Midnight Crawl
by Paul Batchelor

Midnight crawl is a prickly warm little patch featuring string resonators, and warm FM and subtractive saw sounds.

Tables and variables

• Table seq is the sequence for the warm saw sound.
• Table prog is used together with seq to build a "progression"
• Table seq2 is used to sequence the warm fm sound.
• Variable clk is used to store the clock signal
• Variable send is used as a send signal for the delay.
• Variable env is used to store an envelope signal

 _seq "0 2 7" gen_vals 
 _prog "0 2 5 7 9" gen_vals
 _seq2 "0 2 7 11 14 19" gen_vals
 _clk var
 _send var 
 _env var

The Clock

First, the clock signal is created. The clock is generated using dmetro. It is set to 125 BPM, then multiplied by 4 to get sixteenth note subdivisions. The clock is sent into a maytrig with an 80% probability rate. All of this is assigned to the variable clk.

 125 4 * bpm2dur dmetro 0.8 maytrig _clk set 

Warm Saw Sound

The warm sound saw is a sequenced subtractive sawtooth patch. The sequence is created via tseq, using the values inside of seq, and the clock clk.

 _clk get 0 _seq tseq 58 + 

To add a sense of a chord progression, another "progression" sequence is added to the note as a bias. The same clk signal is used, but it is fed into clock divider tdiv so that the "chord" changes every 4 notes. To keep it from changing that frequently, the clock signal is fed into a maytrig.

 _clk get 4 0 tdiv 0.1 maytrig 1 _prog tseq + 

To add occasional octave leaps, maygate scaled by 12 is added into the sequence signal.

 12 _clk get 0.2 maygate * + 

To simulate analogue "drift", jitter is added to the sequenced signal. This entire signal is converted to frequency via mtof.

 0.1 0.1 5 jitter + mtof 

Portamento between frequencies is controlled via port. The portamento time is randomized via randi to pick a value between 1 and 20 milliseconds every 5 seconds (inv efficiently inverts a number).

 0.001 0.02 (5 inv) randi port 

With the frequency set, the amplitude of the saw is set to 0.4.

 0.4 saw 

The sawtooth signal is fed into butterworth lowpass filter butlp. The cutoff frequency of the filter is determined via a sinusoidal LFO sine, modulating between 200 and 800 Hz.

 15 inv 1 sine 200 800 biscale butlp

The amplitude of the filtered saw signal is modulated by envelope generator tenv. The envelope used is very long, and is meant to be the contour for many notes at a given time. Since the clock signal is way too fast for this, it is sent through maytrig.

 _clk get 0.05 maytrig  1.4 1.1 2.3 tenv *

Streson Prickles

Now for the instrument I like to call "streson prickles". These are the most novel sounds of the patch, created using a series of string resonators. The start of this sound are short little noise bursts, created from clk, maytrig, tenvx, and noise.

 _clk get 0.7 maytrig 0.001 0.001 0.001 tenvx 1.9 noise * 

Streson, is a string resonator. When fed an impulse (like from the noise bursts ) it will produce a karplus strong pluck sound. In addition to the input signal, there are two arguments to streson: the filter frequency determines the frequency of the string, and the feedback amount determines how long the note stays on for. The first string resonator has both the frequency and feedback parameters being randomized by clk-synchronized random number generators trand.

 _clk get 100 1000 trand _clk get 0.8 0.9 trand streson 

This signal is fed through two other string resonators in series. The decay times are set to be constants at 0.9, and 0.8, respectively.

 _clk get 400 4000 trand 0.9 streson 
 _clk get 100 1000 trand 0.8 streson  

To keep samples numerically reasonable, this signal is fed through a DC blocker dcblk and attenuated by 10 dB. Some of the high end is shaved off using a lowpass filter butlp.

 dcblk -10 ampdb * 2500 butlp 

This sound is complete, but copy of it is sent to the variable send to be used as a throw signal. The throw mechanism below is built out of a maygate, tdiv, and port filter.

 dup _clk get 4 0 tdiv 0.1 maygate 0.01 port * _send set +

Pinging FM Sound

The final instrument in the mix is a pinging FM sound. It is a sound buried in the mix. As the trigger signal will show, it is very sparse, having clk being set through a very large clock divider and maytrig.

 _clk get 16 0 tdiv 0.5 maytrig 

A copy of the signal maytrig is made with dup, and it is fed into the envelope generator tenvx. This generator sets the envelope env.

 dup 0.005 0.01 0.3 tenvx _env set

The clock signal that was dupd before feeds into a tseq, where it is then biased by 58. When I wrote this patch, I wanted to bump this synth up an octave, so I added 12 to it too.

 1 _seq2 tseq 58 12 + + mtof 

The amplitude of the FM synth is is modulated by a slow moving LFO.

 13 inv 1 sine 0.01 0.2 biscale 

The C:M ratio is set to a typically very bright 2:5.

 2 5 

The modulation index is modulated by envelope stored in env. This basically controls "brightness" in the FM world. This is the final argument in fm, the sound generator.

 _env get 5 * fm 

In addition to manipulating the timbre, env also controls the amplitude. this signal is then fed into butlp.

 _env get * 1500 butlp 

The instrument signal generated is copied and added onto the send signal, which will be sent into some effects.

 dup 0.3 * _send get + _send set +

Effects

The first effect used in this patch is a feedback delay, with a very large feedback parameter. The delay time is based on the BPM, and is set to approximately dotted quarters. I say "approximately" because I made it slightly longer to make it go out of phase with the dry signal.

 dup 0.9 125 bpm2dur 1.51 * delay 2000 butlp -8 ampdb * + 

Next, the send signals are processed. First, send is fed into revsc. It is fed into a dc blocker dcblk, and a high pass filter buthp.

 _send get dup 0.98 5550 revsc drop 0.9 * dcblk 200 buthp + 

The send signal is fed into a tempo-synced feedback delay line delay. It is attenuated and then added into the rest of the patch.

 _send get 0.3 125 bpm2dur delay -3 ampdb * +