SOUND FUNCTORS APPLICATIONS Jônatas Manzolli2, Adolfo Maia Jr.1,2 Mathematics Department1 Interdisciplinary Nucleus for Sound Communications (NICS)2 University of Campinas (UNICAMP) - Brazil ABSTRACT In this work we show that the concept of mathematical structures called Functors can be useful to search and develop a large number of procedures and compositional algorithms in Computer Music. We start introducing Functors, Categories and some generalisations. It follows two applications: a functor between plane curves and sound generating structures based on MIDI events, and a functor between C(x) defined in a finite interval I and the class of Fourier Spectra A(). INTRODUCTION It is well known that music composition is plenty of rules that could be related to mathematical symmetries. An interesting discussion about it is presented in (HOFSTADTER, 1989), a study relating music, design and mathematical structures. Music symmetries could be better understood if we study the underlined or hidden mathematical structure, which generates them. In this work we show that, for the most cases, the intuitive use of symmetry and associations in music can be studied and explored through the mathematical concept of Functor. This way of thinking is interesting, not only because it allows us to classify sound objects mathematically but also, and perhaps more importantly, it furnishes new relations and associations that point out to an enormous variety of sounds and composition devices. In short, we claim that Functor, in the same way it was firstly devised in mathematics (Maclane & Birkohoff, 1953, 1979), can be a kind of universal tool to construct mathematical models for music composition and sound synthesis. On the direction of applying mathematics to build sound structures there has been a series of approaches such the use of 1/f noise fractal distribution (Voss & Clark, 1978; Bolognesi, 1983), non-linear dynamical systems and iterated function systems (Pressing 1988; Scipio, 1990; Gogins, 1991) and there is a study about these systems in (Manzolli, 1993a). 2 In line with these concepts, our group has worked on Computer Music and our focus is driven to: new methods for Sound Synthesis using non-linear dynamics (Manzolli, 1993b), algorithm composition using Markov Chain and Boundary Function (Manzolli & Maia, 1995), and development of interactive desktop and gesture interface for composition in real time (Manzolli & Ohtsuki, 1996). In this paper, we discuss Functors as generalised Algorithmic Compositional tools for sound construction. It follows a section in which we introduce categories and functors, followed by applications and the computer implementation. CATEGORIES AND FUNCTORS As its own name suggests, a Functor is a kind of function, but not an ordinary one. The difference to the usual ones is that a Functor carries the underlined structures of the sets, which is applied. These sets are called Categories. More precisely, a Category is defined with three kinds of data a) A class of objects A, B, C… b) For each pair of objects A, B we have a set of applications (morphisms) M(A,B) from A to B. c) For each triple of objects A, B, C we have a composition law for the morphisms M(A,B) M(B,C) M(A,C) (f,g) g O f Now, for these data we have the following axioms, which are the primary properties of these categories: A1) The sets of morphisms M(A,B) and M(C,D) are mutually disjoints unless A = C and B = D. A2) Associative Law: h(gf) = (hg)f . A3) Existence of Identity: For each object A there identitysuch that for any f: A and f of and g = g. exists a morphism g: C we have 3 Usually, the theory of categories and functors are mathematically involved. Here we use only the consequences of the above general definition and properties in order to show that we can map underlined structures from categories of mathematical objects to structures in categories of sound objects which undertake some formal properties from their mathematical counterparts. There are too many examples of mathematical categories and functors. Since we are most interested in music, in the next section we give two simple applications of the concept of functor to be applied to sound categories. Given two categories and , a functor F between and is a map which associates each object A to an object F(A) F: A F(A) and for each morphism f M(A,B) associates a morphism F(f) M(F(A), F(B)) with the properties F(gf) = F(g) F(f) F(1A) = 1F(A) Observe that the functor F operates on the morphysms as well as on the elements of a category. The properties above simply mean that a structure of the product (or composition) between two morphisms in category is transported for the morphisms in category via functor. There are too many examples of mathematical functors. Since we are most interested in music, in the next section we give two simple applications of the concept of functor to sound categories. APPLICATIONS EXAMPLE 1 - Sequence of MIDI Events We begin with the mathematical category = continuous finite curves in a bounded region U R2. Given two curves C1, C2 we define a morphism in as an application f: C1 C2 x f(x) which means to deform C1 to C2. In other words f M(C1 ,C2). The product gf is defined as the composition gf = gof 4 f g C2 C1 C3 Figure 1. Composition of Plane Curves Also we define as the category of functions which control sound parameters. We use MIDI (Music Instrument Digital Interface) parameters to define the category . So F(f) M(F(C1 ),F(C2)) and F(f) deforms F(C1) to F(C2). Given a curve C in , the function F(C) can be defined in several different ways. In this example we define F(C) as the distance function between a fixed point in the plane to the points of the curve C. It is easy to see that the functor property is satisfied, namely F(g(f(C))) = F((gf)(C)) = F(g)F(f)(F(C)) C2 C1 f F g F F F(f) F(C1) C3 F(g) F(C2) F(C3) Figure 2. Diagram of the Functor operation EXAMPLE 2 - Sound Synthesis Derived from the traditional Sound Synthesis method called Waveshaping (LEBRUN, 1979), we show an application of a functor, but in this case a kind of Transfer Function is used to build Spectral Envelops. We take as mathematical category, a class of smooth functions C(x) defined in a finite interval I and as sound category, the class of Fourier Spectra A() of a finite set of sounds. The function C(x) acts as a shaper of the Set of Spectra. Starting from a fixed spectrum, which we call Input Spectrum, we use deformation of curves as morphisms between both classes of functions. We construct the following Functor: 5 F: C F(C) = C(A()) Where the parameter x = A(), with 0 x Amax, and Amax is the maximal amplitude considered . The morphisms in are deformations of smooth functions and the morphism in can be chosen as F(f)=f. This means that the morphisms in can also be used as deformations on the Set of Spectra. By the above definition of our example of functor it is easy to see that they satisfy the property F(g) F(f) = gf = F(gf). 1 C(x) 0 Amax Input Spectrum A x 0 Output Spectrum A F 0 Figure 3. Spectrum Functor Diagram It is an endless work to construct examples like the above ones. Other mathematical structures like groups, lattices, algebras, several different geometrical and topological structures can be used as generating categories (see MACLANE 1971) for sound outputs through a suitable choice of functors. In this sense Functors allows to construct a virtually infinite number of sound outputs, with the additional advantage of reflecting the structure (or symmetry in several cases) of the external mathematical category used to generate the sound material. Musicians who work on Computer Music, aware of the properties of functors, can expand their sound tools in a way, certainly, not yet explored by the authors of this paper, depending on their mathematical skills which, in order to produce good results, are not required to be too much sophisticated, added to music capability and their own aesthetic view. 6 CONCLUSION The use of mathematical tools in composition brings new possibilities for composers envision the construction and development of Compositional Systems. Research in Computer Music provides powered tools for constructing these systems. This union of artistic and mathematical knowledge creates a framework for investigation and music production, a environment for applying mathematics to sound structures manipulation. Mathematical models presented here could be expanded using graphic interfaces similar to the computational implementation we discussed. These will create new musical performance situations making mathematical design to produce computer music instruments. REFERENCES Bolognesi, T. (1983). "Automatic Composition: Experiments with Self-Similar Music." Computer Music Journal 7(1):25-36. Gogins, M (1991). "Iterated Functions Systems Music." Computer Music Journal 15(1):40-48. Hofstadter, D. R. (1989). Gödel, Escher, Bach: an eternal golden braid. Vintage Books, New York, ISBN 0394-75682-7. LeBrun, M. (1979). "Digital Waveshaping Synthesis." Journal of the Audio Engineering Society, 27(4):250266. MacLane, S. (1971). Categorias for the working mathematician, New York, Springer-Verlag. MacLane, S. & G. Birkhoff (1953). A Brief Survey of Modern Algebra, New York, the MacMillan Company. MacLane, S. & G. Birkhoff (1979). Algebra, New Yok, Macmillan Company. Manzolli, J. (1993a). Non-linear Dynamics and Fractals as a Model for Sound Synthesis and Real Time Composition. PhD Thesis at the University of Nottingham, England. Manzolli, J. (1993b). "Musical Applications Derived From FracWave Sound Synthesis Method". Proceedings of the 94th Audio Engineering Society Convention, Berlin. Manzolli, J. & A. Maia Jr (1995). “Interactive Composition Using Markov Chain and Boundary Functions”. Procedings of the XV Congress of the Computer Brazilian Society, II Symposium on Computer Music, Canela, Brazil. Manzolli, J. & W. Ohtsuki (1996). “INTERASOM: a desktop for Algorithmic Composition”. Procedings of the XVI Congress of the Computer Brazilian Society, III Symposium on Computer Music, Recife, Brazil. Manzolli, J. (1994). "FracWave: Non-linear Dynamics as Timbral Constructs". Anais do XIV Congresso da Sociedade Brasileira de Computação, I Simpósio Brasileiro de Computação e Música, Caxambú - Brasil. Pressing, J. (1988). "Non-linear Maps as Generators of Musical Design." Computer Music Journal 12(2):35-45. Rowe, R. (1993). Interactive Music Systems. MIT Press, Cambridge, Massachussetts, ISBN 0-262-18149-5. Scipio, A. (1990). "Composition by Exploring of Non-linear Dynamic Systems" Proceedings of the 1991 International Computer Music Conference. Voss, R. F. & J. Clarke, (1978). "1/f noise music: Music from 1/f noise". Journal of the Acoustical Society of America 63(1):258-263. 7 ACKNOWLEDGMENT We thank Prof. Raul do Valle for fruitful discussions and the Foundation for Research of State of São Paulo - FAPESP for supporting this work. 8