`t Shakespeare, but clearly, this was on the right track. Only now, instead of monkeys on typewriters, it was me, typing into my computer, turning it into a "million fast-acting monkeys.".

As the music I compose has always been in a classical or baroque-like tonal style, the program was, naturally, developed to write music similar to what I loved most.

But to my surprise, the marvels of computers added their own dimension, composing music using combinations that I would not (nor ever could) have thought of myself.

For example: While the computer-produced melodies (which are lightly edited only here or there by my human judgement) will play out over the speakers only by one note at a time, the "raise-an-octave" feature in the 2nd program actually produces, especially at brisk tempos, the effects both of harmony and counter-melody (or "counterpoint"). This is not something I expected, nor something as a composer I was able (nor even considered trying) to do before with a single line of music. Thus serendipity -- or Fate, not just Fink, must share the credit. While learning the music at slow tempos, It may sound somewhat dissonant or disjointed, but not so at fast tempos, where patterns may be perceived that cannot be noticed at slower speeds.

I have often scored the music so that the raised counter- melodies can be enhanced by being doubled by one or more instruments (and sometimes with a simple added bass) for the advantages of tone-colour or fullness of sound, but little else has been done in the Serendipity series. The music remains basically playable as unaccompanied solos, and written virtually all by computer.

The music melody's first and last notes are automatically made the same to aid the sense of tonality, and a note or two is added manually at the end, or changed, if the final cadence is too abrupt.

In planning for a Book II of the Serendipity music, for those curious to make comparisons, I will include the last work I composed manually "Before the Computer" ("B.C."). I actually now get to date my stuff B.C.!

Also, in successive editions, Book II will include a growing number of "Development" pieces, based on the original Serendipity pieces, in which I compose additional music by combining the best segments of the computer melodies into longer works (cut & paste), or by developing the musical suggestiveness inherent in the computer's melodies. This Development series may include harmony, additional counterpoint, other rhythms for the same material, thematic contrasts, &c. The human input will be greater in Book II than in book one, which will remain examples of the program's capacities on its own.

Aside from the book, many of these pieces will also appear as individual flyers.

All I need now is 4oo more years to catch up to the prodigious computer output which covers me like a ticker-tape machine gone mad. A "Finkenstein" monster, out of control! King Midas was blinded by glitter, but I am being mauled by magnificent melody.

[See page: How Many Melodies Can Be Written?]

II. ON HOW THE PROGRAMS WORK

As far as I know, since 1995, only some computer research people at the KTH Institute in Sweden and two other people (Calif. and England) have announced programs they say will 'learn' the various styles of music including the old masters, like Bach, Beethoven, and so on. And then the program could write out or recompose music in any of those styles.

From the information I have about the Swedish effort, it's a program that arranges or re-arranges pre-composed music rather than composes. That is, you'd feed in a melody and the computer could then arrange it in a jazz or Bach-like or other style. But as far as actually composing melodies, I think I was among the first to develop such a program.

I wrote the program on a small Tandy laptop. I didn't need one of those new pentium turbojet warp-drive cyber-snot micro-zapping-type super-fast computers to do it.

The two others programmers in California and England did work on programs that compose, but it's unclear how much of it truly is "original" -- that is, generated from scales and basic musical structures as much as possible, rather than based on computer-analyzing of pre-composed patterns taken from existing music.

How does the program work?

My background as a musicologist was a key factor in the program's development. It takes not only notes of the scale and makes melodies of them, but uses 2 or 3-note sub-sets based on how frequently certain basic music structures are used in the music style desired, and also draws upon these sub-sets, which are usually segments of the scale or arpeggios of various chords.

I analyze styles of music and this gives me statistical bases for frequencies of sub-sets. Different styles of music use melodic and chord structures in different configurations and quantities typical to that style. It's mostly a statistical thing, and therefore easily programmable.

Some sub-sets could make music that might sound Scottish, others would sound jazz-like or like country fiddle music.

My program, called "Serendipity" (which means discovery by accident) is actually two smaller programs, used one after another.

The first of the two programs (originally called "MZ-COM," short for "Music-Composer") uses sub-sets designed to produce music without human intervention that is classical, like Bach, similar to Bach's unaccompanied violin solos.

In addition, as noted above, the speed of the playback, allows me to create the effect of harmony (or what's called "counterpoint"). Indeed, the "auditory illusion" of counterpoint coming consistently from merely a one-note-at-a-time melody is likely also a "first" in the world of musical composition: Even though the computer is actually playing the tunes back one note at a time, the effect arises when a second program (called "Replay") imposes various generalized formats and rhythms on the best of the first program's computer-composed melodic lines, like raising certain notes of each measure by an octave, producing an upper countermelody and more esthetically sophisticated quality results from which to choose.

The original first program produces about 3o-4o% music that has "possibilities," 10% are really beautiful, and the rest is trash. The second companion program (offering very rapid trials of different formats) allows a few seconds of human choice, but this is enough to re-introduce what a computer can almost never achieve: Inspired results. This provides a much higher percentage of esthetically attractive music.

This is about 5oo times faster than Bach or Mozart could have done it, and often just as good or better. I'm probably now the fastest composer in the West.

Finley said "My reading of the reports of the other programmers, working independently, seems to show similar approaches. In some ways the others go beyond what Fink has achieved, while in other features, Fink seems well in advance of the others at this point."

At the present time, the program works well on the old Tandy M100 laptop computer (with 32K, a portable disk-drive and DiskPower operating system) -- mostly obsolete equipment now in 1999. Sorry to say, a lack of interest has served to prevent making an updated PC-compatible version of the program available. The Basic used in the Tandy laptop is not standard Basic language, unfortunately, and translating and testing is not worth the work at this time....

A sample tape is available ($10 postpaid).
FOR CREDIT-CARD PURCHASES & RUSH DELIVERY SERVICE, click here
.
Bob Fink can be reached at:
GREENWICH PUBLISHING
1829 Arlington Ave., Saskatoon,
Saskatchewan,
Canada S7H 2Y9
TEL# (306) 244-0679 or 931-2189
In order for us to reduce getting spam from junk-mailer programs,
click on the e-mail button, and then
before e-mailing us, remove the '+' symbol from our address
to make sure you reach us.
(All web logos, art & illustrations: © 1996 by Greenwich)
.
Return

AUTHOR'S BACKGROUND

Bob Fink is the author of:

* The Origin of Music (3 editions) This work was reconized as useful by Dr. Anne Draffkorn Kilmer, chair of Assyriology at the University of Calfornia at Berkeley, and who deciphered the oldest song known ln the world (4,ooo years old). This book on music's origins remains the only full-length and conclusive work on the subject (to our best knowledge), receiving consistently excellent reviews.

* "Some New 'Old -Musicke'" (musical compositions). -- Well-reviewed.

* "Lysistrata and the War" -- (An opera in the style of Mozart). Written to be performed at Wayne State University in 1968 (from which Fink has his degree).

* Numerous articles on musicology, including in the scholarly journal Archeologia Musicalis published from West Germany in 3 1anguages. Also published famous essay in Spring, 1997, showing the oldest musical instrument known (43,ooo years old) played part of the do, re, mi scale: See Neanderthal Flute.

.
(Click here: REQUEST FOR HELP FROM VISITORS & READERS about NEANDERTHAL FLUTE ESSAY)

____________________

Fink has also given several concerts of his music in Detroit, New York, and in Saskatoon. His music has been broadcast on TV and radio and by the University of Saskatchewan and some was re-published in a local anthology of composers.

His lecture on music's origins was sponsored by the Public Library and his writings have been used in some university courses.

Fink's work in computer programming has included a poor-man's typesetting program getting excellent reviews in a laptop magazine and the press.

Among his other programs: A Perpetual Calendar (which uniquely translates "Old Style" and "New Style" English colonial dates, and calculates missing parts of dates to allow historians to quickly date old documents); A ticket numbering program; and various word-processing utilities.

Fink is also a noted heritage artist. Many of his detailed realistic line drawings of beautiful heritage sites are widely used by preservationists & published in newspapers, magazines books reviews and posters. His art appears in his web pages and his published music and books. He has written regular columns for some years in the city's daily newspaper and in other weeklies and journals, on the arts and other issues.

if image fails to appear, try RELOAD

Click here to hear piece above
or
GREENWICH SUBJECT LISTINGS-INDEX
(Click here: REQUEST FOR HELP FROM VISITORS & READERS about NEANDERTHAL FLUTE ESSAY)
.
In order for us to reduce getting spam from junk-mailer programs,
click on the e-mail button, and then
before e-mailing us, remove the '+' symbol from our address
to make sure you reach us.
(All web logos, art & illustrations: © 1996 by Greenwich)

Or Return to Top

(*P.s.: I didn't win)
Vz  x7.19.97