\|\\M0NGR30;
\F0\CThe 'SCORE' Program for Musical Input to Computers
 
\CLeland Smith
\CStanford University
 

 
Abstract:
 
\J	The score program has been in use at Stanford for
 about ten years now.  What began as a system for
generating parameters for a particular piece of music has
evolved into a general purpose musical input language that
has been used at several computer music centers.  Since the
program is in FORTRAN it is not machine dependent.  SCORE
can be adapted to create note lists for a variety of
computer sound generating programs.
 
	With SCORE it is possible to use much of the standard
terminology of music when dealing with pitch and rhythm.
However SCORE by no means limits one just to the tempered
scale and conventional rhythms.  This discussion will give
an overview of the principal features of SCORE with
emphasis on the program's newer features.  These include an
added capability to aid in the simulation of musical
phrasing and increased flexibility in regard to the use of
several different procedures in a single parameter.\.

 
\C____________________________________________________________
 


\J	The Stanford sound generating system, in use from the
late '60s until 1979, evolved from a form of Max Mathews'
MUSIC4 into what came to be known as MUS10.  In the late
1970s Stanford put into operation Peter Samson's real-time
digital sythesizer, the "Samson Box."  These two systems,
along with the well-known MUSIC5 and several other computer
music systems, use for input some kind of list of
"instrument" definitions followed by note lists which
include a name or number which refers to a particular
instrument followed by a list of parameters which convey
detailed information about the nature of each sound to be
produced.  In a very simple case there might be as few as
four parameters: begin time, note duration, pitch,
amplitude.  However we often work with instruments which
have as many as 99 parameters.  These added parameters will
refer to things such as quadriphonic position, envelopes,
vibrato, reverberation, glissando limits, various ways of
creating different timbres, etc., etc.

	A rather short piece of music can include more than
1000 notes.  Clearly then we are faced with a massive
typing job unless we can get the computer to help us to
deal with the many redundancies which are inherent in
music.  SCORE has been created for the purpose of
facilitating input to the computer of data for any musical
style.

	For example, the begin time (parameter 1) of each
successive note, or sound event, for a given instrument is
usually dependent on each preceding note duration
(parameter 2).  Thus, after stating an instrument's initial
begin time, P1 is always generated automatically by SCORE.
Any parameter that is to remain unchanged need only be
stated once.  If all the notes in a single voice were to be
.1" long, the statement\.

\CP2 .1;

\Jcould generate the durations for thousands of notes.

	However we usually want a little more variety than
that, at least in the parameters of rhythm and pitch.
SCORE allows you to specify strings of input for any
parameter.  The required syntax for this is that a
parameter number be given, followed by one of the special
code words and then the desired sequence of data.  There
are several code words that can be used in this regard.  I
will give brief examples of the three most often utilized,
NUM, RHY, NOTES

	A string of time values, in seconds, might be entered
as follows:\.

\CP2 NUM/ .5 / .2 // 1.3 / 6.2 / .1 //// 4;

\JThe slashes are delimiters, and when no new number appears
between slashes, the last one seen is repeated.  The
semicolon is always the terminator.  There are several
features available to indicate various kinds of repetition.\.

\CP2 NUM/ .35 / .12X8 / 1.5;

\JIn this case the duration .12" will repeat eight times.

	Usually when we use the code word NUM we are dealing
with real time values which will not be transformed in any
way.  However if we prefer to enter time data in terms of
musical rhythmic notation we will be dealing with relative
time values which become real time values only after being
multiplied by a tempo factor.  The default tempo factor in
SCORE is 1, or, in musical terms, MM=60, i.e. there will
be 60 quarter notes played per minute - one per second.
This mode of input is invoked with the code word RHY.  Now
the numbers entered will be the denominators of musical
fractions; /4/=a quarter note, /8/=an eighth, /1/=a whole
note, etc.  A dot can be placed after any number to
increase its time value by 50%.  (Several dots can be added
if desired.)  However dots followed by numbers are taken as
decimal points.  /.5/ represents a double whole note.
/.5./ represents a dotted double whole note!

	Any complex rhythmic relationship can be expressed.  A
half note tied to a sixteenth would be /2 16/.  An ordinary
triplet is a /12/ (there are twelve of them in a whole
note); a quintuplet is a /20/.  Each note in a group of
seven in the time of a dotted quarter would be a /18.6667/.
(This latter is arrived at by means of a simple formula:\.
 
\CR=n*4/T

\Jn is the number of equal units to be heard in the time span
of T quarters.  In the example above we have n=7 and T=1.5;
18.6667=7*4/1.5 .)

	Of course all these numbers are relative time values.
SCORE produces the real time values by dividing the number
into 4 and then dividing the result by the tempo factor.
Thus the real time value of an eighth note at a tempo of
MM=60 is 1/2 second.  [(4/8)/1=.5]   At a tempo of MM=120 it
would be:  (4/8)/2=.25 second.  SCORE provides the
capability of frequent tempo changes, ritardando and
accelerando at any rate.  There is even the possibility of
having different streams of tempo changes in each voice.
The tempo feature can be used both with the RHY and NUM
code words, but then of course the data given after NUM
will also be treated as relative values which will be
transformed by the tempo factor.

	At Stanford, P3 is usually reserved for pitch input.
SCORE has three different modes for pitch input.  The NUM
code word can be used if frequency numbers are to be used.
This mode would have to be used for micro-tonal scales of
an irregular (non-tempered) nature.  If the conventional
tempered scale is used the ordinary letter names of the
notes are used.  A suffix of S is used for sharps and F for
flats.  The octaves are numbered by a system which sets the
lowest C on the piano keyboard as C1.  This makes middle C
into C4 and the highest C into C8.  It is easy to keep
track of these by remembering that the low C of the cello
is C2, the low C of the viola is C3, the high C of the
bassoon is C5, the high C of the trumpet is C6, and the
high C of the flute is C7.  Once an octave number has been
stated it need not be typed again unless the notes move
into a new octave range.  The code for a mid-range Eb scale
would appear as follows.\.

\CP3 NOTES/ EF4/ F/ G/ AF/ BF/ C5/ D/ EF;

\J	A third mode of pitch input allows for the
establishment of microtonal scales made up of any number of
equal tempered steps to an octave.  This kind of input
requires three parameters.  The fundamental pitch is given
in P3.  This can be stated as either a frequency number or
the letter name of any note of the ordinary twelve note
scale.  Then in two higher parameters we must give the
number of equal divisions of the octave and a list of the
particular scale steps desired.  In the following example we
will give the code for a seventeen tone scale starting from
a basic pitch of 440 hertz.\.

   P3 440;
   P21 MICRO 17;
   P20 NUM/0/1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17;
 
\J	Both the basic pitch in P3 and the scale division
number in P17 can be changed as often as desired.  In P20,
minus numbers can be used for notes which descend below the
basic pitch.
 
	Time does not permit a full description of all the
features of the SCORE program.  (Much of this is covered in
my article "SCORE - A Musician's Approach to Computer
Music" which first appeared in the Journal of the Audio
Engineering Society, January 1972, and later in NUMUS West,
l973.)  However I would like to give examples of a couple
of the newer features of SCORE.

	One of the most difficult problems in computer music
is to achieve good, musical articulation and phrasing.
SCORE helps to solve this problem by giving complete
control over the "duty factor," that is, the actual
duration of each note irrespective of the perceived rhythm
as defined by the time interval between attacks.  The time
values given in P2 are the intervals between the attacks.
The actual note durations are established by use of the
code DF used in conjunction with other SCORE code words.
Take for example the conventional musicial situation where
we find a string of quarter notes with the indication,
"poco a poco piu staccato."  Here the attack interval
remains constant but the sounding duration of each note
becomes less and less.  This could be notated in SCORE:\.

   P2 RHY/ 4X12 ;
   P15 DF NUM/ 1/ .93/ .86/ .79/ .72/ .65/ .58/ .51/
   .44/ .37/ .3/ .23;
 
\J	In this case the duty factor list will serve as a
simple multiplier applied the real time values produced by
P2.  For the third note the output of SCORE will give P1 as
time 2 even though the previous note began at time 1 and
the value .93 was found in P2.  In other words, a rest of
.07 seconds has been inserted at the end of the second
note.

	It is also possible to create a legato effect by
increasing the duty factor to a number greater than 1.
Since the actual duration of the note is greater than the
time between attacks, this causes any given note to overlap
with the following note.  In the next example the MOVE
feature causes DF to gradually decrease from 4 to .08.
This will cause the passage to begin with extreme legato
and progress to the point where the notes are so short that
they are heard almost as clicks.\.

     TOOT 0 20;  P2  .12;      <Tape example 1>
     P3 NOTES/ EF/ F;          < These 2 notes will repeat.
     P4 .6;  P5 F6;  P8 F1;      < Amplitude, envelope, timbre
     P9 .05;                       < 5% reverberation
     P10 DF MOVX/ 16  4  .08;  < Move exponentially from 4 to
     END;                          < .08 during 16 seconds.
 
\J	In the cases described above the duty factor was a
simple multiplier.  There are several other ways of using
the duty factor.  A fixed amount of time can be added or
subtracted from the notes in a string of varied P2 values.
Or the sounding duration of the notes can be set to a
constant without regard to the time between attacks.  This
last can be especially useful when dealing with undamped
bell-like sounds.  The following example is made up of
glockenspiel sound.  The wave form used is a simple sine
wave and the envelope is an exponential curve.  The rhythm
of the attacks includes fixed values, 20th and 12th notes,
etc., as well as some random elements.  (This will be
explained below.)  The pitches and amplitudes are chosen at
random from fixed ranges.  The duty factor is established
in a user-added subroutine which causes the sounding time
of each note to be dependent on the randomly chosen pitch;
the higher the pitch, the shorter the note.\.

       <Tape example 2>
     TOOT 0 24;  P2 RHY/ 20X5/ P12// 12/// P12/ 36/ P12;
     P3 1 G4, C7;   P4 1  .05, .6;
     P5 F5;  P8 F1;  P9 .05;
     RD  .1;             <Random deviation of attack = + or - .1".
     P11 DF SUBR;     <Call subroutine to set the duty factor.
     P12 1  1, 4;       <Choose random numbers between 1 and 4.
     END;
 
          <Crucial lines of subroutine.>
              X=P(3)-69                ;69=GS5 (C5=61)
              IF(X.LT.0)X=0             ;USES RANGE GS5 TO DS7 (88)
              IF(X.GT.20)X=20          ; SETS NOTE DURATION
              P(IPAR)=207.-5.*X/20. ; FROM 2" TO 7"
 
\J	In the example you will notice that P12 appears at
three places in the rhythm list.  When SCORE sees a
secondary parameter designation within one of its lists, it
then places whatever the current value of that parameter
may be in the list.  In this case the effect was that a
rhythm of longer, varied duration was inserted at several
points in P2's string of values.  This is a bit like using
the fermata sign in conventional musical notation.

	In the next example secondary parameter calls are made
in the pitch list, P3.  First of all a string of 10
tempered scale notes is established as motive "Z".
Following this, P12 is called upon five times in
succession.  Looking down to P12, we find that this will
produce randomly chosen notes between F6 and D7.  Then the
statement /@Z-1/ brings back motive "Z", but 1/2 step (-1)
lower.  After this P13 will be called upon for the next 7
notes, random frequencies between 1200 and 1400 hertz.  And
so forth.\.

       <Tape example 3>
     TOOT 0 17.7;  P2 .1;       <Will play 177 notes.
     P3 NOTES/ Z( F6/G/A/FS/EF7/DF/G6/B/D/GS) / P12X5 / @Z-1 /
          P13X7 / @Z9 / P12X4 ;   <The string cycles until end.
     P4 .5;  P5 F6;  P8 F1;  P9 .05;
     P10 DF NUM/ 1X10 / .1X5 / 2X10 / 1.5/ 1.3/ 1.1/ .9/ .7/
          .5/ .2/ .1;
     P12  1  F6, D7;   P13  1  1200, 1400;
     END;
     TEMPO/ 70;          < Converts attack durations to .085"
 
\J	The last tape example incorporates all the features I
have discussed.  The timbre input for all the notes in
this, as well as all my previous examples, is a simple sine
wave.  I hope this will serve to demonstrate that a
relatively wide variety of sounds can be attained from very
limited means.  In all, three different envelopes are used,
a sine wave, and the features of the SCORE program.\.
 
         <Tape example 4>