Index   Sounds
D. Keller - touch'n'go: Ecological Models in Composition


Unlike science, art, and specifically music, does not center its attention on the observation of natural phenomena. On the contrary, the main trend in European and North American music composition and theory seems to be the production of art objects completely unrelated to everyday sounds. The reproduction or description of environmental sounds falls broadly outside the field of music composition and analysis. Nevertheless, during the last thirty years there has been a growing interest on mundane sound as source material for composition and even as an object of study by itself. Based on the work of R. M. Schafer (1977), two new areas of music research opened: acoustic ecology and soundscape composition (Truax, 1996).

Another aspect that has been noticeably neglected in compositional and theoretical practices is the social context in which a given musical piece is produced. How this context influences musical thought has been generally approached from the anthropological and/or sociological perspective. Nevertheless, with few exceptions, little attention has been paid to the structural aspects of music while focusing on the relation between sound environment, cultural context and music structure.

Just as natural sound does not generally have a place in compositional techniques, perceptual processes in music listening have only recently been regarded as relevant for music composition. In this respect, J. Tenney's and P. Oliveros’ work stand as an example of an alternative approach. A healthy exchange between music composition and music psychology has timidly started, but a full-blown interaction is yet to come. In fact, a great corpus of research in music perceptual phenomena has been gathered during the last ten years, creating a firm ground for the development of music psychology. This knowledge could eventually be applied to music creation processes and music theory. Similarly, new developments in music composition and theory could raise well-focused questions in experimental research. Our work is intented as a humble contribution to bridge this gap.

We need to make several non-trivial assumptions to advance in our argument. As put forth by Kelso (1995, 92) and tested in various experimental settings, sound production mechanisms and perceptual processes are inextricably interrelated. Keeping in mind the innovative profile of his theory, it would be useful to provide specific examples which pinpoint its implications in environmental sound compositional techniques. We will not deal directly with these issues here. The reader should refer to (Keller, 1998b) for an in-depth discussion. We propose that musical processes cannot be isolated from sound structures. Thus, morphology, syntax and sound (Keller & Silva, 1995) are just levels of musical structure which can be used for analytical purposes. Nevertheless, no hierarchical relationships can be implied since all time levels interact influencing the final musical percept. This is shown in several synthesis examples included with this paper, and is documented in experimental research (Tróccoli & Keller, 1996). This interaction among levels provides the key to a dynamic-pattern formation process which underlies time-varying percepts (Kelso, 1995, 224).

Finally, we define sound source recognition as the axis that links sound production mechanisms, music perception processes, and natural and cultural context. These processes can be observed at various simultaneous time levels. The interaction of these processes results in musical phenomena which we can manipulate compositionally. Therefore, understanding the processes of sound source recognition becomes a key aspect of composing with environmental sound.

Given the breadth of issues relevant to an ecologically-informed compositional approach, we will summarize in a few paragraphs our two-year research work in ecological models. After discussing the theoretical foundation of our work, we will briefly describe the technical issues involved in the use of ecological models in composition. As a case study we will present the organizational procedures, the underlying compositional concepts, the socio-contextual references, and a few sound examples of our tape piece, touch’n’go.

The four domains

The problem of sound source recognition can be studied from four perspectives: (1) the formal domain: structure of systems and models of sound production (theoretical and algorithmic), (2) the acoustic domain: sound parameters, organization in time, frequency, and space (3) the perceptual domain: processing of sound stimuli and correlated mental processes, (4) the social domain: the natural and cultural sound environment.

Table 1. Perspectives in sound modeling.
Perspectives in sound modeling
formal domain
acoustic domain
perceptual domain
social domain
structure of systems and models of sound production (theoretical and algorithmic).
sound parameters, organization in time, frequency, and space.
processing of sound stimuli and their mental representation.
the natural and cultural sound environment.


Although the formal domain belongs mainly to epistemology, as exemplified in Kampis (1991), music theory and sound synthesis methods can deeply affect this domain (Truax, 1992). System theory has already been applied to musicological studies (Georgescu & Georgescu, 1990; Boon, 1995), unveiling a wide range of questions that need to be confronted. How do the microproperties of acoustic events influence the perception of macroproperties? Could we hypothesize the existence of a meso-level representation? Is there a signal space defined by social, perceptual and physical properties of the sound environment? Are sound events categorized in relation to a "perceptual grain?" Are dynamical models more suitable than fixed categories? Is musical time irreversible?

There is a wealth of research in acoustics relevant to synthesis and sound organization techniques (Smith, 1992). Physical modeling generally places emphasis on the study of resonant structures that respond dynamically to different types of excitation. As J. O. Smith (1997, 221) puts it, "a mathematical model of a musical instrument becomes the instrument itself." Simplified resonant systems are strings, tubes, membranes, either elastic or stiff. Sound-producing systems are the source of energy fed into the resonant system (Fletcher, 1992, 6). Another useful technique for modeling sound-producing systems is granular synthesis (Roads, 1996; Truax, 1988). Since granular synthesis can be applied to produce complex excitation patterns, bringing together both approaches provides a comprehensive way to deal with dynamic excitation of existing resonant structures.

During the last few decades, auditory perception has become a flourishing area with well-grounded and, often, conflicting theoretical approaches. The oldest paradigm belongs to psychoacoustics, a branch of psychophysics that focuses on sound. Psychophysics attempts to relate physical descriptions of stimuli with measurable behaviors to sensation (McAdams, 1987, 5). Its methods and underlying philosophy are tightly linked to the behaviorist approach. In other words, psychoacoustics concentrates on the problem of finding models that relate subjects’ sensations with physical representations of stimuli (Parncutt, 1989). Since an exact definition of the sound object is needed to implement a computational model, this acoustic representation may be useful when mapping sound parameters onto perceptual results - a fancy way to say ‘composing.’

A body of theories that has gained weight during the last fifteen years is what has been called ‘the cognitive approach’ to music. Many disparate methods with dissimilar objects of study share some basic assumptions which entitle them to be labeled as ‘cognitive.’ Krumhansl (1990, 5) describes cognitive psychology as "a subarea of experimental psychology concerned with describing human mental activity." "One of the major contentions of the cognitive approach to psychology is that all mental activity is mediated by internal (or mental) representations." (McAdams, 1987, 18). As we discussed in (Keller & Silva, 1995), this mental representation can take two forms, symbolic or subsymbolic. The idea of a mental representation as separate from the actual perceptual process is the main difference between cognitive psychology and ecological psychology (Kelso, 1995, 194; Keller, 1998b). Ecological psychology studies acoustic phenomena by observing the physical characteristics of a sound event, the high-order configuration of variables, and the listener’s ability to detect the information provided by the event (Gibson, 1966; Kelso, 1995; Michaels & Carello, 1981).

Within the ecological approach, we have discussed the relationship between music and its social environment (Keller, 1998a). These are the conclusions of this study: (1) music production and perception are dependent upon the social structure where they take place; (2) social structures are not only based on economical and political dynamics but also on their cultural representation. Music is a key factor in the shaping of this representation.