Granular Synthesis for Rookies (Special Guest Writer Simon Stockhausen)
Simon Stockhausen is a formidable sound designer. He generously shares his expertise in the field of granular synthesis in the in-depth tutorial (with videos!).
SoundBytes Magazine is delighted to have Simon Stockhausen as a guest writer for this issue. Simon is probably best known to our audience as a noted sound designer (going under the name of Sampleconstruct on the KVR forum and elsewhere). He has designed sounds for a number of synthesizers including Alchemy, Chromaphone, HALion 5, IRIS and more than a dozen others. His work is available on his website http://www.patchpool.de/.
Simon does far more than sound design. He is a prolific composer for concert hall and soundtrack and is an accomplished musician as well. But today he dons his sound designer hat to explain what to many of us will be the exotic topic of granular synthesis. Since his most recent sound set embraces granular synthesis in a preset library for HALion 5, one of the more capable granular synthesis instruments available today, his expertise is all that more to be valued.
Granular (re)synthesis still seems to be a somewhat exotic form of manipulating audio as the knowledge about it is not as widespread as the knowledge about subtractive synthesis, for example. This article tries to shed some light on the basic principles of this very versatile and interesting form of synthesis.
Composers, audio scientists and other audio abusers in the stone-age of electronic music were stuck with tape machines when they wanted to manipulate sounds like natural instruments, vocal textures and field recordings. When they wanted to transpose their audio up or down several octaves they had to speed up or slow down the tape machine and then copy the result to another tape machine. This procedure resulted in the chipmunk-effect so typical for traditional sampling, as the faster the tape would run the higher the pitch of the audio would be, the slower, the lower the pitch would drop. In the early 1960s the first hardware granulator Tempophon was built by the German company Springer. This device allowed to transpose the pitch without changing time and vice versa by using a specially designed rotating tape head in combination with transition-free tape speed-control and it was based on the principles of Dennis Gabor, the father of granular synthesis.
Granular synthesis detaches the parameters time and pitch, the audio information is dissolved into many more or less small grain particles which can then be manipulated in numerous ways. As these two parameters are detached from each other, transposing audio will not yield the chipmunk-effect, at extreme transpositions other audio artifacts will be introduced though, depending on the algorithm used in a particular device or application. So time stretching and time compression are basic aspects of this synthesis form, but there is much more to it. Creating more or less dense clouds of granular audio particles is a further aspect, but let’s look at some basics first.
This determines the point in the audio file at which the grains are created. When grain Speed is set to “0“ the audio is “frozen“ at that position, with positive speed values the audio stream plays forward, negative values reverse the audio. Grain Position can also be modulated, e.g. with an LFO or an envelope; in this case the “Speed” value is usually set to 0 so that the direction and speed of the grain stream is only determined by the modulator. If the modulator, e.g. an LFO with a ramp up shape is synced to a certain tempo-synced length – for example 4 bars – the modulation of Grain Position can also be used to sync musical loops and the like to the tempo of a given project. The LFO will then scan through the sample once within 4 bars. Most granulators also offer a function for randomizing Grain Position, so that grains are created at random positions in the audio file depending on the range of the randomization. Some granulators like HALion 5 or Padshop Pro allow for multiple grain streams occurring simultaneously, these streams can also run with a position offset, each stream scanning the audio at different points of the processed file making for extremely dense and rich grain clouds, especially when slightly detuned. In granular synths the user can also modulate the grain position via Aftertouch, so when, for example, the sample of a musical phrase played on a duduk is used, Aftertouch will scan through the phrase making for some expressive solo playing.
As already mentioned above this parameter determines at what speed the sample “play head” runs through the sample. Some granulators use positive/negative speed values to also determine the direction of the grain stream (forwards/backwards), other granulators like HALion 5 have a dedicated direction parameter. Interesting results can be achieved with Grain Speed set to 0, and then only slightly modulating Grain Position with an LFO, so that the audio is frozen but still has some movement inside, like a spaceship gently hovering over a landing point.
The definition of this parameter also varies between the different apps. The shorter the duration of the grain size, the less distinct the original pitch is audible. With sizes below 50ms the grain stream turns into noise without a distinct pitch; very short grain sizes below 10ms in combination with some randomization of grain position create more white-noise-like effects; without randomization sounds reminding of ring-modulation can be achieved. If the grain duration is longer than the intervals between the creation of each grain, a sustained sound will be created; if it is smaller the sound gets perforated/fragmented. Some apps like crusherX allow Grain lengths of up to 20 seconds yielding similar results to an advanced delay unit with very long delay times and several delay lines.
This determines how many grains are created over a given period of time – the higher the density the more continuous the sound will become. In some apps this parameter is also defined as “grain birth” determining how often a new grain is born, a nice analogy…
Most granulators offer a variety of shapes applied to the amplitude of each grain. The most common shapes are Hann, Hamming, Triangle, Sine, Rectangle, Blackman Harris, Welsh. Generally speaking, smoother waveforms with round curves create more organic sounding results, while shapes with more right angled forms create more edgy and distorted tones. Some apps like crusherX also let the user draw the shape. The more complex the shape, the more distorted the grain stream will sound.
This parameter determines the pitch of each grain, Grain Pitch can also be randomized within a given range making for some very interesting chaotic grain pitch clouds, or they can be pitch-quantized to certain scales like pentatonic, major, minor, diminished, dorian, etc. which produces tonal grain clouds. Most apps allow a more or less complex modulation of the amount of pitch modulation, so one can for example use the modulation wheel in a granular synth to introduce subtle grain detuning, similar to chorus FX with the wheel slightly up and totally chaotic pitch randomness over several octaves with the wheel fully engaged.
I have made videos explaining these basic functions using various apps which may explain all this more clearly. They are embedded below for your convenience.
Today we find numerous granulators, either as playable synthesizers, where the user can load a single sample or multi-sampled key maps and play the grain stream on a keyboard, or as FX plugins where the incoming audio is processed in real time. The advantage of granulating multi-sampled instruments with several velocity layers, variations per sample (round robin) and several sampled notes per octave in the case of a melody or chord instrument is obvious, as there are no artifacts occurring due to extreme pitch transpositions. So the expressiveness of traditional multisampling is combined with the versatility of granular synthesis, a field I find particularly interesting.
I hope to have given the reader some insights into granular synthesis and I can only encourage electronic musicians out there to grab a field recorder, record your environment and start granulating away. If you play a musical instrument, just hit record, improvise for a while and then start granulating the result, this will certainly lead you to sonic places that you haven’t explored before. With the advanced music technology of today and the powerful processors available in modern computers, granular synthesis has become a very powerful synthesis form with an infinite range of results ranging from experimental, scientific, strange and alienated to organic, smooth and very musical.
Dennis Gabor – 1946 – Theory Of Communication.
Curtis Roads – 2002 – Microsound