Another sporthling created for the AudioKit sporth editor. It features a Karplus-Strong pluck and a phase-vocoder buffer shuffler. The controls are the following:
- Control 1: Feedback
- Control 2: Tempo
- Control 3: Dry/Wet
- Control 4: Probability
A buffer for audio is created and zeroed out.
'buf' 80000 zeros
An argument table is created to store some parameters.
_tk 2 zeros
A table is created containing a musical sequence.
'seq' '0 2 7 11 14 4 2' gen_vals
tick will output a single trigger at the start of a sporth patch. It can only be used once, so the output must be copied somewhere else, in this case our table tk.
tick 0 _tk tset
Our clock consists of 3 parts: the tick signal, guaranteeing a trigger at the beginning of the patch,
0 _tk tget
the metronome signal, whose rate is set by p-register 1,
1 p 1 5 scale metro
and the metronome goes into the maytrig, whose probability is determined by p-register 3.
3 p 0.4 1.0 scale maytrig +
This clock signal is copied into index 1 of the table tk for later use, then duplicated for more imediate use. (It is also possible to use table instead of the stack, but the patch evolved this way.)
dup 1 _tk tset dup
The Karplus Strong pluck is the fundamental sound source in this patch. The other aspects of the patch process this sound.
The clock signal from the previous section triggers a tseq, whose mode is 0, which means it will move sequentially through the table.
0 'seq' tseq
This sequence is biased to 61 (Db major ish), and converted to a frequency via mtof.
61 + mtof
The rest of the pluck module is pretty typical. It is put through a lowpass filter to smooth out the pluck sound a bit.
0.9 400 pluck 1000 butlp
The pluck is split and put into a static delay line, whose feedback is determined by p-register 0.
dup 0 p 0.01 0.9 scale 0.01 port 0.75 delay
The output of the delay is put through a highpass filter and attenuated a little bit. Then, it is mixed back into the dry signal.
1000 buthp 0.7 * +
The pluck is fed into the zita-reverb with a very long decay time. A dc blocker is put on the end of it for good measure.
dup dup 10 10 8000 zrev drop 0.5 * + dcblk
The entire signal created so far is sent into tblrec, which "records" the audio input into the buffer. The tick signal created earlier is needed here for tblrec, which will leave it recording indefinitely.
dup 0 _tk tget 'buf' tblrec
Mincer is a phase vocoder, which allows control of position and pitch of an ftable independent of one another. In this case, the ftable being used is the buffer, which is being constantly rewritten via tblrec
Wavetable position is a slow random walk.
0 'buf' tbldur 0.3 randi
Ampltiude is 1. No attenuation.
Next argument is the playback rate. The clock signal drives a random number generator, which picks values between 1 and 3. These numbers are truncated to be integers via floor. This is done so the play back rate can be one of three states: normal (1), one octave up (2), and one octave and fifth up (3). Restricting these states makes things sound more pretty and in key.
1 _tk tget 1 3 trand floor
The last argument of mincer is the buffer.
The output of mincer is sent into this chorusing delay effect. It is a feedback delay sent through a (surprise) chorus made up of pitch shifters.
The feedback delay has a somewhat high feedback, and a very long delay time.
dup 0.7 1.5 delay 0.7 *
The output of this delay is put into two pitch shifters in parallel. The first pitch shifter raises the pitch by 0.11 semitones, the second drops the pitch by 0.1 semitones. The dry delayline signal and pitch shifters are then mixed together.
dup 0.11 1024 512 pshift rot dup -0.1 1024 512 pshift + +
+ -10 ampdb *
At this point, there are two items on the stack: the pluck signal, and the chorused mincer signal. A crossfade is used to switch between these two signals. The position of this crossfade is determined by p-register 2. The subtraction is needed to flip the signals. The final signal is duplicated to work on a stero signal.
1 2 p 0.01 port - cf