Any digital algorithm can be described as a filter, for at a basic level a filter is any process which in some way affects the amplitudes and/or phase values of the input's spectral components. More specifically, a filter can be thought of as an algorithm that resonates or attenuates a particular area or areas of the spectrum. The psychoacoustic side-effects of this are largely dependent on the sound itself, but one can make some broad generalisations. The problems that are experienced in quantifying the aural effects of the filter are mentioned by Curtis Roads:
The theory of digital filtering is a rarefied speciality, couched in mathematics that are removed from human experience. The equation of a digital filter, for example, does not necessarily reveal its audio qualities. This is unfortunate, since perception and emotion are sharply attuned to the effects of filters. The profound subject of the aesthetics of filtering is seldom addressed in the signal-processing literature... even though the impact of filters on musical sound can range from the sublime to the horrific.
However, I believe it pertinent to begin making some generalisations, backing them up with musical examples, where possible. Firstly, a low-pass filter can have the effect of listening to a sound source in some medium with a higher density than air, or imbue it with a feeling of great distance, where higher frequencies have been absorbed by the intervening air. This filtered sound if heard on its own may place the listener in some distorted perspective especially where the sound is referential. Basic low-pass filtering may then be seen as initiating interplay on an environmental indicative field. (Sound Example 3).
A time-varying filter, however, is more problematical, in that it is often hard for the listener to decide whether it is the environment that is changing or the object itself. As can be seen I have drawn lines of influence from the spectral typology and morphology levels to both environment and object/substance. This ambiguity becomes important in creating interplay on both an object/substance field and the environment field as in the next sound example, from my recent composition Ombrisages. The highly shredded vocal source rises in tessitura until it whites out (a similar technique found in Tongues of Fire, discussed later in the text). A time-varying filter is applied to this, until it transforms into the deep water resonances, such that the object seems to grow and 'submerge' the listener (Sound Example 4).
A similar example from Chrysalis: firstly, the noise-based source is heard, perhaps wooden in substance. Following that is the same sound with a time-varying filter applied, the overall filtering becomes more and more flat, eventually articulating the higher frequencies. There is a certain amount of ambiguity as to whether the sound is changing or the listener's spatial or environmental relationship to the sound is changing (Sound Example 5).
If the time-varying filter changes rapidly enough, or is offset by some sound which is filtered quite differently, then we will hear the filtering as being a perceptual part of the sound's own identity, not the environment's. We most often equate this sort of rapid filtering with a change in vocal formant (an area of the spectrum that is resonated)-I shall deal with the manipulation of vocal formants in my discussion of source-filter cross synthesis.
Trevor Wishart: Tongues of Fire
Trevor Wishart is one of the few composers who freely documents their studio techniques; in fact, his book Audible Design is largely devoted to the processes and ideas that went into the composition of his piece Tongues of Fire. Two examples from Tongues of Fire demonstrate the way in which he utilises filterbanks to traverse the note-to-noise continuum. The first example demonstrates a continually shredding texture, which is shredded to the point of 'white-out' (ie approaching white noise). This noise, however, is also treated to a successively narrowing filterbank, thus focusing on pitched features and pitch trajectories. This section (from 17:10) demonstrates how the pitch material gradually takes over in perceptual focus from the noise-based material. The second example demonstrates the use of a static filterbank to achieve a similar traverse (18:30). Wishart equates this interplay on the note-to-noise continuum with the classical notion of modulation; rather than changing keys, however, we are moving into a new sonic area, one where the pitched features of the sounds become new sonic foci.
Creating transformations on the amplitude morphology of a sound, I have put under the general heading of 'Enveloping'. Enveloping a sound can be very powerful, especially in initiating change on the energy/motion and concurrently on the gesture field. One special case of the enveloping algorithms, is the process known as envelope substitution, which analyses the morphology of one sound and applies it to another. In order to effect this substitution, we must first normalise the amplitude envelope of the second sound, and then multiply the envelope of the first by the second. By substituting amplitude morphologies, we are theoretically changing the gesture and energy characteristics of the second sound; in practice, however, this will only sound convincing if the second sound has stability characteristics of the first. For example, a noise-based will sound fine if enveloped by a rapid corrugation, whereas a stable pitched sound is rather unattractive. Another problem with envelope substitution is that it may try to amplify low-level noise to a considerably loud volume. The result of this is that when using this process, we should bear in mind that the first sound (the filtering sound) should have a strong, characteristic amplitude morphology, while the second sound (the filtered sound) should have a fairly flat amplitude morphology to avoid noise artifacts, and thus reduce the perceivability of this very useful algorithm. This process is essentially a time domain analogy to the source-filter cross synthesis algorithm, which I will discuss this algorithm later in the text. However, first I will play four extracts. Firstly the sound whose morphology I wish to capture. Secondly will be a flatter noise-based sound, to which the morphology will be applied. Thirdly, the result of the substitution. After this will be the same substitution, but onto a more stable sound, which is less effective. (Sound Example 6)
A delay line is most often used in a sort of feedback loop-that is to say, the input is delayed by n samples, multiplied by some gain factor g, and then added to the next input signal. This is known as the recursive comb filter. The essential essence of a delay line is that we hear some portion of the input sound again after it has been sounded. At large enough time increments, we can hear these as echoes, while at smaller increments we tend to hear them as a reverberatory effect.
Reverberation is important for its ability to initiate interplay on the spatiomorphological level; with differing amounts of reverberation, we are able to place a sound in terms of spatial depth. For example, this sequence from Ombrisages: (Sound Example 7). We must not be confined to thinking that reverberation is useful for spatiomorphology only; it can also create a continuous amplitude morphology out of one that is transient in nature. This new continuous morphology may then become the basis for further transformation. The next example demonstrates how a short phrase played on the flute is elongated by reverberation. The reverb is then enveloped, changing the gesture percept to that of a rubbed trajectory, and finally this is further processed and its percept changed completely. (Sound Example 8)
Márta Grabócz senses, though doesn't completely satisfactorily articulate, the distancing in both time and space that tends to occur with a delay-line based process:
Thanks to utilization of echoes, decays, loops, filters, and other electroacoustic transformations and special effects, the compositions... stress and important characteristic of their narration: this technology attributes to these compositions an aura, a distinctive atmosphere that creates a separation in time between the present (the moment of the writing of the work) and the past (the period of the story told). This aura may evoke either an ironic, nostalgic tone, and atmosphere of alienation, or, contrarily, of empathy.
Curtis Roads notes a rather more useful approach to the effect of reverberation, that of articulating sound-objects in a spatial manner:
...space is not merely a cosmetic appliqué for sounds. Spatial depth can be used to isolate foreground and background elements in a compositional architecture. Furthermore, reverberation is not a monolithic effect; there are many colors and qualities of reverberation-as many as there are natural spaces and synthetic reverberators.
Certainly, there can be a sense of nostalgia: a distance not only in space, but in time. For this reason, musical extracts, when put through sufficient reverb, always seem to have the effect of a half-remembered phrase.
Brassage is the name given to encompass all types of time-domain fragmentation. Segments of the input sound are 'lifted out', then perhaps processed, before 'repasting' into a new sound file. The manner in which this occurs varies from application to application; some possible parameters include: manner of fragment redistribution (linear or random), pitch-changing of fragments, gain-changing, panning of fragments, changing fragment size, changing 'hop size', etc. All of these parameters greatly affect the final sonic percept. For example, small fragment size, small 'hop size', and linear ('synchronous') redistribution can create a time-stretching effect if so desired, whereas the same algorithm, but with a random ('asynchronous') redistribution instead creates a cloud-like percept, similar to "those created in the visual domain by particle synthesis".
This short work from French Canadian composer Robert Normandeau is essentially a brassage vignette. It is a short work utilising a single sound source and a single transformation technique. It is important to notice that the brassage technique does little to disrupt the source recognition-we can generally distinguish the voice of the young girl. The fragmentary nature of the sounds, however, betray the underlying sense of unrest which Normandeau wishes to capture, a sense of anger or confusion. This is rather similar to the effect that Wishart seems to capture, of the "slightly angry, disgruntled, comic mutterings". One must say that this certainly seems to be the initial affective response to the paralinguistic result of such vocal fragmentation. This effect, however, has deeper resonances with the shift from unification to a fragmented pluralism that has dominated artistic thought over the twentieth century.
Christian Calon: La Disparition
Christian Calon, a French Canadian composer now living in Berlin, enjoys exploiting the extrinsic shifts resulting from sound-fragmentation. An particularly arresting example of this is La Disparition (1988) which provides an interesting subject from several perspectives: firstly, he uses 'musical' extracts (from Beethoven's Grosse Fuge, to traditional music of Africa, Melanesia and the Far East) as well as environmental sounds as his brassage sources, and secondly, the fragmentation is achieved with a programmatic agenda:
Rooted in this work lies the desire to hear those multiple and deep voices, that now belong to what we call History, rise together in song. Through the sometimes violent embrace of sound materials, as distant and far away as our great music can be from the traditional music of other civilisations, one will recognize this vain desire to break the wall of silence, of the erasure, of the dissappearance (sic).
Through the fragmentation of 'musical' extracts, then, Calon achieves a shift in emphasis from the sound-symbol (coded expression) to that of the sound-object-that is to say, the extract, divorced of its context, through its brevity and repetition, and often through its unusual reverberatory setting or pitch shift, becomes objectified rather than playing a linear part in some musical discourse. This objectification is further supported through the interplay between an aural and mimetic syntactical discourse, which I covered earlier. For example, in the Présentation (0:00-1:35), the flute sound from the (presumably) Melanesian extract is removed from its context and treated in various ways: pitch shifts, time-stretches, filtrations, spatialisations etc. which endow the opening minute with the sounds of breath and of pitch. While the sounds a reasonably recognisable, they are treated in such a way as to emphasise the aurally discursive nature of the sounds. At 1:00 some high metallic resonances quietly enter; both the flute-sounds and the metal-sounds prepare us (aurally, at least) for the introduction of the first sizeable musical fragment at 1:35, which comprises the metallophones and flute. Furthermore, the metaphorical associations with both a codified syntax and a purer sonic syntax come into play at, for example, 14:04 (La Mort) where the traditional affective vocabulary of the Grosse Fuge suggests the 'death-like' qualities of the section title; the progressive fragmentation/decay of this extract, however, suggests a more far-reaching metaphor of the impermanence of all things.
Brassage is also used in La Disparition for transformative effects: at 2:28 the highly 'shredded' material, with short fragment time and high pitch variance, while still being in the sonic context of the co-existent fragmentary interjections and drones, does in fact pre-empt the water sounds which enter at 3:40. This percept transformation occurs again at 7:00, where the water sound is followed by another high brassage texture, resembling the stochastic play of high partials in the water sound. Out of this texture, however, grows the sounds of the cicadas, providing a link between the texture-based environmental sounds and the progressive fragmentation of the sound material.
The progressive transformation from fragmentary texture to identifiable extracts is used in both La Disparition and Wishart's Tongues of Fire. At 6:00 in La Disparition, the noise-bands dynamically transform into a sort of rhythmic counterpoint to relatively literal rendering of the Grosse Fuge through progressive fragment-lengthening. In Tongues of Fire this occurs in both directions (Example 8.17 from Audible Design). Calon describes the process of brassage in his own unique, and somewhat prosaic, manner:
The orchestral and heterophonic quality which can define stylistically this work has been reached by following this compositional method: from the sound materials of environmental and musical origin ... short Elements have been extracted. These sound elements were broken down into Fragments. These related-sounds families were transformed and accumulated into Clouds. These clouds of multiple materials were finally mixed to form the Starry Wheel through which hierarchical connections are then made possible in all directions.
Calon subscribes to the idea of a 'web of interrelations' in his music by presenting the sound-sources in a roughly egalitarian manner, and allowing the listener to form the 'hierarchical connections'. He is not adverse to making a few metaphorical 'guidelines' along the way, however. For example, in the section Le Sommeil, Calon juxtaposes the environmental sounds with the sounds of the voice (7:54), which is interrupted by what sounds like the droning of a light plane (8:11). This drone, however, drops in pitch until it drowns out the singing, slowly transforming to become part of the Totem section, resembling and resonating with the sounds of the cello (Western music).
Granular synthesis is essentially a specialised form of micro-brassage; that is to say, the grains are enveloped sound-fragments which generally lie within the range of 20 to 60 milliseconds. These fragments may be redistributed either synchronously (in a linear fashion) or asynchronously (randomly). When we wish to produce a time stretching effect, we can employ synchronous redistribution of grains.
Granular synthesis is useful for very large time stretches, in the order of 100 to 1000 times. My first example of granular synthesis demonstrates how it creates interplay on the amplitude morphology field, from a fluctuating typology to a continuous, granular percept. It also demonstrates change on the spatiomorphology level: both from a close field to a mid-distance field (due to its reverberatory nature) and also, by its continuity, to a feeling of immensity. We will first here the original phrase, then the first half time-stretched by sixteen. Finally the word 'born' will be stretched out to one-hundred times its original length, the quasi-reverberatory aspect now acts as a 'chorus' effect so that rather than the one individual, we have the percept of a massed choir (Sound Example 9). On the utterance level, this creates an obfuscation of meaning. The best example of this that springs to mind is Susan Frykberg's Birth/Rebirth/Bearing Me where an entire ten-minute piece is derived from the sentence 'Born in Hastings', although the sentence is only gradually alluded to towards the dénouement of the piece. Two more examples of the initiatory indicative and spectromorphological interplay follow. These illustrate the change to the object/substance field possible. Two small gestures will become the basis for an immense, continuous object; the first from a metal-stringed harp, the second from a small metal ball in a metal bowl. (Sound Example 10)
Barry Truax remarks:
When granular synthesis is used to produce continuous textures, it has no resemblance to instrumental and other note-based music; instead, its sound world is more closely related to analog electroacoustic music, but with greater precision of control. In certain cases, the acoustic result resembles environmental sounds in terms of their inner complexity and statistical texture. However it is used, granular synthesis is clearly situated in a different psychoacoustic domain than that occupied by most computer music; it creates a unique sound world and suggests new approaches to the way music made with it is formed.
Certainly the unique method of control allows for some interesting sonorities; but one must again question whether the over-unifying effect of a total reliance on granular synthesis can somehow rob the music of a certain degree of musical dialectic.