perm filename USEMUS.LCS[UP,DOC]5 blob sn#197349 filedate 1976-01-21 generic text, type C, neo UTF8
COMMENT ⊗   VALID 00010 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00003 00002	**********  Using the Music System -- MUS10  **********
C00021 00003	A type of flow-chart diagram for SIMP would appear as follows:
C00024 00004
C00028 00005	In the next example a unit generator will be added  above  the  right
C00034 00006		Various types of noise  and  other  random  fluctuations  are
C00040 00007		Frequency modulation allows for  the  production  of  a  wide
C00045 00008	********** APPENDIX XXX NOT COMPLETE!!!XXX  ************
C00048 00009	********* SOME INFO RE. THE 'FUNC' PROGRAM ***********
C00051 00010	Information re. WAVES.   To run it type R WAVES.
C00053 ENDMK
C⊗;
**********  Using the Music System -- MUS10  **********

******* WORK IN PROGRESS -- JUNE 75 -- LELAND SMITH *********

This manual is designed for use with the PDP10 at the Stanford AI lab.
It  is  quite  possible  that  several features discussed will not be
operative on other installations.  See appendix for some details.


MUS10 is a complete sound generating  package  which  exists  on  the
disk. In its basic form it includes no storage area for sound output,
so when the program is to be run, a  number,  establishing  the  core
size, must be given following the program name.

For first attempts type:

	R MUS10 10	
	(All lines must be terminated with the 'RETURN' key.)

At this point the program will type the message:

	INPUT?

Basically  there  are  two responses possible.   If the program is to
receive  further  instructions  from  another  file  which  has  been
prepared with an editing program, type:

	NAME -- where NAME is the name of the file to be read.
			(If NAME has an extension, it must be used!)

If instructions are to be entered by means of the  teletype  keyboard
(TTY mode), type carriage return (<CR>).

At  this  point  a  star  (*)  will appear which means the program is
awaiting input.

Most complete statements  to  be  read  by  MUS10  must  end  with  a
semicolon.  Several  complete  statements  may be entered on a single
line but it is best not to have the lines too long.   More  than  one
line  may  be used for a single statement.  If the less-than sign (<)
appears everything following on that line will be ignored.  Use  this
for entering comments.

   	*****  Note  that  the above rules DO NOT apply to the syntax
	of the SCORE program.

Already  present  in MUS10 is an "instrument" known as SIMP which has
been set to play a test tone of 'A' (440 hz) for 1/2 second.

In  order  to  play  this  tone, first get into TTY mode as described
above, then type:

	PLAY;SIMP;FINISH;

When the computation has finished the note 'A' will be heard  if  the
various  switches on the sound system are all in the right positions.
(See appendix for this  information  and  information  regarding  the
filter settings.)

Assuming  there were no typographical errors, you will know the comp-
utation is done by the appearance of another star (*).  The test tone
may now be repeated several times by typing:

	$P N where '$' indicates the 'ALT' key and 'N' is the
		number of times the note will play.

If no sound has been heard up to this point repeat the  above,  using
some  large number (e.g.20) for  'N' and go test the various switches
until the tones are heard.

Occasionally a message will appear:

	HUNG DEVICE AD

If this happens get expert help.

******************************************************

The instrument SIMP has five parameters.

	P1 = begin time of note (in seconds)
	P2 = duration of note     "     "
	P3 = pitch
	P4 = amplitude
	P5 = wave form (or timbre)

P1 and P2 will have the same significance in all instruments but  all
higher   numbered   parameters   are  assigned   roles  according  to
convenience.  (However it will prove useful to consistently apply  P3
and P4 as above.)

Internally all pitch entries become  numerical,  however  the  twelve
frequencies  of  the  tempered chromatic scale, from middle C (261.62
hz) up to B may be used in MUS10 by typing the letter  names  of  the
notes. The letter S = #; flats are not used in this program.

Since these letters merely represent the  frequencies of  each  note,
the  octave range may be changed multiplying or dividing by multiples
of two.   Thus C or A in the octave below middle C would  be  entered
as  C/2  or  A/2.   In the octave above the basic middle octave these
notes would be C*2 or A*2.

	C -- 2 octaves down would be C/4
	C -- 3 octaves down would be C/8
	C -- 2 octaves up would be C*4
	C -- 3 octaves up would be C*8 etc.

To test the use of these letters try:

	P3←C;PLAY;SIMP;FINISH;

Now instrument SIMP will play middle C instead of A.   The left arrow
(←)  indicates  that  the value of C has been placed in P3, replacing
any value that was previously there.  (The left arrow and the  equals
sign[=] are interchangeable in this program.)

PLAY;SIMP;FINISH;  must  be typed so the new note will be computed.
After it is first heard it may be repeated as indicated above.

If frequencies other than those of  the  tempered  scale  are  to  be
played, a number may be used instead of a letter.

	P3←1000;PLAY;SIMP;FINISH; will play a tone at 1000 hz.


The amplitude scale available is the range of number from 0 to  2047.
(This  upper  limit  is  set  by the number of bits [12] used for the
sound samples.  See appendix.) P4 has been set at 1200 for  the  test
tone.  This may be reset using the same method as described before.

	P4←100;P3←GS*2;PLAY;SIMP;FINISH;

This will play a G# above the middle octave at amplitude 100.

The duration of the tone may be changed be resetting P2.

	P2←.1; etc. will play a note of 1/10 sec. duration.

If  longer  notes are desired the core being used must be expanded to
allow for more storage space.    At  the  established  sampling  rate
(10000,  see  appendix.)  3.33k  of core is needed for each second of
sound.   Since 1/2" may be played when MUS10 has 10k, 14k  is  needed
for   1 1/2",  17k  for  2 1/2",  20k for 3 1/2", etc.   (Methods for
playing much longer examples are described in SCORE.MAN.)

********** SETTING CORE SIZE *********

To  increase  the  core  size type CALL, then C N where N is the core
size desired.   Then type CON (=continue)  to  return  to  where  you
previously were in the program.

When several parameters are to  be  changed  at  once  the  following
type-in format should be used:

	PLAY;SIMP 0 .2 FS/2 850;FINISH;

This will play F# below middle C for 2/10" at an  amplitude  of  850.
(Please  note  that  P5,  the  wave form for SIMP, will be dealt with
later.)

**********  COMMAS  **********

Commas may be used to separate the parameters and if nothing precedes
a  comma  the contents of that parameter remains unchanged.  Also any
parameter numbers higher than the length of  the  list  will  not  be
affected.

	PLAY;SIMP ,.3,,1200;FINISH;  changes only P2 and P4.

******************************************************

A string of notes may be played with the following input:

	PLAY;SIMP 0 .2 C 1500;SIMP .2,,D;SIMP .4,,E;
	 SIMP .6,,C;FINISH;

In this case P1 must be updated for  each  note.  (Never  overlap  an
instrument  with  itself.   Distortion  will  occur.)  P2,  the  note
duration remains unchanged, so the commas suffice for the last  three
notes.  P4, coming at the end of the list for  the  first  note  need
only be stated once if it is not to change.

Rests  are created by simply leaving some time between the end of one
note (P1+P2) and the beginning of the next (the new P1).

	PLAY;SIMP 0 .2 C;SIMP .5;FINISH; will play C  for  2/10",
rest for 3/10" and then play another C for 2/10".


**********  FUNCTIONS  **********

The  wave  form  in P5 is entered by means of a name which is used by
the program to locate a  list,  or  array,  of  numbers  (512)  which
describe  the wave.  The names used for this purpose will always be F
followed  directly  by  a  number.    These  arrays  will  be  called
"functions."

Only  one  function  is  in  MUS10,  although more can be added.  The
function present is known as F1 and describes a sine  wave.   To  see
this wave on the CRT type:

	SEE(F1);

To clear the screen hit the 'ESC' key followed by  a  'C'.  (This  is
called ESCAPE C.)   <CR> will bring back the typescript display if it
is gone.

Functions may be created with an  external  program  called  FUNC  or
within  MUS10  itself  by means of two routines called SYNTH and SEG.
SYNTH is used to create composites made by adding  various  harmonics
together.  The form of F1 could be changed in the following manner:

	SYNTH(F1); 1,1 2,1 3,.5 999;

In the three pairs of numbers the first of each pair  represents  the
harmonic  number  and  the  second  the  relative  amplitude  of that
harmonic.   Thus the ratios of harmonics 1, 2 and 3 will be 1:1:.5 .

The size of the second number of each pair is important only  in  its
relation  to  the other amplitude numbers.  The number 999 is used to
signal the termination of a string of entries.

Several pairs may entered and harmonic numbers up to 256 may be  used
but  in  practice  great  care  must be taken to avoid the "foldover"
effect which  occurs  when  frequencies  higher  than  one  half  the
sampling rate are present. (See appendix.)

It should be pointed out that the fundamental (harmonic #1) need  not
be present in a wave.

	SYNTH(F1); 10,1 12,1 15,1 999; will give the three notes of a
minor chord.  After this has been entered the following will cause  a
C minor chord to play:

	PLAY;SIMP 0 .5 GS/8;FINISH;

While the lowest Ab (or G#) on the piano keyboard has been indicated,
since  the wave form includes only the 10th, 12th and 16th harmonics,
the notes middle C, Eb and G will be heard.

Several  experiments  with  different  wave forms should be made.  Be
sure to SEE the waves so a visual-aural connection might be made.

A function may be changed in the middle of PLAY routine but  it  must
be  noted that the new wave definition must follow! the note which it
is to affect.

	In PLAY;SIMP 0 .3 D 1000;SIMP .3;SYNTH(F1); 1,.7  3,.2
	    5,.1  999;SIMP .6,,E;FINISH;  the newly defined wave
     	will be heard in the second and third notes.


If you wish to have several functions with different names  available
and you do not create them with the FUNC program, their names must be
"declared" to MUS10.  Suppose you wish to have F2, F3  and  F4.   You
must type directly to MUS10 (or into an  EDIT file which will be read
by MUS10) the following:

   ARRAY F2,F3,F4(512);

The "(512)" indicates that each function array will require 512 words
of  storage.   Thus  at  this  point  the free storage left for sound
samples has been reduced by 1 1/2K so you must increase the  size  of
your  core  if  you  want  to have the same amount of playing time as
before.

The following example will play a sequence of notes wherein are heard
the 10th, 14th and 18th harmonics of a low C, then the 10th, 13th and
16th, and finally the 10th, 12th and 14th harmonics.
	As  each  SYNTH  function  is  typed  in  the  wave  will  be
displayed.   If  you  wish  a  clear  screen  after  the last one has
appeared, hit the 'ESC' key followed by a 'C'. (This is called ESCAPE
C.)

	ARRAY F2,F3(512);
	SYNTH(F1);10,1 14,1 18,1 999;
	SYNTH(F2);10,1 13,1 16,1 999;
	SYNTH(F3);10,1 12,1 14,1 999;
	PLAY;SIMP 0 .3 C/4 2000 F1;
	SIMP .3,,,,F2;SIMP .6,,,,F3;FINISH;



	From this point on it would probably be better to prepare any
input  for MUS10 which requires more than a couple of lines of typing
with the SOS editor.  Typographical errors are inevitable and when an
error  is made near the beginning of a string of input typed directly
to MUS10 you most likely will have to retype everything.
A type of flow-chart diagram for SIMP would appear as follows:

		 P4       MAG*P3
		  |         |
		  ↓         ↓
		***************
		*             *    OSCIL
		*             *    U1 (UNIT GENERATOR ONE)
	        *     P5      *
	         *           *
		  *         *
		   *********
		       |
		       ↓
		     *****
		    * OUT *
		    *  A  *
		     *****

The top left input, P4, serves simply as a multiplier for the numbers
found  in  the  wave  form array, P5.  The particular number from the
array to be multiplied is determined by the number in the upper right
input.   The  upper  right  input, in this case P3, when processed by
"MAG" (the "magic" number) becomes the increment, the rate  at  which
the  wave form array is stepped through.  The "magic" number is found
by dividing the array length, 512, by the sampling rate, 10000.

   512/SRATE=.0512 (Higher sampling rates will be discussed later.)

The  maximum  size of the numbers in the wave array is + or -1.  Thus
if P4 is set to 1000 the output of the OSCIL will be numbers  in  the
range  +1000  to  -1000 which will describe the wave form put into P5
cycling at the rate given in P3.


The code for entering this instrument follows:

	COMPILE;
	INSTRUMENT SIMP;
	OSCIL(P4,MAG*P3,P5);
	OUTA←OUTA+U1;
	END;
	FINISH;

Several instruments may appear between COMPILE; and FINISH;.


This  instrument  has  only  one unit generator (the OSCIL) hence the
output of U1 is added to the contents of OUTA.  If there are  several
instruments  the  outputs  of all the instruments will be combined in
OUTA for each sample.

It will be noticed when playing instrument SIMP that the sound begins
and ends quite abruptly.  This is because  no  attack-decay  envelope
has been applied to the tone.  The sound begins at the full amplitude
of P4 and remains at that level for  its  total  duration.

To apply an envelope, another unit generator must be added.


		 P4       MAG/P2
		  |         |
		  ↓         ↓
		***************
		*             *    OSCIL
		*             *    U1 (UNIT GENERATOR ONE)
	        *     P5      *
	         *           *
		  *         *
		   *********
		       |
		       | 	 MAG*P3
		       |         |
		       ↓         ↓
		     ***************
		     *             *    OSCIL
		     *             *    U2 (UNIT GENERATOR TWO)
	             *     P6      *
	              *           *
		       *         *		COMPILE; 
		        *********  		INSTRUMENT TOOT;
		            |			OSCIL(P4,MAG/P2,P5);
		            ↓			OSCIL(U1,MAG*P3,P6);
		          *****			OUTA←OUTA+U2;
		         * OUT *		END;
		         *  A  *		FINISH;
		          *****

Now that the instrument has been expanded you will note  that  it  is
the output of unit generator two (U2) which goes to OUTA.

P5 will now contain the envelope array.   This array is best  defined
by the SEG routine.   SEG defines the positions of line segments used
to approximate a curve.   With SEG several pairs of  numbers  may  be
entered.   The first number of each pair is an amplitude, normally in
the range of 0 to 1, and the second is the step number in the  array.
The step numbers 1 through 100 are used in SEG.  (However internally,
512 array locations are  used.)  Straight  line  segments  are  drawn
between  each  of  the  points  defined.    The following would put a
triangular envelope shape into F2:

	ARRAY F2(512);
	SEG(F2); 0,1  1,50  0,100;

    Note that the routine is terminated when step 100 is reached.
    DO NOT USE 999 with SEG.

After having typed in the code for instrument TOOT and the definition
for an  envelope  in F2, the following will produce a note using that
envelope:

	SYNTH(F1);1,1  2,.4   3,.1 999;< Sets the tone color.
	PLAY;TOOT 0 .5 A 2000 F2 F1;FINISH;

If two envelopes are to be contrasted add another function and define
it.
	ARRAY F3(512);
	SEG(F3); 0,1  1,7  .2,25  .1,60  0,100;< Staccato
	PLAY;TOOT 0 .2 1000 2000 F3 F1;  < P5 has envelope
	TOOT .2 .5,,,F2;FINISH;<Plays stac. then sust.(F1 then F2)

In the next example a unit generator will be added  above  the  right
side  of  the  bottom,  tone producing unit generator.  In this way a
function may be used to describe fluctuations  of  pitch  within  the
duration  of  a  note  --  much  as  the  previous  example  gave the
possibility for changing the amplitude during a single note.


				 MAG*P7-MAG*P3       MAG/P8
	 P4       MAG/P2		   |         |
	  |         |			   ↓         ↓
	  ↓         ↓			 *************** 
	***************			 *	       *   OSCIL
	*	      *	 OSCIL		 * 	       *   U2
	*             *  U1     	 *     P9      *
	*     P5      *      		  *           * 
         *           * 			   *	     * 
          *         * 			    *********
	   ********* 	   MAG*P3		|
	       |	       |   _____________|
	       |________      _↓___↓_
		       |      \     /
		       |       \ + /
		       |        \_/
		       |         |
		       ↓         ↓
		     ***************
		     *		   *
	     OSCIL   *             *
	     U3	     *     P6      *
	              *           * 
		       *         *    COMPILE; 
		        *********     INSTRUMENT GLISS;
		            |	      OSCIL(P4,MAG/P2,P5);
		            ↓	      OSCIL(MAG*P7-MAG*P3,MAG/P8,P9);
		          *****	      OSCIL(U1,MAG*P3+U2,P6);
		         * OUT *      OUTA←OUTA+U3; END; FINISH;
		         *  A  *
		          *****


In order for this instrument to perform glissandos, a third  function
must  be  defined  for P9 (the "shape" of the glissando).  A straight
line slope will suffice for a simple glissando.  After typing in  the
instrument definition set up the three functions.

	ARRAY F5,F6(512); <F1 is already present.
	SEG(F5);0,1  .8,7  1,12  1,90  0,100;<Envelope
	SEG(F6);0,1  1,100; <Slope

In the preceding, the ARRAY declaration is needed only when some  new
function names are to be used.

The following will play a glissando up two octaves, from C to C*4.

	PLAY; GLISS 0 1 C 2000 F5  F1 C*4 1 F6; FINISH;

If P8←.5; (while P2 remains at 1) two glissandos will be heard.



This  instrument  may  be  used  for  a  dramatic  demonstration   of
"foldover", the phenomenon  which occurs when a frequency exceeds the
upper limit of one half the sampling rate.  (See Mathews' book for  a
technical explanation.)

For this purpose it is best to use a Sine wave in P6.

	SYNTH(F1); 1 1  999;

	PLAY; GLISS 0 1 1000 2000 F5 F1 4000 1 F6;FINISH;

This first note will slide up from 1000 hz to 4000 hz.


	PLAY; GLISS 0 1 1000 2000 F5 F1 9000 1 F6;FINISH;

Due to "foldover" (at 10000/2 hz.) this note will slide up to 5000 hz
and return to the 1000 hz level even though 9000 hz was given in  P7.
The  general rule for "foldover" is that any frequencies which exceed
one half the sampling rate will be heard at (SRATE-F) hz.


Try this one!

	PLAY; GLISS 0 1 0 2000 F5 F1 30000 1 F6; FINISH;


This same instrument may be used to produce a vibrato  by  putting  a
sine  wave into P9, setting P8←1/7; (the vibrato rate will be 7 times
per second) and making P7 some very small amount different from P3.

	PLAY; GLISS 0 1 C 2000 F5  F1 C+2 1/7 F1; FINISH;

		(It is assumed that F1 is a sine wave.)
	Various types of noise  and  other  random  fluctuations  are
produced  by the two random number unit generators.  These are called
RANDH and RANDI.  RANDH (H=hold) produces in effect a  function  made
up  of  horizantal  lines at various levels with a perpendicular jump
from one level to the next.  There are  two  inputs  to  RANDH.   The
first   (left  hand)  gives  the  range,  plus  or  minus,  of random
selection and the second (right hand) gives the rate (per  second) at
which the selections are to be made.

	Care  must  be  taken with the number in the first input.  If
the number 100 is given, the output of RANDH will  fluctuate  between
+100  and  -100.  Thus if a range of 100 to 200 is desired, the input
number should be 50 and the number 150 must be added to the output.


				        MAG*P7     MAG*P8
	 P4       MAG/P2	           |         | 
	  |         |		           ↓         ↓
	  ↓         ↓			 *************** 
	***************			 *	       *   
	*	      *	 OSCIL		 *    RANDH    *   U2
	*             *  U1     	 ***************
	*     P5      *      		        |
         *           * 			        |
          *         * 			        | 
	   ********* 	   MAG*P3		|
	       |	       |   _____________|
	       |________      _↓___↓_
		       |      \     /
		       |       \ + /
		       |        \_/
		       |         |
		       ↓         ↓
		     ***************
		     *		   *
	     OSCIL   *             *
	     U3	     *     P6      *
	              *           * 
		       *         *    COMPILE; 
		        *********     INSTRUMENT NOISE;
		            |	      OSCIL(P4,MAG/P2,P5);
		            ↓	      RANDH(MAG*P7,MAG*P8);
		          *****	      OSCIL(U1,MAG*P3+U2,P6);
		         * OUT *      OUTA←OUTA+U3; END; FINISH;
		         *  A  *
		          *****

			ARRAY F2(512); <F1 is already present.
			SEG(F2);0,1  .8,7  1,12  1,90  0,100;<Env.




	The following will produce white noise.

	SRATE←25000;MAG←512/SRATE;SPEED←2;
	PLAY;NOISE 0 .5 C*8 1000 F2 F1 P3 P3*4;FINISH;

	Actually  P8 (given as P3*4) can probably be left at a number
like 4000 for  noise  purposes.    As  P7  is  changed  the  apparent
band-width  of  the  noise  will  be changed.  As the band-width gets
narrower the center frequency becomes more apparent. Thus if P7←P3/16
and  P3 is up in the range of C*8, something of the effect of blowing
across an open tube will be produced.  The  pitch  is  clear  --  but
quite windy.

	The  SRATE  (sampling  rate)  must  be  increased  for  noise
production since very high frequencies are essential.  At SRATE←25000
the high frequency cut-off will be at 12500 hz.

	If P8 is set to a  low  number  (e.g.  8)  individual  random
pitches, instead of noise, will be produced at that rate.


	If the random unit generator  is  replaced  by  a  RANDI  the
random  function  produced  will  be  made  up  of a series of slopes
(I=interpolating) up and down from one random point to  another.   In
the case of noise production there is little difference between RANDI
and RANDH.  However RANDI is necessary for  getting  such  things  as
random  vibrato.    The following will produce an acceptable, "human"
sounding vibrato.

	PLAY; NOISE 0 1 C*2 1000 F2  F1 P3*.01 16; FINISH;

	The random rate of 16 per  second  (in  P8)  is  considerably
faster  than  the  human  vibrato rate of 5 to 8 per second.  In this
case however since  the  full  band-width  (in  P7)  is  only  seldom
attained and the heard effect is that of a rate much slower than 16.
	Frequency modulation allows for  the  production  of  a  wide
variety  of  tone  colors  using relatively little compute time.  The
INTRP unit generator is really just a  combination of an OSCIL and an
ADD box.  The left input represents the output when the function (P10
below) is at zero and the right input represents the output when  the
function  is  at  1  (peak  amplitude).   No time input is given with
INTRP.  The speed of stepping through the function  array  is  always
taken  as  being  P2,  i.e. the note duration.  In this case P10 will
contain an envelope which  will  control  the  changes  in  frequency
modulation.   For  a  full  explanation of FM see John Chowning's AES
Journal article on this subject.


				 P9*P7*MAG    P8*P9*MAG
				      |         |
				      ↓         ↓
				    ***************
				     *           *
				      *   P10   *   INTRP
				       *       *    U2
					*     *
					 *   *
					  * *
				           *         P9*MAG
	 P4       MAG/P2		   |         |
	  |         |			   ↓         ↓
	  ↓         ↓			 *************** 
	***************			 *	       *   OSCIL
	*	      *	 OSCIL		 * 	       *   U3
	*             *  U1     	 *     P11     *
	*     P5      *      		  *           * 
         *           * 			   *	     * 
          *         * 			    *********
	   ********* 	   MAG*P3		|
	       |	       |   _____________|
	       |________      _↓___↓_
		       |      \     /
		       |       \ + /
		       |        \_/
		       |         |
		       ↓         ↓
		     ***************
		     *		   *
	     OSCIL   *             *
	     U4	     *     P6      *
	              *           * 
		       *         *    COMPILE; 
		        *********     INSTRUMENT FM;
		            |	      OSCIL(P4,MAG/P2,P5);
		            ↓	      INTRP(P7*P9*MAG,P8*P9*MAG,P10);
		          *****	      OSCIL(U2,P9*MAG,P11);
		         * OUT *      OSCIL(U1,U3+P3*MAG,P6);    
		         *  A  *      OUTA←OUTA+U4; END; FINISH;
		          *****



    The following functions should be set up to test the FM instrument.

	ARRAY F1,F2,F3(512);
	SYNTH(F1); 1 1 999;   < A sine wave.
	SEG(F2);0,1  .9,4  1,8  1,72  .8,88  .5,95  0,100; < Envelope
	SEG(F3); 0,1  1,100;  < An upward slope or ramp.

	The following will produce a shift from a pure sine tone to a
highly modulated tone over a period of 2 seconds.

	PLAY; FM 0 2 100 1000 F2   F1 0 10 100 F3  F1; FINISH;


	To reverse the procedure, i.e. change from the modulated tone
to the pure tone, reverse the values of P7 and P8.

	P7←10; P8←0; PLAY;FM;FINISH;


	Change F3 (the ramp) to make the modulation  emerge  only  in
the mid-part of the note.

	SEG(F3); 0,1  1,50  0,100;  < Makes a pyramid.

	PLAY;FM;FINISH;

	Try several of the variations suggested in Chowning's article.
********** APPENDIX XXX NOT COMPLETE!!!XXX  ************

The main program for sound generation is currently  called  MUSIC.FAI
or MUSEXP.FAI (EXP=export).  The export version is designed to run on
a standard PDP10 DEC system (which has a "FAIL" compiler.)

The main program must be loaded with two  subroutine  packages  which
are  called MUSF4.F4 (OR MUSEXP.F4) and MUSIO.FAI (or EXPIO.FAI.) The
FORTRAN routines are for the creation  of  function  arrays  and  for
organizing  the  output  of  sound samples to tape or disk after they
have been computed.  The FAIL routines are for fast output of  blocks
of samples.

The STANFORD version can play sound from core or write all samples on
an external device for playback with another program.     The  latter
method is always used for sounds longer than about 10 seconds.

See SCORE.LCS[UP,DOC] for information on longplaying features (RCDFLG
and BIGBIT) and much other information.



************** FILTER SETTINGS **********

	The  filter  switches  are  near the floor roughly behind the
PDP6.  For the purposes of the tests described in this  document  the
setting for channel 1, the right hand switch, should always be at 5K.
This means that frequencies higher than 5000 hz will  be  suppressed.
For  later  experiments with NOISE and FM it may be desired to change
the setting to 12.5 K.  If this is done however sound  run  by  other
users  at  the  lowest  sampling  rate  (10000,  SPEED 3) will always
include the sampling tone of 5000 hz.  The switch at the top  of  the
cabinet should be left at 5 K.



************ LOCAL SPEAKERS ***********


	To  hear  sound  produced by the D-A converter from the small
speaker associated with your TTY console hit the 'ESC'  key  followed
by the '4', and then the 'U'. (ESCAPE 4 U).
********* SOME INFO RE. THE 'FUNC' PROGRAM ***********

CRUNCH:  Any two functions already in  a  single  .DAT  file  may  be
"crunched" together.  Also, a function may be created by  either  the
SEG  or  SYNTH  routines and then if instead of typing "F" for FINISH
the letter "Z" is typed the program will jump immediately to "crunch"
mode.   At  this  point  the  new  function  may be combined with any
function found in the file presently in core.  Note however that once
this new function  is  processed  by  any  of  "crunch"  options  its
original  form  cannot be regained without going back to ordinary SEG
or SYNTH mode.

PLOTTING:  If "SP" (=see on the plotter) is  typed  single  functions
can  be  drawn  on  the  Calcomp  plotter.   The size asked for is in
inches. "SA" (=see all on plotter) will plot all the functions  found
in  a single file. "SX" (=see all on the XGP) will draw all functions
from  a  single  file in the proper size for printing by the XGP.  In
order to use "SX" you must!!! follow the next steps exactly!!!

	Before running FUNC type: A DSK PTP <CR>. This will cause the
	instructions FUNC sends to the plotter to  be  written  in  a
	file on the disk.

	When  the  FUNC  program  finishes  then  type  R X <CR>.
	This runs a  program  called  X  which  converts  plotter 
	information to XGP commands.

	X will ask you 5 questions.  You should answer as follows:

		PLOT.BIN <CR>		(the file name)
		<CR>			(plot slice?)
		5  <CR>			(shift 5 inches)
		<CR>			(use default value of 11".)
		1  <CR>			(1 inch from the left)
		Y			(yes, delete the plot file)



Information re. WAVES.   To run it type R WAVES.

This will allow you to  display  the  actual  wave  shapes  of  sound
computed  using RCDFLG←-1;.  The MUSAA.DMD file thus produced is read
by WAVES.

You may display up to 3072 samples at a  time  but  since  only  1024
separate  positions  can be shown at once, larger  numbers will cause
some samples to be skipped over (but with no great loss.) After  each
group  of samples is displayed a <CR> will move on to the next group.
If a number is typed, that will set the extent of the next group.  If
the number -1 is given, the program will automatically cycle  through
all available samples by groups of the last given number.   (When  no
more samples  are to be found it will end with an error message.)  In
order to make the program  go  backwards  type  any number less  than
-2 for the number of samples you wish to back up.

This  program  should be of use in seeing the wave shapes produced by
amplitude and frequency modulation as well as seeing the  effects  of
foldover,  etc.   Composite  waves  from  more  than one voice may be
displayed but  as  the  complexity  increases  it  may  prove  rather
difficult to glean useful information from what is seen.