home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Simtel MSDOS 1992 September
/
Simtel20_Sept92.cdr
/
msdos
/
txtutl
/
travesty.arc
/
TRAVESTY.MSG
< prev
Wrap
Internet Message Format
|
1988-12-18
|
3KB
From: amlovell@phoenix.princeton.edu (Anthony M. Lovell)
Newsgroups: comp.binaries.ibm.pc
Subject: v01i065: travesty, text/music style simulator w/C source
Date: 17 Dec 88 23:01:05 GMT
Summary: travesty.arc, text/music style simulator w/C source
Approved: dhesi@bsu-cs.UUCP
Posting-number: Volume 01 Issue 065
Originally-from: amlovell@phoenix.princeton.edu (Anthony M. Lovell)
Submitted-by: amlovell@phoenix.princeton.edu (Anthony M. Lovell)
Archive-name: travesty/travesty.uue
[
This is Anthony M. Lovell's implementation of a "travesty" algorithm.
The purpose of travesty is to analyze some text and find statistical
correlations between characters, then create some text that preserves
these correlations. The result will stylistically resemble the
original, though it might make little or no sense.
This could be useful if, for example, you were Bacon and you wanted
to write a play and pretend that you were Shakespeare. You would
do something like:
C:\PLAYS> trav 4 < hamlet > macbeth
The output would be stylistically similar to Shakespeare's style in
Hamlet, though it might not make much sense. Oh what a tangl'd web we
weave when first we practise to deceive...Of course, I have no evidence
that the real Bacon actually did this, or that he even owned a
microcomputer. The parameter "4" to the "trav" program above asks it
to use a fourth order travesty, which means that it will look for
correlations of characters in groups of four.
More interestingly, Anthony Lovell points out that it is possible to
encode a melody into characters, and ask this program to generate new
melodies from the encoded original. If you were to take a number of
Chopin waltzes, for example, and do a travesty on them, the output
might be some fairly decent-sounding music in Chopin's style. This
makes me wonder if it might not be possible to write a travesty program
that would take as its input four concurrent streams of data
representing four-part harmony and preserve their relationships in a
smooth way, so the output would contain some reasonable harmony too.
Naturally, to make the output sound like something by Debussy, you
would probably take the input from a random number generator. :-)
There is a known bug in this program: If a parameter of 2 is supplied,
the output may not be very random. Other values work well.
Source for Turbo C 2.0 is included here. Anthony Lovell wants to
credit the source of this idea thus: He wrote this program after a
friend, who had seen the algorithm in an old BYTE article, described it
to him. He has not seen the article himself.
He suggests: "I recommend you try this out on a textfile you know
well - it's more fun that way."
-- R.D.
]