# Scheale # By Paul Batchelor # August 2015
As the date in original comment header would suggest, Scheale was one of the first compositions written in Sporth. Scheale was written in the mountains of Pennsylvania on an old thinkpad laptop while overlooking Ganoga lake. As it so happens, I find myself writing this documentation up, sitting in the same place a year later.
Many things have changed since this piece was written. Maygate, for instance used to be combination of maytrig and maygate. Arithmetic opererations would words like mul, div, add, sub and not /, *, +, and -. Clip has gone through some fixes as well. Nevertheless, the essence of Scheale remains the same.
This patch is the main component of Scheale. The entirety of the composition contains two Sporth patches and nature samples recorded where this piece was mostly written.
The tables used in this piece contain two pitch sets, which are swtiched back and forth between one another. The first sequence contain the pitches D, A, E, A.
"seq_1" "62 69 64 69" gen_vals
The second sequence contains pitches C sharp, E, F sharp, and B.
"seq_2" "73 76 66 71" gen_vals
The main clock source used is the prop ugen. Prop was a micro-notation language I wrote prior to writing Sporth. I have since then created a port of in to exist inside Soundpipe and Sporth. The rhythm used is "+2(++)--", which can be read is a quarter, two eigths, followed by two quarter rests. The tempo is set to a rather slow 54 BPM.
The output of clock is sent through a branch. This was initially done in order to add a delay to the start of the patch in the larger composition. Since this is only one part of the composition, the branch has been permantly set to choose the prop clock source.
1 54 "+2(++)--" prop 0 branch
The output clock signal is put through a maytrig, which makes the clock trigger 78 percent of the time. It is then duplicated on the stack.
0.78 maytrig dup
The main envelope used is tenv, a triggered linear envelope generator. the release time of the envelope is randomized via randi to be between 500 and 800 milliseconds.
0.004 0.05 0.5 0.8 0.2 randi tenv
After tenv, the clock signal is brought back to the top of the stack and duplicated, so it can be used in the modal filters.
The main means of sound generation uses a technique known as modal synthesis. An impulse (created from the clock) is sent through a series of modal filters, where resonances are added to create percusive sounds that are somewhere between metallic and glass. The first series of modal filters here is used to create an excitation signal. Later on, they will be fed into another series of modal filters to create a sounding pitch.
dup 1000 12 mode swap 3000 8 mode add 0.3 mul dup
The sequencing part of Scheale is by far the trickiest thing to understand, due to the heavy use of stack operations (tables and variables did not exist at this time).
At its core, the sequencer is made up of two tseqs being fed by the clock signal in parallel. The stack operations used are meant to bring shuffle the clock signal to tseq.
rot dup dup 0 "seq_2" tseq swap dup 0 "seq_1" tseq
At this point, the stack has now been prepared in such a way so that the last three items look like this: tseq1, clock, tseq2. The clock is swapped into a 50-50 maytrig, which becomes the trigger that switches between the two sequences via switch. Because trig is the first argument, and not the last, two rots are ended to align the arguments correctly.
swap 0.5 maytrig rot rot switch
The clock signal (yes, there is still a copy of the clock signal on the stack), is used again to drive a counter that counts from zero to two. Every time it reaches 0, 12 is added to the sequence. In other words, every three notes the sequence leaps an octave.
swap 3 0 count 0 eq 12 0 branch add
To add some glissando in between the notes, a portamento filter is added to the signal before they are fed into the modal filters.
More modal filters
Now that an excitation signal and a sequence has been created, these can be fed into the second series of modal filters which will create the pitched sounds.
Slight jitter is added onto the note sequence before it is converted to a frequency. This is a crude attempt to add the "drift" commonly experienced in analogue hardware.
-0.15 0.15 0.2 randi add
The midi note number is converted to frequency. The stack is shuffled in such a way so that the arguments line up for the first modal filter, and so there is also a copy of the frequency on the stack. The Q of the filter is randomized via randi.
mtof dup rot swap 300 2000 0.12 randi mode
Once again the arguments on the stack are aligned for the second modal filter. The Q of this filter is also randomized with a different randi.
swap rot swap 2.01081 mul 100 1000 0.1 randi mode
The two modal filters are summed together, then scaled down, as they are very very loud.
add 0.15 mul mul
The effects used consist of clip distortion, a feedback filtered delay line, and a very large reverb.
Firstly, the sound is put through clip distortion, highly truncating the waveform to give it grit. This is then dulled via a butterworth lowpass filter.
0.3 clip 1000 butlp
This signal is fed through a delay line of 750 milliseconds. It is attenuated and fed through another 1000Hz lowpass filter.
dup 0.76 0.75 delay 0.2 mul 1000 butlp
The dry undelayed signal is fed through reverbsc. Both channels of reverbsc are used, so they are both equally attenuated.
swap dup dup 0.97 10000 revsc 0.3 mul swap 0.3 mul
From this point forward, the signal being treated is stereo. For this reason, many stack operations are used. At the end of the day, what these stack operations are doing is mixing the dry signal and delayed reverb signal together.
This is the first channel.
rot dup rot add rot rot add
This is the second channel.
rot dup rot add rot rot add
For now, all sporth cookbook patches are mono, so we will drop a channel.
That being said, the reverb sounds very good in stereo! The end.