Page  97 ï~~Recent developments with the TAO physical modelling system Mark Pearson and David M. Howard Department of Electronics, University of York, Heslington, York. YO1 5DD. United Kingdom. email:, April 26, 1996 Abstract This paper describes recent developments with the TAO system for sound synthesis by physical models. The system is described elsewhere in the literature and is capable of producing complex, naturalistic sound events, imbued with spatial and physical qualities. TAO provides a holistic and intuitive approach to sound synthesis which includes visually informative and illuminating computer graphics animations of instruments. It has the potential to be controlled purely by physical gestures with future hardware developments. This paper describes three example TAO instruments, and presents a variety of sounds produced by similar instruments. 1 Introduction TAO (Pearson and Howard 1995), (Pearson 1995), is a system for sound synthesis by physical models. It is similar to both modal synthesis (Adrien 1991), (Djoharian 1993) and CORDIS-ANIMA (Cadoz, Luciani and Florens 1993) in that it makes use of masses and springs but has been developed from first principles according to a different set of criteria. The work was originally inspired by recent insights into cellular models and emergent behaviour and puts many scientific developments such as the study of chaos (Gleick 1991), complexity (Waldrop 1994) and dynamical systems into the context of synthesising 'organic', naturalistic sound events. The system has been designed to be accessible to those electroacoustic composers who wish to explore synthesis by physical models but who do not have the technical understanding or inclination to deal with differential equations. The system consists of a synthesis engine and a script language for describing instruments and performances. The structure of the cellular material and the script language are both described elsewhere in the literature. In this paper we will concentrate on some practical instruments and performance techniques which have been recently explored. The examples given are accompanied by sound examples which should convince the listener of the potential of the synthesis model to produce a wide range of naturalistic sounds belonging to different timbral categories. 2 A bowed stringed instrument Figure 1 shows an instrument with four strings which are glued to a rectagular resonator. The strings are tuned in fifths, hence their respective lengths. The cellular material has a constant wave propagation velocity so the frequency of a string is altered by changing its size not its tension. The component stringi is bowed at the point marked. Sound output is taken directly from the movement of the points marked 1 and r. The most important point about this example is that the instrument is a whole, i.e. the strings feed energy into the resonator which in turn feeds energy back to the strings. This does affect the bowed motion of the string as the exact moment at which the bow sticks and slips is affected by tiny variations in wave motion due to the resonator. It is well understood that even pitched instrumental sounds are not purely periodic and this kind of complexity to the model is audible in the results produced. Also the other strings begin to vibrate in sympathy when stringi is bowed. Sound examples from instruments of this family are presented. 3 An instrument with pitched circular sheets The main feature of the instrument in figure 2 is its six tuned circular components, tuned in that their sizes are chosen to produce a distinct set of pitches. In the sound examples presented, instruments from this family are played by striking these components. Microphone output is taken from the one dimensional resonator at the top of the image and the circular components are linked to this resonator by six smaller resonators each end of which is glued to the centre of one circular component at the points marked 1-6 and to the output res ICMC Proceedings 1996 97 Pearson & Howard

Page  98 ï~~resonator string4 st ring3 st ring2 wU" string1 Figure 1: A four-stringed instrument with a rectangular resonator COMMOn ft r d e f resonator._ E 4 2/ Rb e Fd F 4 Figure 2: An instrument with six tuned circular components and other resonators onator at the points marked a-f. By locally varying parameters such as the damping coefficients of each component, a variety of very bright metallic sounds, through cowbell-like sounds to almost wooden instruments can be achieved. 4 A prepared string buzzing against an obstacle The final example shown in figure 3 consists of a single string, or rather a one dimensional resonator which has had its characteristics fine tuned so that it behaves like a string. Two of the cells at the points marked mass have had their masses increased as if the string had been 'prepared' by pegging masses onto it. In addition an obstacle is placed in the way of the instrument forcing it to buzz against it. This leads to sounds which have a very natural evolution when the instrument is left to its own devices after being excited. It is a simple matter for the user to describe such scenarios as the following TAO script used to generate the example shows: String s: E6, 2 min;... pitch and decay time modify amount of damping locally and lock ends s.setdamping(left, 1/40, 0.02'/).lockleft; s. setdamping(39/40, right, 0.02,). lockright; prepare the string s(3/4).mass=50.0; s(1/4).mass=50.0; write sound samples to 'buzzoutfile.aif' Microphone micl: buzzoutfile, stereo; Parameter obstacle.position; Score 20 secs: At 0 secs for 1 msecs: apply force 1/10 of the way along s s(0.1).applyforce(1.0); change obstacle's position throughout performance obstacleposition = expon(9.0,0.5); simulate elastic collisions with obstacle If s(1/2).position > obstacle-position: s(1/2).position = obstacleposition; s(1/2).velocity,= -0.5; generate sound samples from either end of the string micl.leftout: s(0.05); micl.rightout: s(0.95); Pearson & Howard 98 ICMC Proceedings 1996

Page  99 ï~~obstacle d) MOSS obstacle e) b) Ml. obstacle C) obstacle }4",, marS it obstacle /f Figure 3: A prepared string buzzing against an obstacle 5 Conclusions References The three examples given here represent just some of the diverse possibilities available within the TAO system. All of the sound examples were produced directly by TAO with no external audio processing. The graphics examples were also produced by the system and images of the type shown are available to any user of the system. The sound examples illustrate the depth and organic quality of the sounds produced by TAO. The system was inspired by the flexibility of cellular models, many examples of which are given by Toffoli and Margolis (1987), and the success with which they have been applied to simulating the behaviour of a variety of natural systems. Other areas of inspiration include such contemporary areas of study as chaos, complex dynamical systems and the question of how form, order and pattern emerge naturally from systems consisting of large numbers of elements interacting in simple ways. The philosophy behind the whole system is that there is something direct, intuitive and appealing about cellular models. Possible future developments of the system will include: porting it to parallel processing hardware; developing a graphical (possibly virtual reality) interface to allow instruments to be designed and played directly; applying the underlying synthesis engine to speech and singing synthesis; and putting the system through its paces in its intended area of use, as a compositional tool, partly because the sounds have reached a level of quality which begs further creative exploration, and partly to provide feedback for future improvements to the system's musical functionality. Adrien, J.-M. (1991). The missing link: Modal synthesis., Representations of Musical Signals., Cambridge, Massachussetts. MIT press. Cadoz, C., Luciani, A. and Florens, J. L. (1993). CORDIS-ANIMA: A modeling system for sound and image synthesis, the general formalism., Computer Music Journal. 17(1): 19-29. Djoharian, P. (1993). Generating models for modal synthesis., Computer Music Journal 17(1): 57-65. Gleick, J. (1991). Chaos - making a new science., London, Cardinal. Pearson, M. (1995). TAO: a physical modelling system and related issues., Organised Sound. 1(1):-. Pearson, M. and Howard, D. M. (1995). A musician's approach to physical modelling., Proceedings of international computer music conference., pp. 578-80. Toffoli, T. and Margolis, N. (1987). Cellular automata machines - a new environment for modeling., MIT press, Cambridge, Massachusetts. Waldrop, M. M. (1994). Complexity: the emerging science at the edge of order and chaos, England. Penguin. ICMC Proceedings 1996 99 Pearson & Howard