The objective of this article is to provide a detailed analysis of Jonty Harrison’s piece Unsound Objects. An analysis of the whole of Unsound Objects was carried out using a combination of the author’s cognition-based SIAM framework for the analysis of electroacoustic music, a number of software programs for signal analysis, representation and annotation, and careful listening. It begins by analysing the acoustic surface to provide a detailed deconstruction of the first few sections of the work. The analysis then moves beyond the acoustic surface to look for structures, functions and motion, and the two halves of the work are compared. Mimetic archetypes, structural archetypes, and operations on sounds are identified. Functional relations and sonic activity are also summarised. Part of our discussion notes that the beauty of Harrison’s work is the fact that he is exploring and exploiting both the mimetic/semantic implications of the sound world he is creating, and simultaneously weaving musical structures and relations. This analysis of Unsound Objects has used spectral irregularity plots to assist with the identification and isolation of periods of activity within different sections of the work. Irregularity may be a useful tool in the analysis of other works.
The objective of this article is to provide a detailed analysis of Jonty Harrison’s piece Unsound Objects. The work exists in a number of forms, including a version on DAT tape that I was given by the composer many years ago. However the version used in this analysis appears on the Articles indéfinis CD (Harrison 1996). The track notes from this CD provide the background to this version:
Unsound Objects was composed at the composer’s studio and in the Electroacoustic Music Studios of The University of Birmingham (UK) and was first performed at the 1995 International Computer Music Conference (ICMC ’95) in Banff (Alberta, Canada) on September 7, 1995. Unsound Objects was commissioned by the International Computer Music Association (ICMA).
In a response to a question from John Palmer about the nearly 20 year time span of works on the Articles indéfinis CD, Harrison sets out his progression from instrumental composer to acousmatic composer over this period (Palmer 2002). Earlier tape works, such as Klang, ‘are quite abstract – they concern themselves with a (fairly traditional) musical discourse, albeit with a rather broader timbral palette than in instrumental music’. Harrison observes that works from this period are ‘classic Schaefferian musique concrète’ making use of ‘what Schaeffer calls “reduced listening” (where the sound object is defined by its sounding characteristics and not by reference to its source or cause)’ (Palmer 2002). Harrison contrasts these works with acousmatic works produced after his ‘abandonment of “paper” composition’ around 1994 where ‘there is a much more evident (and much more conscious) use of recognisable sound for its very recognisability — reference and recognition are now an active part of my thinking. I call this “expanded listening” ...’ (Palmer 2002). Harrison points out the dialog between composer and the sounds s/he has recorded:
For me, composition is a partnership between material and composer, each interrogating and challenging the other to work out the next step in the process of musical creation. In my case at any rate, the “compositional mind” is not something which imposes abstract “musical” speculation on inert sound material — sound material is not inert!
When specifically asked about Unsound Objects, Harrison says:
... well I guess the materials I chose to use ... were essentially referential and likely to be recognisable by listeners. As it would therefore be impossible to expect listeners to be able to exercise “reduced listening” in such a framework I chose instead to exploit this very recognisability in working on the piece. This is an implicit challenge to Schaefferian orthodoxy about the definition of the sound object through reduced listening;; it is dangerous and, hence, “unsound”. As for the actual processes and so on in the studio — well, I guess you could say I followed my nose (or, rather, my ears): there was no preconceived plan or structure;; the material and I evolved the specifics of the piece as we went along. It was an organic process.
The question for discussion in this current analysis is: How can we gain some insight into how Harrison ‘followed his ears’ and what relationships between the sound objects, the listener, and the composer, has Harrison attempted to establish as the piece unfolds?
Analysis of the whole of Unsound Objects was carried out using a combination of the author’s cognition-based SIAM framework for the analysis of electroacoustic music, a number of software programs for signal analysis, representation and annotation, and concentrated listening. The SIAM framework (Hirst 2008) stands for Segregation, Integration, Assimilation and Meaning. It can be briefly summarised as follows:
- Segregation: identification of sound events and the factors responsible for identification.
- Integration – Horizontal: identify sequential streams and patterns of sonic objects;; determine causal linkages,
- relationships and possible syntaxes;; consider organisation in time and the horizontal integration of pitch (where appropriate).
- Integration – Vertical: consider vertical integration and segregation as a cause of timbre and texture variance;; consider psychoacoustic dissonance and musical dissonance;; consider emergent properties relating to pitch (horizontal overlap).
- Assimilation and Meaning: consider various types of discourse, e.g. discourse on the source-cause dominant (semantic) to typological-relational dominant (syntactic) continuum;; consider global organisation in time and any hierarchical relationships;; consider expectation-interruption, arousal and meaning, semiotic relations, etc.
Using the SIAM process as a starting point, the analysis was then free to take on a life of its own, depending on the direction dictated by the work itself. This adaptive approach seemed to accord with Harrison’s approach to his compositional technique. The winding analytical path is documented in the following text.
3. Performing the analysis
3.1. Analysing the acoustic surface
The first part of the SIAM process is segmentation. The work was first examined to see whether it could be segmented into sections. It was found that the work consists of two large segments. The first ‘half’ lasts from the beginning until about 6’53”, and the second ‘half’ from 6’55” until the end of the work at 12’59” (duration 6’04”). The work was segmented further by choosing moments of silence or points of contrasting changes in texture. We could think of this stage as being akin to identifying phrases in instrumental music. This next level of segmentation yielded eleven sections in the first ‘half’ and nine sections in the second ‘half’. The sections are of varying length. The longest being about 1’30” and the shortest about 12 seconds.
The next part of segmentation involves the identification of sound objects. To facilitate sound object identification, two displays were created in the program Sonic Visualiser (Cannam et al. 2010) – one above the other. The top display consisted of a spectrogram of frequency versus time, with the audio signal superimposed, and the bottom display contained a melodic range spectrogram, used to identify pitch-related elements. I initially worked with 30 second time segments in order to identify as many sound objects as possible. Once identified through a combination of listening and viewing the spectrogram, labels were superimposed over the top spectrogram using words that were either an onomatopoeic representation of a sound, or an estimation of the source-cause of the sound, or an educated guess at the technical method of production of the sound, or a combination of some or all of these methods. The type of representation this yields is shown in figure 1. The labelling of sound objects was methodically worked through for every 30 second segment of the piece, from beginning to end. In essence, this became the first pass through the analysis of the work.
Figure 1. A spectrogram of Unsound Objects with sound objects shown as labels.
It was then decided to construct displays of 60 second time intervals of the work, in order to create a compromise between examining short-term detail and longer term structures. Screen shots of these 60 second segments were taken, assembled on to Microsoft Powerpoint screens, two at a time, and then they were printed out in hard copy. It was also possible to get a time-stamped listing of the annotation layer, which was then opened in Microsoft Excel and printed as a listing of all the annotations (see Table 1). The printed screens and time-stamped sound object listings became the data to facilitate the next part of the analysis process.
Table 1. Extract of the time-stamped sound object listings.
A graphic display was also constructed in the program EAnalysis (Couprie 2012). This program was employed because of its easy inclusion of graphic symbols, particularly those that depict a long, sustained sound. A display was constructed that contained three panes. The bottom pane was a representation of the audio signal for the whole work, the middle pane was a spectrogram showing one minute segments of the piece, and the top pane showed a slightly zoomed out version of the spectrogram of about two minute intervals. They were meant as short, medium and long term views. Graphic symbols representing sound objects were placed in layers over the top of the spectrograms (see Figure 2). The audio could be played back in synchronisation with the visual depictions.
Figure 2. A representation of the first section of Unsound Objects with sound objects shown as graphic symbols, using the program EAnalysis.
I should say at the outset that the intricate manipulation of pitch is not a feature of Unsound Objects, and so will not be discussed in any detail. However, the process of transposition of a sound object, or groups of sound objects, is a characteristic of the work, and will be highlighted where relevant. Initially I will give a detailed description and analysis of the surface of the first two sections of Unsound Objects to provide the reader with a feel of the work at this level. Then I will broaden and generalise more as the analysis proceeds.
3.1.1. Section 1 details
In analysing the surface of an electroacoustic work, segmentation and integration draws heavily on Gestalt principles, as applied to the auditory realm. Some useful principles, highlighted by Bregman (1999) are:
- Grouping by proximity in frequency and/or time
- Spectral integration relationships, such as harmonicity
- Common fate of frequency partials
- Separation via the “old plus new” heuristic where new frequency components are introduced over continuing components, thereby signalling a new sound object has been added
Bregman also claims there is a certain hysteresis in the recognition of sound objects in that the recognition persists for a number of seconds, unless either silence or a sudden change resets this recognition.
The first part of analysis not only consists of recognition of individual sound objects through the grouping or integration of frequency partials, but it will also involve the grouping of sound objects where that process seems relevant or important to the discussion. Ultimately, we will also look at textural features, such as irregularities, in addition to isolating the regularities we may find. A spatial analysis will not be attempted.
The beginning of an electrocacoustic work is always an important phase since the grammar of electroacoustic music is not as firmly established as that of tonal music, for example. Thus in the beginning of a work like Unsound Objects, the composer creates the embryo of a sound world, so that some sort of relationship between the sound objects, the composer, and the listener can begin to be established and the embryo develops into a more complex being as the work unfolds. Unsound Objects begins with two seconds of what I will describe a ‘sonic motive’, which is a collection of sound objects grouped through proximity in time and may function a bit like a melodic motive or a rhythmic motive in tonal music, and it may be reused later. The initial ‘sonic motive’ in Unsound Objects consists of some water sounds combined with a ‘plop’ sound, which may have been produced using some sort of plunger dipped into water.
So in the very first two seconds of the piece we can deduce that water sounds may be a feature, there is a distinctive upward glissando of the ‘plop’ sound, and we have several sounds grouped together to produce a single pattern or ‘sonic motive’ distinguished by the sounds’ mimetic timbrel similarities and by the relationship between the sounds in time (perhaps a rhythmic figure).
Bravely, Harrison follows the initial sonic motive with over three seconds of silence. The initial ‘sonic motive’ is then repeated, but transposed upwardly and quickly followed by several other sound objects I’ve called ‘boing’ and ‘chi’. The ‘chi’ element is a very short broadband noise sound, which will become a feature to be varied through different means of signal processing. Now, within the first eight seconds, we have the presentation of what I will describe as ‘sonic archetypes’ – water sounds and noise sounds.
Coinciding with the ‘chi’ noise sound at 7.98” is the beginning of what I describe as a ‘continuant’ which is a sustained sound consisting, in this case, of some sort of filtered noise, which is subjected to a type of modulation (perhaps frequency modulation) that gets lower in frequency as it progresses, and lasts until about 30.90”. Here we observe another type of ‘sonic archetype’ which I’ll call the attack-continuant archetype that, in Harrison’s case, consists of an ‘attack’ that is actually a collection of sound objects expressed in the form of what I’ve described as a ‘sonic motive’. It’s as though the sonic motive triggers or provokes the sustained continuant. This kind of archetypal pattern is a characteristic used throughout the work.
The first sustained continuant is terminated by a one second swell of broadband noise – perhaps an elongated ‘chi’ sound. It also functions as the provocation of a new sustained ‘continuant’ at 30.82” that is the expression of a much more filtered sound, creating fewer partials and therefore greater purity of sound which is lower in the frequency spectrum. Lasting over 16 seconds, it is punctuated by a solitary ‘drip’ sound at 42.45” and, at its conclusion, the first section is terminated by two events. The first event is like a time-stretched and more quiet version of part of the sound first heard at time 6.91” and the second event terminating this section is a ‘compound sound’ made up of a time stretched rendition of the ‘chi’ sound first heard at 7.89”, combined with the ‘plop’ sound transposed higher. These short sounds function as a full stop or cadence to this first phrase (Section 1).
3.1.2. Section 2 details
After a 1.5 second period of silence, Section 2 begins at 50.97”. Section 2 is slightly longer than its predecessor, with a duration of about 1’08.5”. An examination of the annotated spectrogram reveals firstly that some of the techniques and archetypes from Section 1 are employed in Section 2, and secondly it becomes more active, with more sound objects, over the course of its duration.
Against a background of spectrally filtered, higher, noise continuant, Section 2 begins with a short ‘sonic motive’ of several upward glissando ‘plop’ sounds. At around 58.13”, this motive of ‘plops’ is embellished through transposition and the lengthening of the motive with additional ‘plops’. At 1’11.30”, there is a further multiplication of the ‘plop’ motive, leading to new sound objects at 1’18.15”, which I’ve called a ‘jigger trill’ (sounding like some form of rattle or percussive trill) and a ‘doink’ or plucked-type of sound. Repeated jigger trills lead to a short ‘klok’ (wood block?) sound that announces a more fragmented filtered noise continuant, which is itself a seemingly more active background. Another ‘klok’ sound at 1’25.64” announces a new continuant consisting of a sequence of low water glissandos with active plop motives. Superimposed on this background is a shorter continuant of pitched, sparse frequency components.
Such an increase in sonic activity functions to heighten excitement and create an anticipation that the sound world is building towards some kind of significant event. Beginning at 1’39.10”, the short sounds ‘tap tap’, ‘bonk bonk’, and ‘doink’ introduce yet another ‘sonic motive’ that is then varied and followed by an upward transposition of the ‘plop’ motive. This leads to a much more thinned out background continuant of sparse frequency components built on a frequency of 584 Hertz. As this continuant fades, the ‘plop’ motive is transposed even higher and the final full stop for this phrase is a short drip sound.
We can observe that the trend of Section 2 is that it becomes more active, as time proceeds, with an increasing number of sound objects combined into groups as ‘sonic motives’. These are set against a backdrop of longer ‘continuant’ textures, which are varied much more in Section 2 than in the previous section. The ‘sonic motives’ are varied more than in Section 1 too. With this increasing activity, anticipation builds for a while, but then the section quickly fades, or ‘thins out’, and its conclusion is marked by the full stop in the form of a water drip.
3.2. Moving beyond the acoustic surface – structures, functions and motion
Up until now, I have described and discussed very fine details of the work for the first two sections. Now we must move beyond the acoustic surface to examine the relationship between sound objects in the work, and to zoom out in order to compare different segments of the piece. Through ‘time-span reduction’, we can move from viewing 30 second to 60 second to two minute segments, in order to observe the changing sonic patterns over the course of the work. Ultimately, extending this process this would lead to an analysis of the form of the whole work.
Zooming out one’s attention on Unsound Objects, one does get the sense that there is some sort of foreground, middle- ground, background perspective operating within the work. Here I am not talking about the depth of field often associated with spatialisation aspects of a work, but rather the relationships between musical elements within the work. Others often use the term ‘layering’, but I find that a little restricting. Let me use a different metaphor to describe the relationships emerging from the work. The long continuant textures certainly form a back-drop for sections of the work that we could liken to a scene within a film or play. The foreground sonic motives then take on the role of actors, establishing relationships with one another, through proximity in time, frequency, and/or similarity in timbre. Their ‘dialog’ can be varied, at different time points in the work, against the backdrop of changing or evolving scenes (continuants). The middle-ground is provided by secondary sonic motives, which perform a contrapuntal counter-pattern (rather than counter-melody) to the foreground material.
Unsound Objects can be divided into two approximate halves. In Figure 3, the first half of the work consists of sections 1 to 11 (up to 6’54”), and the second half sections 12 to 20 (6’55” until the end). We can see from this plot that the first half is characterised by sections that are much more similar in length and vary from about 20” to around 1’08” duration. The second half (beginning section 12) consists of much longer sections, especially section 12 itself, interspersed with quite short transition periods. The variation is from 12” (section 13) up to 1’33” (section 12).
Figure 3. Section length (mm:ss) versus section number for Unsound Objects.
The relationship between sound objects in Unsound Objects is beginning to emerge and we have some idea of the overall sections and their duration, but an important question remains: What propels the work along from moment to moment, section to section, scene to scene? To help answer this question, I have already observed that an increase in sonic activity seems to elicit expectation in the listener that an important event is about to occur. But how can we measure and, even better, display activity within a work? Well the Sonic Visualiser program provides access to a suite of plugins of signal analysis, and the type of analysis that seems to correlate best with sound object activity is a plot of ‘irregularity’ versus time. It could be argued that we could simply use a plot of overall amplitude, or one of its variants, but let me contend that simply using amplitude tends to average out ‘activity’, whereas spectral irregularity provides a much more fine-grained representation of activity from instant to instant through time.
There are several different methods for calculating the irregularity present within a spectrum (Jensen 1999;; Krimphoff et al. 1994), but essentially they both give a measure of the degree of variation of the successive peaks of the spectrum. Jensen, for example, calculates the sum of the square of the difference in amplitude between adjoining partials. What I am postulating here is that where there is a large variation across the spectrum, partial to partial, then this can provide us with a depiction of a high degree of activity. Figure 3 depicts an irregularity plot for the first four sections of Unsound Objects.
Figure 4. An irregularity plot for the first four sections of Unsound Objects.
If we take a look at the irregularity for Section 1, we can observe that the irregularity is up and down quite a bit until we see a very large peak at about 30”, and then the irregularity dies away. The peak corresponds to the swell of broadband noise that we described in 4.1.1 above. Let’s look at the next section and see whether our hypothesis that irregularity gives some measure of activity does hold water. In Section 2, the irregularity plot shows an increase in irregularity from about 1’25” until around 1’32”, and above (4.1.2) we noted that at about 1’25”: “an increase in sonic activity functions to heighten excitement and create an anticipation that the sound world is building towards some kind of significant event”. However, our expectations are not fulfilled, and the activity subsides to finish with a single drip sound. Although it is hard to read in the graph shown in Figure 4 within this text, an irregularity plot can actually depict individual events (such as a drip sound) to quite a high resolution.
Let us now shift our focus from details within a section to a comparison between sections. In doing so, it is useful to tabulate the activities and the sound types for each section (a selection of this data is tabulated in the Appendix below), and view the table in conjunction with the graphs of irregularity.
3.2.1. Section to section comparison
Figure 4 shows us that there are roughly two types of activity when we compare sections 1 to 4. The first three section shows a rise and fall in activity, while section 4 displays lots of peaks in activity. Our table of activity lists sound types that are dominated by water sounds, noise sounds, and filtered noise continuants. Running water is introduced in section 3 and section 4 is very active with lots of the sounds and gestures we’ve heard previously.
Figure 5. An irregularity plot for sections 5 to 9 of Unsound Objects.
Figure 5 shows the activity for sections 5 to 9. Sections 5 to 7 are very active, with lots of peaks. Section 8 has a less active beginning and it slowly builds over time, and section 9 is similar in that it begins with less activity and slowly builds to a granulated crescendo, subsiding to individual events. Section 5 (begins 3’3.03”), which is about half way through the first half, introduces lots of new mimetic elements: the first ‘dry’ sounds (crunching paper or leaves), door gestures, footsteps, crunching and crackling. These dry sounds provide a dramatic contrast to the water sounds that opened the work. Section 6 (starts at 3’32.8”) mixes up a return of the frequency modulated filtered noise with a long sequence of dry mixed sounds, footsteps, and a collection of many sounds heard so far. Section 7 (beginning 4’01.1”) has klunks, water plops, dry sounds, chiffs, which then have some modulation treatment, with more plop/gliss sounds. The short section 8 (starting at 4’24.1”) has a continuant of filtered noise, differentiated across the channels, and when a long granular treatment enters, it provides an introduction for the thunder sound at the start of section 9 (at 4’44.2”). Then rain, lightning strikes and granulated noise leads to footsteps in gravel or leaves, a door opens. Here we might speculate that the footstep sounds not only have a sonic archetype ‘dry’ sound, but also function to move the work onwards. The door sounds can have a functional aspect too: they signify a change of some sort – usually a beginning of something different, or an ending.
Figure 6. An irregularity plot for sections 8 to 11 of Unsound Objects.
Figure 6 shows the activity for sections 8 to 11, but we’ll concentrate on section 10 and 11. Section 10 (at 5’35”) is quite short (about 25”) and has much less activity, with a quiet continuant of modulated noise, filtered footsteps, dry sounds, a squeaky door that is opened the closed. Section 11 (at 6’00.4”) begins with a bird call, a door opens and this seems to ‘trigger’ a long, quiet continuant of filtered noise, punctuated by individual sounds events (sequence of clicks, fast iterative sounds). The continuant thins out, the activity subsides and it fades to a door closing sound. The decrease in activity combined with a softer, longer soundscape, and a seemingly conclusive ‘closing the door’ creates an expectation of some sort of conclusion: “Closing the door on the first half of the piece.”
Figure 7. An irregularity plot for sections 12 to 15 of Unsound Objects.
Figure 7 represents the activity for sections 12 to 15 of Unsound Objects. We have seen from Figure 3 that section 12 (starting at 6’55.1”) is the longest section of the work (duration 1’33.3”). It represents the start of the second half of the piece, and it announces itself in a very dramatic manner: A loud ‘smack-like’ sound, very percussive. A sequence of these smack sounds is followed by smeared noise sounds. In terms of activity, the smeared noise sounds swell, subside, swell, subside, fade, and we are left with the sharp, impulsive crack sounds. In terms of events and mimetic references, there is a lot going on in section 12. In addition to the sounds mentioned so far, we hear a hard bouncing ball iterative sound, dry crackling continuants, high-pass filtered smeared noise, hints of previous frequency modulated filtered noise, crackling transposed higher, footsteps, varied crackling, an iterated bouncing marble connected to a deep plunge that leads to section 13. Overall the impression is of a dry soundscape with broad spectral characteristics. There are short, sharp, and sometimes repeating or iterative sounds. Functional connections, where one sound seems to lead on to another sound, are used to create connections to future events – expectation creation. Section 12 establishes a strong contrast with the first section: it is longer, it is dramatic in its use of sounds, it makes extensive use of the dry sound world that dramatically differentiates itself from the watery first half, but it does contain some references to certain sounds used in the first half. We are left wondering: How will this progress?
Section 13 (starting at 8’28.5”) is a short transition with short impulse events and lots of attacks. It contains filtered noise sliced up in time, bouncing ball sliced up in time too, a high pass filtered crack sound and a chopped up bouncing ball. The sound shredding technique is something that will be used again within the work.
Section 14 (starting at 8’40.5”) begins with a crack and continues with lots of iterative activity which is softer and more subtle than the previous section. It consists of two types of iterative sounds interleaved with each other over time. The first type is a high pass filtered bouncing ball (a marble?), and the second iterative type has varied repeated attacks and sounds like a stick being bounced on a string. As they are interleaved, variation is provided through transposition, slowing down, and other types of variation.
Section 15 (starting at 9’30.4”) is a soft, short transition passage, with minimal activity. After some soft iterative sounds, we hear very soft walking and then the unlocking of a door.
Figure 8. An irregularity plot for sections 16 to 20 of Unsound Objects.
Figure 8 represents the activity for sections 16 to 20 of Unsound Objects. With the unlocking of a door, the thunderstorm makes another entrance with cracks of thunder followed by footstep activity. The thunder seems to ‘trigger’ an increase in activity of sliced noise events. The thunderstorm and rain continue with crunching and crackling mixed in. A lightning crack announces a return of the frequency modulated filtered noise continuant, but has it been sliced up too? More crunching footsteps and the frequency modulated noise fades while a flanging thunder-type noise is used as a continuant and segues into section 17, functioning as another short, quiet transition period. It begins with minimal activity and is punctuated by short, crisp events like the sound of footsteps on gravel. Broadband noise moves from channel to channel.
Section 18 (from 11’33.0”) begins with a lightning strike and flanging of the thunder as its continuant moves between the channels. It is punctuated by a sequence of three or four clicks. There is additional broadband noise in motion between the channels and lapping water heightens the return of water sounds. Section 19 continues the waves washing on the shoreline. The wave sounds increase in loudness and they are coupled with voices and the shouts associated with play on a distant beach somewhere. The wind and the waves are sliced up, there are crunching steps and more voices.
Section 20 (from 12’40.8”) begins with a breaking wave that ushers in a final reprise of the multi-band filtered noise. There is a klomp, the voices fade, and the filtered noise fades to the end of the piece.
3.2.2. Comparing the two halves
Zooming out even further, how can we characterise the two halves of Unsound Objects?
The first half consists of sections 1 to 11, and lasts until about 6’54”. The sections within the first half are much more similar in length compared with the second half. Activity within the sections of the first half seems to follow two or three different models. There is the ‘rise and fall’ model of activity (e.g. sections 3, 8, & 9), there are the sections with lots of activity (sections 4 to 7), then there are sections with sparse activity that are longer and quieter (sections 10 & 11).
Mimetically, the first half begins with a water theme and noise variants, and then ‘dry’ sounds are first introduced in section 5, about half way through the first half. Water sounds return in section 7 and then we have a mixture of water and dry sounds with thunder and footsteps added in section 9. So its mimetic scheme is WATER-DRY-WATER-MIXTURE.
Structurally we’ve observed that sonic motives are set against long continuant sounds of filtered noise. Sonic events seem to trigger other events, especially the continuant events. Footsteps serve two purposes: mimetic and functional. They are both dry sounds and imply a forward motion in time. The function of door sounds is to initiate or terminate sections or processes.
The second half consists of sections 12 to 20, according to my segmentation of the work. It begins at 6’55.1” and lasts for about 6’04”. It contrasts greatly with the first half in that its main sections are much longer than those from the first half, interspersed with short transition sections. Each of the long sections begins with a very dramatic gesture (e.g. section 12, a loud crack;; section 14, another loud crack;; section 16, a thunderstorm).
Mimetically, the second half begins with dry sounds. Whereas the first half used a combination of sounds collected into sonic motives and set against a continuant, the second half makes extensive use of iterated sounds such as bouncing marbles, sticks bouncing on strings, noise sources radically ‘shredded’ or chopped up using rapid amplitude modulation techniques. Suddenly, the opening of a squeaky door signals something new in section 16 (9’48.1”). A thunderstorm breaks and there is a return to the water theme, then crunching footsteps, and noise in motion. Section 18 begins with the lightning/thunder sound and it is flanged along with other familiar sounds. Finally there are wave sounds, water lapping and a coda of the multiband filtered noise sound. So the mimetic scheme of the second half is a kind of reverse of the first half: DRY-WATER- DRY-MIXTURE-WATER.
Predominantly the sound world of the second half consists of extended periods of interspersed iterative sounds, triggered by the dramatic gestures of thunderstorms and waves breaking. Footsteps are signifiers of progress and door sounds are signifiers for new sound worlds. In contrast to the first half, the second half diminishes the distance or depth of field between the foreground, middle-ground, and background sounds. The structure is much more tightly integrated and our actor within a scene metaphor begins to break down. Instead, dramatic sound events propel a kaleidoscope of sounds mixed together in a tightly woven fabric.
4. Sonic Archetypes
To begin to round off our discussion, we can summarise what I shall call ‘sonic archetypes’ present in Unsound Objects. Sonic archetypes can be mimetic, or they can be functional or structural models.
- Water sounds: plop, plunger with upward glissando, drip sounds, running water, thunderstorms, lapping water, breaking waves
- Broadband noises with their variants
- Modulated and filtered continuant sounds
- Dry sounds: e.g. crunching footsteps, rustling paper
- Iterative sounds: bouncing marble on hard surface, sticks bouncing on strings, chopped up noise sounds
- Door sounds
- Sounds of utterance: birds calls, people’s voices
Structural characteristics arising from the various combinations of sonic objects:
- Sonic motives – a group of short sounds, grouped together by virtue of their proximity in time and similarity in timbre, almost forming a rhythmic motive.
- Attack-continuant model of a sonic motive followed by a sustained sound lasting a long duration.
- Sonic Collection – a term used to describe a grouping of sounds that aren’t related in timbre, but function as a textural unit. Some literature refers to this aspect as simply texture.
- Sonic Sequence or Stream – a horizontal pattern of sounds related in some manner through proximity in frequency and/or time. Some literature refers to this aspect as simply gesture, however the term gesture is imprecise and can be interpreted in many ways. The term ‘gesture’ is probably best reserved for describing the changing spectromorphology within a sound object and perhaps the connection between individual sound objects.
- Compound sound – a sound produced by combining several different timbres at the same time point, usually short impulsive-type sounds.
Operations on sounds:
- Time stretching
- Modulation – frequency and amplitude
- Shredding or rapidly chopping up sounds
- Proximity in time of short sounds, and/or timbral similarity, produces grouping into ‘sonic motives’
- Foreground-background: Sustained, continuous textures, which I have labelled ‘continuants’, often function as the background to a foreground of ‘sonic motives’. The continuants mark out periods of time we could describe as ‘scenes’ during which, the sound object ‘actors’ conduct their sonic motive dialogs.
- Implication-realisation effects: eg. Increasing sonic activity can lead to an implication or anticipation that the piece is building to something significant
- A short impulse or percussive sound is often used to mark the end of a section, and it serves as a punctuation mark to end a sonic ‘phrase’ or ‘sentence’
- Trigger sounds – loud impulse sounds that seem to trigger the following sequence of sound objects
- Tight integration – foreground and background distance is reduced by tight integration of the mixture of similar types of sound, e.g. iterative sound sections in the second half in particular.
- Spectral irregularity has been used as a marker for sonic activity.
In a keynote address to the SoundAsArt Conference in Aberdeen in 2006, Harrison begins by defining various terms used throughout the history of electroacoustic music, but in doing so, he clarifies some of his own position within this art form. In relation to musique concrète, Harrison (2007) notes:
Musique concrète, then, is so called because it starts from ‘concrete’ (i.e. observable) qualities in sound material itself, from which structural implications are extrapolated and links to other sound materials made. What finally emerges, as Dhomont has pointed out, is an ‘abstract’ musical discourse ... which has been ‘abstracted’ from the concrete starting points.
What we have observed in the above analysis is the deconstruction of Harrison’s concrete to abstract processes in the reverse. In his Aberdeen keynote, Harrison (2007) emphasises the interplay between composer and sound using aural feedback as a compositional method:
... what is ‘concrete’ about musique concrète is the method of working: it is ‘hands-on’, with immediate aural feedback. Its composition is a partnership between composer and material, with each interrogating the other, posing problems and offering possible solutions.
We have found the sonic archetypes identified within the work and noted them above. What is interesting is that these archetypes (or models) can be mimetic/semantic, structural, operational, or functional. In other words, the beauty of Harrison’s work is the fact that he is exploring and exploiting both the mimetic/semantic implications of the sound world he is creating, and simultaneously weaving musical structures and relations.
In his Aberdeen address, Harrison (2007) also emphasises the temporal aspect of composition in this medium:
Time is what makes music out of sound. ... ‘Music’ happens when sounds establish ... relationships in time between each other which make some kind of sense – i.e. they ‘communicate’ ... ‘meaning’ ...
From the text in the previous section we can observe that time is referenced in many ways in order to structure, create connections, and to imply progression through time. Terms like: proximity in time creating sonic motives;; continuants sounds being used to provide a durational backdrop for foreground activity;; a sonic collection of simultaneous sounds, or a sonic stream of horizontally connected sounds;; sounds that act as punctuation points in time, e.g. the drip sound or closing doors;; trigger sounds (like breaking waves) that seem to promulgate other sounds;; mimetic ways of depicting the passage of time, like a sequence of footsteps;; or syntactic ways of depicting passage of time, such as the increase in activity that starts to stimulate our implication-realisation juices.
As Rudy has pointed out: ‘Incongruous juxtaposition of source is evident in Jonty Harrison’s Unsound Objects’ Rudy (2007: 12). Rudy reminds us that: ‘the composer has carefully led the listener from source attention (cinematic listening for a sound source or Schaeffer’s Mode I) to listening to how the sounds interact (reduced listening).’ (Rudy 2007:12).
It is hoped that the above analytical work further illustrates what both Rudy and Harrison contend in their respective quotations.
Finally, we have used spectral irregularity plots to assist with the identification and isolation of periods of activity within different sections of the work. It seems as though it may be a useful measure for the analysis of other works – a hypothesis remaining to be tested.
Table of activity and sound object types for selected sections of Unsound Objects.
Activity – amount and type
Selected Sound Object Types
Activity is up and down quite a bit until we see a very large peak at about 30”, and then the activity dies away
Water, plops, boing, chi, continuant - filtered noise –modulated, broad noise, drip smear, high plop, chik
Sonic activity low, then increases about mid point, then dies away
Spectrally filtered noise continuant (higher), plops, jigger trill, doink, klok, Sequence of low water glissandos - active motive, Low plop sequence continues, sparse freq components, drip
Slow rise and fall in activity
Drip, running water continuant, resonant acoustic ambience continuant, blok, chi, running water continues & fades
Lots of peaks in activity plot
Continuant of bandpass-filtered noise, motion between channels, chiff, transposed lower, jiggle trill, more active version of previous gestures
Breaking wave ushers in a final reprise of the multiband, filtered noise and fades
Breaking wave swells, Multiband filtered noise enters as very pure continuant, Klomp, Voices fade, Multiband filtered noise fades to end
CANNAM, C. LANDONE, C. and SANDLER, M. (2010) Sonic Visualiser: An Open Source Application for Viewing, Analysing, and Annotating Music Audio Files. In: Proceedings of the ACM Multimedia 2010 International Conference, October 2010. Italy, pp. 1467-1468.
HARRISON, J. (2007) Dilemmas, Dichotomies and Definitions: acousmatic music and its precarious situation in the arts. In: SOUNDASART (2006) Blurring of the Boundaries. November 2006. Aberdeen: University of Aberdeen.
HIRST, D. (2008) A Cognitive Framework for the Analysis of Acousmatic Music: Analysing Wind Chimes by Denis Smalley. VDM Verlag Dr. Muller Aktiengesellschaft & Co. KG. Saarbrücken.
JENSEN, K. (1999) Timbre Models of Musical Sounds (PhD), University of Copenhagen.
KRIMPHOFF, J. MCADAMS, S. and WINSBERG, S. (1994) Caracterisation du timbre des sons complexes. II Analyses acoustiques et quantification psychophysique. Journal de Physique IV, 4(5), pp. 625-628.
PALMER, J. (2002) High Spirits with Jonty Harrison. 21st Century Music, 9(1), pp. 1-4.
RUDY, P. (2007) Timbral praxis: when a tree falls in the forest is it music? Organised Sound, 12(1), pp. 5-13.