Chained Delays
P Todd Decker wrote:
This message is directed to Rob Hordijk but I posted it on the list because others might be interested as well.
Rob--I noticed on many of your patches you use chained delays. Most seem to be set to zero and only a few looked like the delay time was modulated. Why do you do this? What am I missing?
Rob Hordijk wrote:
To be short: If the delaytime is set to 0 ms undoubtedly the delaylines are tuned by a keyboardscaler module. And if the modulating input is unconnected and the delaytime is set to some value its to create a resonance at some frequency or maybe create a stereo 'ambience'-type effect.
To be less short:
As most will understand the delay module is not meant for echo-type delays and reverbs, its way to short for that. But it can be used with much pleasure for maybe less known synthesis and soundshaping methods, like physical modelling. (This remark basically for readers on this list that do not have a nord modular, but maybe plan to get one) So most of the time the delaylines are used (by me at least) to create a delay within a single period of a waveform to create modulation, some resonance or the combination of both into the waveform.
There have been very valuable contributions to this list about this subject by several people and there's a number of interesting patches in the archives. But as the Nord is a modular synthesizer everybody creates his own recipies, and I will tell you a bit about mine. First I must say that I'm not interested in playing accurate physical models of existing acoustic instruments. Instead, I'm interested in physical models of objects non-existing in the real world. Simply much more fun.
If the modulation input is connected to a keyboard scaler module, setting the delaytime to 0 ms and opening the modulation input full, the delay lines are tuned to the keyboard. Adding some sort of feedback to the delayline means you can play melodies in a chromatic scale. With a cascade of six delay modules you can go as low as C2. This basic principle you can use to make chromatically tuned resonating bodies or physical models. The difference between the two is that a resonating body does not determine the pitch of the sound, but only the timbre (its more like a complex filter) and in a physical model the delaytime does determine the pitch. So in a physical model of a plucked or bowed string or a tube-based instrument, you actually tune the instrument by changing the delaytime of the delay modules. E.g. with a plucked string the input signal is actually noise (unpitched) and the delay determines the pitch. With a resonating body the output signal inherits the pitch of the input signal. Some physical models can of course be a combination of both, the border between physical models and resonant bodies is pretty vague. Well, at least that's my definitions, maybe not really like the textbooks, but very practical definitions for my personal work. One could also say that a resonant body is the physical model of only the body of an instrument like a violin's body, and a physical modelling instrument is both the resonant body object and the pitch generating objects in one model, so violinbody, -strings and bow.
The attached patch PhysicalTrumpet is an example of a model where the cascade of delaylines is on the brink of oscillating. By feeding a little energy from a sinewave oscillator into the resonating body a complex soundwave starts to build up. The delaylines are tuned to the sinewave oscillator so the complex wave stays reasonably stable. The buildup of harmonics in the wave is generated by ringmodulating the feedback signal with the input sinewave.
Damping is done by adding the last two samples in the delayline cascade together and divide them by two and use this for feedback signal. This is the original '
Karplus-Strong' method.Filters like the 6dB lowpass set to a cutoff frequency between 4kHz and 20kHz can also be used, but changing the cutoff frequency also changes the total delaytime of the cascade slightly, resulting in sometimes unstable or detuned instruments (which can be nice by itself as in the FM bleep patch, that exhibits true 'chaotic' behaviour in the small resonating element, tweak knob one of the mixer, the delaytime and the FilterD1 cutoff frequency). 12 dB and 24 dB filters with a Q-setting different from 0 can easily lead to unwanted oscillation and delayline overflow, as Q (resonance setting on a filter) might be slightly different over the frequencyrange of the filter. Also they add significantly to the delaytime with lower cutoff frequencies, making the tuning much more complicated.
An extra example of use of the delay lines is added in the trumpet patch, to add a stereo panning effect to the sound, a little bit like leslie. This type of panning, that does not influence the volume but the phase relation between the L and R channel, gives a full spatial effect. Regrettably its not mono-compatible, so it shouldn't be used on a recording that might be broadcasted. But its a good panning effect on stage, with the advantage over volume-panning that the audience close to a speakercabinet will hear the instrument with a constant volume, instead of with a tremolo as with volume-panning.
Hope this solves some of the mysteries.
Len Sasso wrote:
Wonderful explanation Rob, thanks!
Here is the result of a little playing around - knob 1 morphs to a 'french horn', knobs 3,5 & 6 control the 'spit tone'.