Master's Thesis

Sine wave:
The dry test affected sine grain shape greatly - envelope shape can induce extra frequency modulation with different shapes clearly. Also can turn noisy at high grain speeds with randomised pitch - almost watery sound at very small grain size (less than 20ms). Long grains good for low pitches.

For the wet recording, distortion adds harmonic colouration. Delays add sense of layering and space - spectral complexity from multi-channel delay stacking. Super pure spectral swells - almost chime like. Stereo spread creating full soundscape. Short reverb, saw envelope, short delays, random filtering - creating complex timbre - bubbly texture. Reverb expanding long creates huge space. Random pure sine tones being resonated by the random filtering causes chord-like stacks of sound to ring out in the space. Delays adding repetition, lots of spectral gesture (up and down; resonant tones). Free panning with small reverb sound and long delay-line able to beam moments smoothly around the stereo field.

Noise:
Dry relatively limited. Grain speed and panning are the main effecting parameters here. Grain shape relatively negligible at high speeds as it approximates a noise signal anyway.

Wet is more sonically convincing with the added spatialisation, multi-fx signal routing, delay, and spectral manipulation of the effects. Resonant tones, lots of gesture, thick grain streams possible.

Drips:
Similar to sine in dry - noise at high grain speeds - can get icy sounding with the glassy tone of the source sound.

Wet - RAIN EXPLORATION - capable of starting very simple and pointillist, like a dripping tap. Pitch randomisation able to approximate organic dripping well. Can build to a thick watery stream mass with faster onset time, added delay, and reverbed swells. mc.delay to Degrade output to reverb can induce noisy clouds in background.

For all sounds, powerful preset controller enabled consistent parameter application across different sounds to see how the same parameter settings would affect different input signals.

Due to the shear scale of parameter values allowed in every parameter, the simple input sounds had elegantly varied outputs from the various presets I made. Sonic complexity is quite dependent on the number of modulators used and grain density, but a vast range of densities can be achieved - from pointilist sparse grains to very dense and varied drones/walls of sound, with dozens or even hundreds of grains playing at once each with discrete parameter values.

Sine source is heavily affected by grain size and envelope shape, and the intense grain streaming, intermittency, and async variations can induce noise or even amplitude modulation style sounds. The custom LFOs enable all kinds of pitch, grain, and filtering sounds, with cyclical and random LFOs to introduce cyclical patterns, random fluctuations, or chaotic parametric value blasting. Filtering resonance works best with filter frequency movement to make individual sine grain resonances heard more clearly (especially in a pitch spread texture).

Likewise, noise benefits from the vast parameter ranges - the highly microsonic ranges in grain properties introduce great variations of noise timbre. Filtering works particularly well at introducing focused streams of noise. Noise can easily fill up the stereo field in dense, stormy/cloud-like textures, and create very crackly variations with intermittent short grains. Comb filter-style sounds can be achieved with short grain and pitch/scan speed variations.

The intermittent audio content of the drips source seems to be a bit more difficult to immediately get lots of complex textures when there is scan spread. When focused on the ping of each drip though, the sound can be controlled in similar ways to the sine grains to get lots of pitch varied textures. Fast scan speed across the whole waveform enables more chaotic cloud mass variations to form as the silence intervals are smoothed over from fast reading.

Sound file switching enables textural composite sounds to emerge - fast fluctuations between sound files can create complex sound objects or clearly layered texture streams built from an even spread of sound content from the different sound inputs.

For these sine, noise, and drip studies, all adjustments were done manually by hand. Additional exploration of QLists could be done to automate much of this and create more extreme dynamic gestures, however, the following information is still a grounding point for the kinds of emergent sonic variations immediately possible:

Sine: static textures possible with different levels of chaos covering much of the spectral space - pure synchronous sine tones or chaotic, water-like flurries of mixed pitch and asynchronous sound. Can go endlessly low or high pitched. Grain sizes nearing 0ms can create added harmonic tone with trapezoid grain envelope. Gesture between different textures and dynamic movements can occur with individual range adjustments on the grain properties or window shape.

Noise: Like sine, static and dynamic textures possible with synchronicity or asynchronisity. Grain randomisation/min-max spread on grain density with maintained pitch and sharp grain envelopes can create tonal qualities. Wide pitch, size, gain, and density spread creates chaotic gaseous textures - stormy and airy. Long grains and high densities with different pitch variations creates more wind-like/ocean-like swells - wave breaching. This can be spacially exacerbated with wide panning variation.

Drips: Static texture with static pitch emphasises tonal qualities, while asynchronous and varied ranges creates different kinds of glassy/ceramic rain-drop textures. Lower pitches become more gong like - with static pitch and long dense grains, the drip attacks get lost in the texture. Sparseness achieved with much longer grains (where multiple drips occur in one grain). Hard panning left and right with the manual panner can introduce spatial centralisation, with scattering around it.

All together and presets (Predominance Grid): multiple sound object streams can play concurrently, creating rich composite textures (this is perfect for generating more elaborate layering of sound within single instrument interactions). Fading between them to get different average grain balances allows for multiple spectrums of texture. Shuffling between multiple presets enables very sudden textural juxtapositions of each composite textural space. Alternating back and forward between presets while adjusting the predominance values creates alternating gestures and interesting moments of sound-object repetition. Scrolling through presets quickly can produce very hectic sounds.

(note that more variations of emergence/texture/complexity could be explored with different voice counts - in all the above examples, 64 grain voices were used)

Spindrift Playground is a very capable granular tool for creating varied morphing musical textures, specialising in it's powerful spatial controls and filterbanking to enable complex or musically rich harmonic/inharmonic atmospheres and drifting trajectories of the sound mass. For each individual sound test, manual mouse control was used to hone in on the individual parameters. A final full sound test made use of presets, preset interpolation, and random grain animation for more dramatically varied textures. Each temporal grain parameter dial being skewed to the microscale pushes the user manipulation towards more microsonic granular manipulation (smaller grains, denser grain spacing). Grain voice overlapping is possible, with longer grains fading into each other at short spacings, but the sense of creating multiple textural streams of granular sound is difficult to achieve using simple samples (more complex audio input is needed to create enough grain variety to approach a sense of multiple grain 'layers'.

The synth encourages spectrally latticed musical emergence with its scale-preset filterbank system.


Sine:
The temporal and spatial parameters of the tool were able to provide the most variation in pure granulation. The sine tone sample was more limited in the pitch realm in the dry recording, only allowing a few octaves range maximum; but still, the pitch variance provides a range of static to very noisy/dynamic sine pitching. Both static and dynamic textures of high density are smooth and noisy - envelope shaping can increase or decrease the timbral harshness of the grain mass. At lower densities the sines can be clicky, sporadic, or regularly pulsing.

Adding the reverb and filterbanking opened up the spectral and spatial possiblities of the sine tones (interesting side note: the pureness of sine waves made the the filterbank easy to blow up in resonance at times (constructive addition of each filter-band resonance approaching single sine bandwidth), so had to be careful with levels). Density/temporal granular settings likewise had a strong effect on the filterbank and spatial response, with higher density gestures and motions leading to loud excitations of pitch and space. Resonance controlled the filterbank timbral quality, from pad-like long resonant tones to more percussive, woody notes. Adjusting the filterbank spread with inharmonic values and sliding the frequencies led to some more interesting atonal timbre variation outside of the grain pitch randomness control.

Noise:
As with sine, the density and time controls gave the most variation of the noise textures. The pitching was especially limited on the dry noise grains, only capable of producing a moderate range of noise darkness to brightness in the spectrum at absolute maximums. Rumbly or very high pitched noises were not immediately possible. Synchronous grain generation was a powerful timbre-colouring setting for the noise grains, turning a chaotic signal into thinner tone streams. Panning quite effective with the noise grains - as a spectrally busy signal, the perceived width and image localisation of individual grains or moving masses is quite clear. Envelope shaping with attack skew creates popping, crackling grains, and with decay skew creates swipier textures. Pitch randomness may be best heard here with sharp attack, discrete popping grains with audible spacing.

The filterbank and reverb are very responsive to the noise signal - with the noise spectrum being very full, any filterbank setting will get clear tone output from the noise grains, enabling highly spread and full tonal atmospheres to play. This is most noticeable with standard filterbank mode (every grain hits most of the filterbands). Random filter mode provides a rich liquid-like trickle; adjusting resonance and grain density/length can morph the tricklling mass into more smeared, airy tonal pads.

Drips:
As a dynamic audio source, Spindrift was able to find further temporal power in the dry grain generation. On top of the same time-based parameters controlled in the noise and sine grains, the drips jitter, read speed, and position/spread on the source file created lots of dynamic granular variations, from closely matching grain reproduction of the raw audio, to a wide gamut of temproal and pitch-controlled textures: noisy-pitched playback of waveform, smeared rhythm of waveform, dense clouds and airy currents, cyclical mass motions, super-trickly rainstick-style streams, sporadic chime like undulations, cascading pitch grains (synchronous grains, but random pitched, leading to cascading phase offset of each grain's chime pulsing).

The filterbank and reverb, like noise and sine, added good spectral and spatial atmospheres and variations. The drip waveform shape in the grains naturally emphasised the trickly sound of the random, low-resonance filterbank tones, leading to the most water-like musical sounds of the three audio samples. Interestingly, using a small, low-end inharmonic frequency spread on the filterbank leads to spring-like excitations of the grains.

All with Presets:
As with some of the other synths, the preset controls here provide a powerful form of dynamic trajectory generation and juxtaposing gestures. Interpolating between presets with smooth ramping creates strong parametric dives and ascents, while manual sliding and fast ramping increases force and chaos of change. Meanwhile, the animate mode randomly automates parameters to produce a slowly evolving constant cloud of sound. These preset controls, combined with a complex waveform, reverb, and filterbank variations, leads to a impreassive gallery of tonality-focused sculptural atmospheres and grain clouds.

Spindrift Synth, much like its sibling Spindrift Playground, can achieve some of the same granular sound spaces as Spindrift Playground; from a static texture persepective, its emergence is quite similar. However, because the synth is MIDI-note voice based (where a triggered midi note creates a new granular stream), parameter setting of a given voice can only happen before the onset of the note; the synth is thus somewhat limited in its ability to create fast changing textures and sonic emergences (still possible, but new notes need to be triggered every time). But in the face of this apparent lack of parameter control during note playback, the synth specialises in its ability to create impressively complex drone textures, enabling a large polyphony of inidividualised grain streams each with their own unique parameter settings. Thus, from a simple sound source, a very complex static (or slowly note-shifting) texture can be generated.

For Soundgrain, only 1 'all with presets' recording was made, as its sketch window control interface made it very appropriate to do waveform mixing through multiple granular trajectories. From these simple sound sources, sound grain is able to produce very dynamic, moving, and spectrally complex grain clouds. Multiple free-flowing trajectories looping in their own timeframes means that the collective grain texture is constantly phase-shifting in each grain trajectory gesture. Combined with global grain engine manipulations, such as using long grains and switching short grains can completely change the textural fill and timbre of grain clouds.

The recording builds up from a single grain stream, pitch-shifting and waveform shifting in a low density loop. Further trajectories are slowly added, bringing more textural, spectral, and dynamic thickness/intensity. After multiple streams are added, various global controls are manipulated, leading to numerous overarching texture changes as described above. This additive trajectory approach could be carefully constructed too, whereby loops may be drawn to last the same length, leading to more consistent, cyclical granular motions, with macro-stasis (but localised motion and activity)."