Page  00000001 AN INTERACTIVE AURAL APPROACH TO THE ANALYSIS OF COMPUTER MUSIC Michael Clarke University of Huddersfield, Department of Music and Drama, Queensgate, Huddersfield England, HD1 3DH j.m.clarke@hud.ac.uk ABSTRACT Computer music presents particular challenges for the analyst. This paper discusses the approach taken by the author to tackle these issues in analysing Jonathan Harvey's work Mortuos Plango, Vivos Voco [2]. The focus of the paper is not the analysis itself but the problems posed for the analyst by such a work and the methods used to try and resolve them. It describes a number of different techniques that were used, both in the making of the analysis and in its presentation. In particular it focuses on the interactive aural approach developed for this analysis using specially designed software and discusses its potential for wider application. 1. INTRODUCTION Jonathan Harvey's Mortuos Plango, Vivos Voco was composed in 1980 at IRCAM. It is an 8-channel piece derived entirely from computer transformations of two sound sources: recordings of the great tenor bell at Winchester Cathedral, and of the composer's son, Dominic, at that time a chorister at the cathedral. The timbre of the bell forms the basis of the whole work. Different partials of the bell form the focal pitches for each of the 8 sections of the work. The bell is transposed (by changing the speed of sound file playback) onto the frequencies of its own partials. Mediation between the two contrasting source sounds takes various forms, for example: modification of the characteristic envelopes of each source; synthesis of 'bell' timbres using vocal sounds placed on frequencies of the bell partials; transformation of one spectrum into the other by means of glissandi. Glissandi, in conjunction with pivot frequencies, also play an important structural role in the work, creating 'spectral modulations' somewhat analogous to modulation in tonal music. 2. THE CHALLENGE From the above it will be clear that Mortuos Plango is far more pitch-oriented than some electroacoustic music. This might give the impression that traditional approaches to analysis might be appropriate and certainly there is some role for these. However, as with many 'tape' pieces, the normal starting point for analysis of Western classical music, the score, does not exist. The 'trace' here is not a written score but an audio recording. Furthermore, whereas a score normally provides a set of discrete data (pitches, durations etc.) which can form the basis for analysis, with Mortuos Plango it is not always possible to identify discrete events: a bell timbre might transform continuously into a voice; a component of one timbre (e.g. a bell partial) might evolve into an event in its own right. In the absence of a score, analysts of computer music often resort to creating various forms of visual representations, but these are not without problems. Clearly the creation of a score, or even the identification of discrete events, is in itself an act of interpretation, of analysis, and not a neutral act. Sonograms, for example, have frequently been used in the analysis of computer music and can provide useful insights. However, they work best in relatively simple textures, such as the glissandi of sine waves in section 3 of Mortuos Plango. However, the depiction of more complex polyphonic textures is often confused. Partials of different sounds become entwined and the perceptual grouping that happens aurally is not shown. In such situations sonograms result in the opposite of what analysis is trying to achieve, disaggregating the material in a way that hides relationships. Symbolic pictures are also used sometimes in place of a score. Although useful at a basic level of description, two-dimensional black and white drawings are rarely able to capture the subtle detail of the sound and can frequently trivialise the musical structure. Especially with a work as oriented towards pitch as Mortuos Plango, transcription into musical notation is another possibility. Indeed this approach was found useful for various aspect of the work. However, traditional notation is designed to place events on a two-dimensional lattice-work (striated pitch and rhythm). Other dimensions (e.g. timbre) and continuous, smooth data are not so easily represented. Two further factors are important. Firstly, whatever system of visual representation is used, the reader is unlikely to be able to read the music from the page in the same way as would happen with the use of notation in traditional analysis. Secondly, the reader is less likely to be familiar with the techniques used in the work. In analysing tonal music an analyst can assume a good level of basic theoretical knowledge and build on this. The techniques of computer music are less widely known and different composers use a wide variety of different techniques so that it is not possible to assume the appropriate knowledge even within the computer music community.

Page  00000002 3. AN INTERACTIVE AURAL APPROACH 3.1. CD playback Given that the work exists as sound and the difficulties already described with visual representation, it seems appropriate to consider the value of an aural approach to presenting the analysis. Sound is used in a variety of ways in the analysis of Mortuos Plango. Firstly, it is used to play precise extracts from the work itself. Although primarily an 8 -channel work Mortuos Plango is widely available in a number of stereo CD recordings. The software used for the analysis (written using MaxMSP) can play back extracts from such commercial recordings (the CD is inserted into the computer's CD drive) without the need to copy the audio. No copyright is infringed and the computer simply acts as a sophisticated CD player. The reader is expected to have access to the CD just as one might expect someone reading a traditional analysis to have access to a score. The written text of the analysis makes frequent reference to numbered 'Sound Locations' in the work and the software permits these very precise extracts to be played at the click of a mouse. 3.2. An aural paradigmatic analysis Ruwet, Nattiez [3] and others have developed distributional analysis in relation to notated music. The music is first segmented and these segments are then classified according to similarity. Typically the results of such an analysis are presented in musical notation. Different classes of material are represented as different columns and the score is divided between columns according to its classification. Reading across the columns left to right progressively down the page reconstructs the original score. Reading each column separately from top to bottom reveals the development of a particular class of material through the piece. The aural paradigmatic chart presented as part of the analysis of Mortuos Plango works in a similar way. Labelled buttons are used to represent different segments of material and these are organised in columns according to their classification. Clicking the buttons plays the relevant extract. In the context of this work a large-scale paradigmatic analysis seemed most appropriate. In analysing other electroacoustic works it might well be appropriate to use smaller segments and this would be entirely feasible with the software. Many of the well-known examples of paradigmatic analysis are of monodic works. Segmentation becomes more difficult to present in this way when different voices with different types of material overlap. A similar issue arises in this analysis and as a result some of the segments overlap and it is not possible with CD playback to isolate a particular strand of the texture and play this back independently of other material sounding simultaneously. Nonetheless, the paradigmatic chart provides a useful means for readers to orient themselves to the shape and structure of the work, and also a way of tracing the development of different types of material through the work as it progresses. 3.3. Reductive aural sketches A common feature of many types of analysis of notated music is the use of reductive sketches to reveal the deeper structure underlying the foreground detail of the music. Such sketches can be studied by the reader and compared to the original score. In the analysis of Mortuos Plango such notational reductions are used where appropriate and, given the importance of pitch in the structure of this work, they play an important role. Although microtonal inflections need to be indicated there are places in the work where notating such examples is feasible. However, given the nature of the work, an important addition is the adaptation of the same principle to the aural domain. An example is the presentation of the structure of the glissandi which underpin the third section. These are presented as musical notation in outline form, but the microtonal nature of the passage makes it difficult represent every aspect clearly. So, in addition, synthesised tones are used to present the glissandi aurally. The reader can play the component elements separately, or play the whole passage, and the synthesised reductive sketch can be compared immediately with the relevant extract from the work itself. In fact it is even possible to play the sketch at the same time as the original. In this way the reader can gain insight into the underlying structure and also hear how this is elaborated in terms of the surface sound of the work itself. 3.4. Learning the style In analysing a piece from the mainstream of Western classical music an analyst may well assume that readers are likely to be familiar with many of the main technical features of the compositional style, for example, tonality, cadential formulae, sequential progressions, sonata form, fugue or serialism. They will also therefore have an understanding of the potential of these techniques and the range of options facing the composer within a given framework. They will have knowledge of a certain range of 'norms' and of the significance of deviations from these. In their inner ear they will be able to imagine alternatives to the solution chosen by the composer and therefore have an understanding of the significance of that choice. However, the analyst of computer music cannot safely make such assumptions. Not only are the techniques of computer music less widely known, but also the range of different techniques and approaches used by different composers is such that the reader, even a specialist in the field, may not be familiar with all the techniques used. In some cases a technique is invented for a particular composition, it is part of the formation of the individual work. All this means that part of the analyst's work may well need to be clarification of the basic components of the compositional technique. Such techniques can be described in words and diagrams, but to thoroughly understand them and grasp their significance the reader will often need aural experience and the opportunity to try out the techniques interactively. In this way the reader can learn more of the potential of the techniques and can place the compositional choices made by the

Page  00000003 composer in a wider context of possibilities using these techniques. In the analysis of Mortuos Plango the accompanying software offers just such opportunities. The software provides a series of interactive exercises illustrating the techniques used by Harvey in the work. In some cases synthesis is used to simulate the processes used by the composer. In other cases source recordings, kindly provided by the composer, are manipulated to illustrate the transformational methods he used. The exercises range from the relatively simple example of filtering a recording of the bell at Winchester to pick out the individual harmonics (Fig. 1) to much more complex manipulation of the glissando technique used by Harvey. In the first case the reader is learning about the structure of the source material (the bell) used by Harvey and about the way this can be turned into a compositional structure, prolonged over time. In the second case, it demonstrates how Harvey uses glissandi not simply as a gesture or colouristic effect but as a means of linking passages through pivot frequencies, and of modulating from one pitch centre to another. An exercise provides readers with the opportunity to create their own glissandi, using the same principles as used by Harvey, and to compare these with those in the work. Other exercises, for example, demonstrate the effect of some of the techniques used to manipulate the bell recording, for example, changing the envelope, reversing playback, inverting the dynamics and changing the decay patterns of the partials. All this is designed to develop in the reader a level of familiarity with the technical world of the composition and to provide an understanding of the opportunities and choices facing the composer in this context. All this is important in understanding the way that the work is shaped, and is best achieved not through text or diagrams but through this sort of interactive play with the sonic materials and transformational tools. 3.5 Summary The interactive aural approach therefore fulfils a number of different roles: It provides a means of orientation within the work; it provides an aural paradigmatic chart and reductive aural sketches. In addition it provides the reader with a mean of gaining familiarity with the means used by the composer to structure the sound. It helps to develop a deeper understanding of the compositional process and of the way in which the work is structured. 4. A PRAGMATIC APPROACH Although the interactive aural approach is important in the analysis of Mortuos Plan go it is not the only approach used. In addition to the computer software more traditional means are used, including text and visual representations. Similarly, no one analytical method is employed in this analysis. A pragmatic approach is taken, blending a range of methods as appropriate to the issues under consideration. Musical charts are used to illustrate the large-scale pitch structure of the work and these in some ways resemble Schenkerian graphs, representing different levels within the hierarchical structure by using different note-heads. As already described, a distributional (or paradigmatic) approach is used to demonstrate the way in which different materials evolve in the course of the work. Recent discussions of musical analysis have been critical of analytical methods which focus too exclusively on the musical 'trace' without proper regard for the context in which this has been created or received. In this analysis detailed attention has been given in particular to the poietic element. The composer's aesthetic and technical ideas are discussed and related to the music. The composer also generously made his hand-written sketches available and these greatly aided the process of analysis. 5. WIDER APPLICATION The software developed for the analysis of Mortuos Plango is easily extensible for someone with knowledge of MaxMSP programming (which is an increasingly large percentage of the computer music community). It is constructed within the Sybil [1] environment, developed at the University of Huddersfield by the author together with Ashley Watkins, Mathew Adkins and Mark Bokowiec. As a result it is possible for the interactive aural approach described here to be extended to the analysis of other works. The fact that the audio is played from a standard commercial CD means that there is no infringement of copyright. Mortuos Plango is perhaps not typical of much computer music in the great significance attached in the work to complex pitch structures. The interactive aural approach may be even more important in other works where pitch is less significant and, consequently, description of the work using traditional notational methods more difficult. In acousmatic works where the focus is on the spectro-morphological aspects of the sound such an aural approach may be particularly useful. Beyond computer music there may also be a significant use for an interactive aural approach in ethnomusicology, in the study of aural traditions. In such work issues of notation arise and the ability to reference extracts directly in terms of sound rather than via an artificially imposed notational system could well have advantages. In traditional Western classical music the potential for an interactive aural approach may not be so clear. Notation is the means by which the music is transmitted. Furthermore, analyses are likely to be read by those trained in reading notation. Therefore the presentation of musical examples or structural charts using musical notation is effectively in aural domain in any case. Nonetheless, recent trends in analysis have directed attention more towards issues of interpretation and performance. In theory different interpretations could easily be compared using the software. However, either a multi-CD drive would be needed for the computer or else all the interpretations would need to be available on the same CD for ease of use. This might

Page  00000004 involve the creation of a CD especially for the purpose and copyright would then become an issue. The equivalent of the interactive exercises may again not be so relevant within this tradition, both because readers would already be likely to be familiar with the style and because musical notation could be 'heard' by most readers. Even so, it could be useful to provide the reader with alternative versions, either in terms of interpretation or in terms of presenting alternative compositional choices that were open to the composer. In the case of works no longer in copyright this could be done using MIDI versions of the scores (many standard works in the repertoire are now available). The author could present alternatives, or indeed provide the reader the tools to construct their own versions. These could then be played using internal MIDI instruments and the result compared with a reference version played from CD. All these possibilities require further investigation to determine their practicality and usefulness. The software itself, however, is highly flexible and takes full advantage of the range of options available within the MaxMSP programming language. To a large extent it is up to the imagination of analysts as to how far this approach can be used to enhance and enrich a range of analytical ventures. 6. REFERENCES [1] Clarke, M., A. Watkins. M. Adkins, and M. Bokowiec, M. "Sybil: Synthesis by Interactive Learning", Proceedings of the International Computer Music Conference, Miami, USA, 2004 [2] Clarke, M. "Jonathan Harvey's Mortuos Plango, Vivos Voco". In Analytical Methods of Electroacoustic Music, Mary Simoni (ed.), Routledge, New York, USA, (forthcoming). [3] Nattiez, J.-J. Music and Discourse.- Towards a Semiology of Music. Princeton, USA, 1990. Figure 1. A screen shot of one of the interactive exercises.