Richard Bettis
1) Introduction
The AMPLE program 'U.KSplit' contains the bare essentials of
a utility to produce a sound 'split' when playing a Music 4000 keyboard.
The program will work with a Music 5000 (it doesn't use any
of the words in the M4 module), but the same effect is easier to obtain by
writing multiple parts.
The main purpose of the utility is to allow the 'live'
playing of more than one sound; in the example provided these are 'Upright' on
the higher notes and 'Slapbass' on the lower ones.
The actual position of the split can be defined by the
player when the program is RUN.
2) RUNning the program
When the program is RUN, a prompt is given to define the
'split-point'. When a note is played, this becomes the lowest note for the
'top' instrument.
Any number of notes may be tried, but only the last one
played will count.
When RETURN is pressed to select the 'ok' message, the split
keyboard is brought into operation, and can be played.
In use the current program allows up to four note polyphony
in each of the two sections.
3) How it works
As you may have guessed, the split function is based on the
ACT structure. The rest of the program may also merit some description.
'setup' is merely a definition of a single 'player' with 8
voices, the first 4 voices are used by the higher section of the keyboard,
while voices 5 to 8 are for the lower section.
'select' restores the default ACT, then uses 'getsplit' to
prepare for finding the splitpoint. After that it displays a MENU with just one
option, to tell the program that a splitpoint has been selected. (In effect
this just provides a simple and familiar format for a 'pause' while the user decides).
'getsplit' redefines ACT for the first time. All that it
actually does is to make a copy of the PITCH number, and store this in a
variable, 'splitpoint'.
'ok' just confirms the note that has been chosen (the
numbers work in semitones above or below middle C), and cancels the 'getsplit'
ACT command.
'splitact' redefines ACT to give the split keyboard effect.
One problem to be overcome is the allocation of different voices. Each key
pressed while another is being held adds to the VOICE number generated;
however, we want to split 8 possible notes into two groups of 4. This is done by
'change' (see below) on each voice number, which renumbers 5 to 8 as 1 to 4
again!
The PITCH value of the note is then compared with the
selected value in 'splitpoint'. If it is above the splitpoint the ACT continues
as normal (with any revisions to voice number already made by 'change').
If, on the other hand, the PITCH is below the splitpoint
then 4 is added to each (none zero) voice number.
The net effect of all this is to ensure that higher notes
are played on voices 1 to 4 and lower notes on voiceS 5 to 8, thus effecting
the 'split'.
'change' reduces the VOICE numbers to the range 1 to 4. For
speed of operation, and to avoid problems with the value 0 generated for both
the LEVEL voice at all times and the GATE voice for an 'OFF GATE', the change
is made using a 'look-up table'.
This is an ARRAY, 'vtabl', which has 9 elements (including
element 0). 'change' takes each voice number and replaces it with the value
held as that numbered element in 'vtabl'. For element 0 to 4 this makes no
change, but elements 5 to 8 contain the values 1 to 4 again, so these voice
numbers are re-allocated.
Finally 'RUN' coordinates the procedure. It does a 'setup',
uses a word 'vset' to set up the 'look-up table' for VOICE numbering, allows
the user to 'select' a splitpoint, and then sets the split keyboard going.
4) In conclusion...
Obviously, the actual voice allocations in 'setup' can be
simply changed to any other instruments.
By changing the pitch played in each section, either by
using SHIFT in the instrument definition or by altering the pitch voice in ACT,
it is possible to play two melodies in the same register at the same time using
different instruments on the same keyboard (at least I presume that it is
possible - neither my brain nor my fingers could cope with it!).
I did originally write a more general version that allowed
for a definition of the total number of voices, and how they were allocated
between the two sections. However, the additional processing required led to a noticeable
slowing down of the response to the keyboard. Notes sounded just late enough
after pressing the keys for me to be aware of the delays, and block chords sounded
like fast arpeggios!
This version, with values fixed both by constants in the ACT
definition and by the contents of the look-up table, does not produce that much
of a delay in note processing - it is only just audible on block chords.
(My next essay may well be called 'Swings, Roundabouts and
Compromised Ideals').
Related file on this disc:
Published in AMPLINEX 007, September
1988
