Axis allies cd rom patch version 1.331

Reply Reply as topic. This topic has been deleted. Only users with topic management privileges can see it. Here is a help forum. Thanks, KURT. Will it work on Vista? M: the main menu is all glitchey.

See you on the Warzone! So it is unstickied now. First post. Suggested Topics. I assume many of you have this game. Which version do you feel gives the Axis the best odds without going through a whole pile of trouble using the editor feature to level things out?

I am currently on the look-out for the Pacific with , and printing, the Spring Anniversary Edition, and the Jedko Games editions. Is it possible to save the game state on TTS and send a link or something between players?

Register Login. Reply Reply as topic. This topic has been deleted. Only users with topic management privileges can see it. Update: All router and firewall issues have been defeated! Thank you very much for the info and your time. Will this patch make the version work on the windows Vista platform? But thank you very much for the help. I also run XP, so does the latest patch provided here work? After you download the latest patch go into your axis and allies directory.

Wow, great to see a new patch for this classic CD!!! I installed the patch and followed all of the directions, but after launching the game I get a jumbled screen that looks like this: Anyone know what could be causing it? Thanks for the reply, kurt.

First post. Suggested Topics. So it is unstickied now. Any help is appreciated. D Hi Folks! Note that with this scheme no member of the population is excluded from breeding, they all have some chance of contributing to the next generation. However, this method can result in too strong a selective pressure in favour of individuals that are relatively good at the early stage but may actually be far from optimal; the population prematurely converges to be dominated by copies of such individuals.

Rank-based selection is a particularly straightforward alternative scheme that provides more control over the selective pressure and allows strong differentiation of the population, even at later stages when their fitness values are very close.

Using this strategy the population is ranked, or ordered, according to the fitness values of its members. Selection is then performed by following a pre-determined probability distribution function, such as the ones shown in Fig. This may be a simple linear function that constrains the first-ranked fittest individual to be twice as likely to be selected as the median-ranked individual, or something more complex.

An alternative form of selection, that makes most sense in the context of parallel EAs, was alluded to earlier — the use of local selection rules.

Typical rank-based selection probability distributions. Local schemes may be based on the methods described earlier or may be simpler. For a more detailed discussion of possible selection schemes see Eiben and Smith and Mitchell Interactive EAs, employing human-based aesthetic selection, effectively dis- pense with a separate selection method: individuals are picked out by the user to act as parents for the next generation.

The Genetic Operators The genetic operators maintain variation in the population and create new indi- viduals from old ones. Myriad specialised operators have been developed over the years and there are numerous variation on the standard ones.

Hence only a few of the most common generic operators will be outlined here. The two most common are cross-over and mutation see Fig. Like most widely used oper- ators, they have strong stochastic elements to their operation. Simple cross-over involves choosing at random a cross-over point some position along the string for two mating chromosomes — two new strings are created by swapping over the sections lying after the cross-over point.

Variations include two-point cross-over where randomly selected sections of the strings are swapped over and special oper- ators that rearrange genes during the crossing over, either in order to keep the new solutions legal or to make them better Michalewicz et al. Mutation changes the value of a gene to some other possible value.

Depending on the encoding, this might entail assigning a new value at random from the entire range of possible x-over point cross-over mutation FIGURE 1. Schematic of popular genetic operators. Mutation operators can be heuristically guided, rather than com- pletely blind e. For complex encod- ings, it often makes sense to have several different mutation operators acting in parallel. Other operators sometimes used are inversion, which is simply a matter of reversing a randomly chosen section of a single genotype; translocation, which involves moving a randomly selected section to another place on the genotype; and duplication, which entails adding extra copies of genes or groups of genes.

The latter operator makes sense only in circumstances where a variable length encod- ing is being used; it often functions in tandem with a deletion operator. Specially designed cross-over operators can also be used to allow genotypes to grow and shrink Harvey Special domain specific operators are regularly employed to good effect.

For instance, in the application of EAs to musical composition operators based on musical transformations such as inversion and transposition can be very useful Biles The operators have assigned rates that determine how likely they are to be used. They are applied at the offspring creation stage according to a routine like the following: When two genotypes are selected for breeding, first apply crossover with some high probability to create two new genotypes. Next apply inversion to these with a medium probability.

Finally, each gene on the resulting genotypes undergoes mutation with a low probability. According to the encoding scheme and problem area, different combinations of operators with different rates are used. In some circumstances it makes sense to dispense with cross-over, for instance if it is difficult to devise an encoding that works with this operation, and just use one or more mutation operator.

It is common to have to experiment with operator rates to find good settings, which can usefully be made to vary during the search — in some cases the rates themselves are put under genetic control Back et al. The Replacement Scheme In some EAs enough offspring are produced on each cycle to replace the entire population in one go. In others, sometimes called steady-state algorithms, new individuals are introduced one at a time, as long as they are fitter than at least the worst member of the population which is then replaced.

This allows a more gradual search. Other schemes use an inverse selection method to choose members of the current population to be replaced with a bias towards the least fit.

It should be clear from this brief outline of EAs that there are many choices to make in deciding how to apply them to any given domain and many parameters to tweak once the basic algorithm has been designed. The various elements of the EA must all work well together in order to achieve good results.

The best choice of operators, genetic representation, evaluation function and so on can be either guided by what has been shown to work in the past or by experimentation with different settings and options. An Introduction to Evolutionary Computing for Musicians 9 theoretical understanding of how EAs work can be very helpful and save time spent down the blind allies of poor representations or inadequate fitness functions.

EAs are complex nonlinear stochastic systems, which makes them extremely difficult to analyse. Hence the theoretical literature tends to be rather inconsistent and is often contradicted by empirical results.

However, there is useful information to be gleaned and good sources include Schmitt , Vose and Wright et al. Related Developments EAs have played an important part in the development of the related fields of Arti- ficial Life, which is concerned with the synthesis and analysis of lifelike processes in artificial media Langton ; Pollack et al. These areas have seen interesting explorations of phenomena and techniques that have found applications in artistic endeavours.

This is a direction which might hold some promise as far as computer music is concerned. Multiple species could represent different voices in a composition or it might be possible in some situations to have coevolving species of compositions and critics who evaluate the compositions developing towards some interesting end Werner and Todd , ; Hillis In this system, simulated beings populate an artificial world in which they can move around and make and hear sounds.

These sonic agents must compete for limited resources in their environment. The agents generate sounds to attract mates and also to capture the imagination of the audience, since its response has a direct affect on the virtual environment, particularly the growth rate of food. In this work McCormack has demonstrated the successful use of an open-ended automatic evolutionary process to generate a highly engaging interactive artwork.

This system illustrates a more implicit approach to fitness evaluation, with a fairly oblique interaction element. Such pieces suggest a wealth of opportunities for musical developments.

Applications of Evolutionary Computing in Music There is a growing body of work involving the use of EAs in musical applications see Horner and Goldberg ; Burton and Vladimirova ; Bilotta et al.

In music, the two areas that have attracted the most attention are composition and sound design. However, extending such work to more creative and original compositions is challenging, for reasons including those discussed in Section 1. In the area of sound design, researchers have demonstrated the efficacy of the technique in controlling sound synthesis methods, both to explore new sounds and to develop synthesis algorithms for existing target sounds Johnson ; Dahlstedt ; Garcia ; Mandelis Various aspects of these topics, in relation to specific systems, will be dealt with in detail in later chapters.

Musical composition with EA will be discussed in more detail in Chapters 6—8 and EA in sound synthesis and design will be discussed in more detail in Chapters 3, 4 and 8. The remainder of this chapter is intended to raise a number of important issues in these areas as background and context to the rest of the book. Fitness evaluation turns out to be a particularly thorny issue in relation to compositional systems and it is not a trivial matter in sound design. Evolutionary Computing in Musical Composition 1.

Introduction The main purpose of this section is not to attempt a comprehensive survey of evolutionary computational approaches to musical composition see Burton and Vladimirova for a good overview as well as later chapters in this book but rather to highlight some of the potential problems, apparent in the literature, of too close a marriage between the development of compositional computer programs, and an approach to musical form derived primarily from academic theory, rather than what many composers demonstrably do.

This is a very real problem since textbook musical form is by its nature algorithmic and has often been seen as the ideal starting point for the development of composition programs, particularly those based on pre-existent models of compositional practice see Wiggins et al. This is not to suggest that the production of an interesting sonata structure per se is a primary goal of more than a minor- ity of practitioners in this field.

Rather, that sonata form itself was a significant tool whether algorithmic in nature or not in the evolution of complex musical structure for more than years in the history of Western Art Music.

Its potential to encompass so many elements that inform the creative process — exploration, contrast, development, transformation, motivic mutation, etc. An Introduction to Evolutionary Computing for Musicians 11 importantly the genotype, the genetic operators and the fitness function. If an au- tomatic fitness evaluation method is used, the desired musical outcome must be somehow formally codified.

Deriving sets of rules to describe particular forms or styles is fraught with difficulties, as discussed below. If the automatic fitness function problem is sidestepped by using human evaluation, the search space de- fined by the genetic representation and operators must be sufficiently constrained to avoid impossible bottlenecks in the time needed to perform the evaluations Biles ; Gartland-Jones and Copley ; As this will entail encoding musical knowledge into the representation and operators, the difficulties do not disappear.

Algorithmic Composition In his book The Algorithmic Composer, David Cope stated that throughout the history of Western Art Music, composers have used algorithms as part of the creative process. However, while it is perfectly possible to define some compositional processes as algorithmic, not all fall so neatly into this category.

A necessary preliminary step would be to attempt a delineation of boundaries, as to what extent, which compositional processes can or cannot be so defined. Part of his support for this proposition is a series of examples of compositional processes defined as algorithmic.

This is a wide- ranging set of examples although, with the exception of the motet, all bear only a peripheral relation to musical composition as actually practised by fully fledged composers.

In the seventeenth century, Giovanni Bontempi proposed that his rota as a guide by means of which one thoroughly ignorant of the art of music can begin to compose; a sort of musical equivalent of painting-by-numbers. For example, imagining a song form of the medieval period, a dance form of the baroque, or a sonata allegro form of the classical period of Western music history as symbols in a flowchart — one way to describe an algorithm — does not seem unreasonable.

Cope , pp. This is not to suggest that Fux is valueless as an example of a producer of musical algorithms, rather simply that the process of modelling anything more than the most elementary compositional process is rather more complicated than his citation by Cope might suggest. Historically, a large claim made for the ben- efit of strict counterpoint study of the Fuxian variety was that it provided what amounted to an algorithm for composing in the style of the composer Giovanni Pierluigi da Palestrina, who had been regarded for centuries after his death as a byword for purity of contrapuntal style.

Unfortunately, this claim was largely unfounded and was completely exploded by Morris as far back as the s: Yet the rules of Mr Rockstro [another author of a book on strict counterpoint] are not peculiar. They are, more or less, the same as those found in almost every textbook of counterpoint. Who invented them, goodness only knows: why they have been perpetuated, it passes the wit of man to explain. Music written to meet their requirements is something altogether sui generis, a purely academic by-product.

The rules of counterpoint are found to have no connexion with musical composition as practised in the sixteenth century: are we to abandon the rules or to abandon the sixteenth century? Follow Byrd and Palestrina, or follow Mr. Rockstro and Professor Prout? Morris , p. Is Sonata Form an Algorithm? Sonata form expressed as an algorithm brings similar problems in its wake. Is the algorithm to be based on textbook definitions or on what significant composers actually produced?

Furthermore, there remains the question of which variety of sonata form as practised is to be taken as the starting point. Many commentators see, e. Rosen , pp. An Introduction to Evolutionary Computing for Musicians 13 in the second half of the eighteenth century, the sonata form not termed as such was an elaborated binary structure characterised by differentiated key areas. The first part contained a tonic area and a dominant or related key area, although the first area could be characterised by a modulation to the tonality of the second area.

The second part consisted of an area of rapid modulation or episode followed by a return to the home key in which tonality the movement remained until its end. The two-part view of sonata structure is confirmed by the prevailing eighteenth-century practice of repeating both sections, rather than just the first part, as is usually the case in contemporary performance.

What is set out above is just about the fullest extent of universal common ground in composing practice that can be extrapolated from the majority of later eighteenth century sonata structures and a composing algorithm extracted from this would be little different from one derived from baroque binary dance patterns, despite the two forms being in reality quite distinct from each other.

The beauty of the form lay in its flexibility. Its greatest merit, which enabled it to hold a commanding position over a period of years, is its extraordinary flexibility in accommodating the widest variety of musical ideas, long or short, many or few, active or passive, in almost any combination.

The internal details may be subjected to almost any mutation without disturbing the aesthetic validity of the structure as a whole. Schoenberg , p. For a truly distinctive sonata algorithm, resembling neither the simple binary nor the ternary model, we would need to turn instead to the traditional theorists, who would state that Sonata Form consists of firstly, an exposition, comprising first and second subject groups, respectively in the tonic and dominant or related keys and linked by a transition or bridge passage; secondly, a development section, in which the original thematic material will pass through a variety of related keys and may be extended by episodes; this will be followed thirdly by a recapitulation, in which the material from the exposition returns but is mostly confined to the original key.

Actually, this theoretical description does indeed correspond to more conser- vative later nineteenth century practice and this lends a depressing sameness — from a purely formal point of view to the majority of sonata-type structures from this period. It is difficult to avoid the conclusion that once the rules had been encapsulated in a detailed formal scheme or algorithm, the sonata began to lose its dynamic and developmental possibilities and its various sections took on the character of moulds into which appropriate music could be poured.

Such an approach to potential sonata material would have been psycholog- ically impossible for any major eighteenth or early nineteenth century composer of whose structure-building creativity generally went beyond simply following formulae devised by others.

The Dangers of Too Many and Too Few Rules It may seem that several of the preceding paragraphs address issues more cen- tral to the concerns of musical historians, analysts and aestheticians than those of designers of computer programs for musical composition.

However, if we are modelling musical creative processes to any degree of sophistication, it is crucial that we base our model on something close to what composers actually did, rather than on theoretical constructs, often established long after the creative event, that oversimplify or distort complex thought processes in the interests of pedagogical expediency.

An excessively rule-based system stands in grave danger of produc- ing little more than schoolroom exercises or, at best, stolid replications of good craftsmanship because no facility has been provided for expanding a given search space to accommodate the possibility, indeed the desirability of the unexpected, or even iconoclastic but still meaningful musical idea or development.

An Introduction to Evolutionary Computing for Musicians 15 Although the explorative and stochastic nature of evolutionary search are help- ful, this is perhaps the most challenging problem facing the designer of an EA-based composition program, whether for general use or tailored to one particular set of preferences. But without the most stringently defined search space an unman- ageably large amount of potential material, mostly unusable, is apt to be produced.

Biles has described this situation as the fitness bottleneck. As Werner and Todd pointed out: More structure and knowledge built into the system means more reasonably structured musical output; less structure and knowledge in the system means more novel, unexpected output, but also more unstructured musical chaff. These are not isolated, eccentric examples but the essence of a truly creative use of form, wholly characteristic of their respective composers, which can lend musical compositions their enduring power to fascinate and hold the attention.

Meaningful contradiction of expectation is one expression of individuality that distinguishes specific pieces and composers from the more typical cultural products of whatever age in which they lived, giving the music an intrinsic value that can transcend time and place.

To attempt to model this level of creativity is asking much of a process still in a comparatively early stage in its development but it seems vital that the possibility of overriding rules must be provided for in composition programs with any preten- sions to model creative, rather than reproductive musical thought.

The historical fact that theory so often followed, and in the process distorted, practice should in it- self be warning enough of the pitfalls of regarding compositional processes purely as algorithms.

It is natural to have recourse to algorithms when modelling cre- ative processes, as every computer program ever devised is in essence algorithmic. However, it must also be recognised that if the algorithm employed is reductive and constricting in relation to the process it is modelling, the musical interest of what emerges will be at best limited, if not utterly predictable.

The system uses virtual blocks, which have the ability to both play and compose music. As the blocks are arranged in various structures they interact with each other in ways that influence the emerging music.

To do this, it uses an EA that is initialised with the home music and has the incoming phrase as its compositional target, allowing the use of an automatic evaluation function that measures the closeness of fit to the target.

The path taken by the EA generates intermediate material related to the home and target pieces. The user can stop the evolutionary processes at any stage and restart it with new incoming phrases, as well as set parameters that control how far the evolutionary process will travel between the two pieces.

To quote the designer any number of blocks may be chained or grouped in any 3D structure. If a block is passed some music from its neighbour, it first recomposes itself, and then passes its new music on to all of its neighbours, and so forth within a pre-specified range. It is important to clarify that each block holds on to its home music throughout, enabling any music composed by it to remain thematically related, despite the constant process of re-composition undertaken by each block.

In this way the composer of the music for all blocks maintains a compositional thumbprint on the evolving musical structure. Gartland-Jones and Copley, , p. Evolutionary Computing in Sound Design The use of EAs at the sound level is concerned with the manipulation of parameters that define a sound, using a particular sound synthesis technique SST , or with parameters that define a particular deformation on an input stream sound effects.

An Introduction to Evolutionary Computing for Musicians 17 new sounds. These areas are briefly introduced in this section while highlighting pertinent issues. In the optimisation case a sample of sound, often from a traditional instrument, is used as a target waveform.

An EA is put to work to derive the parameters of a particular SST to produce a sound as close as possible to the target. A fitness function that measures the difference between a candidate sound and the target is usually employed and there are many technical issues involved in how best to define this.

There are a number of examples of successful uses of this approach e. Garcia Sound definitions usually describe a singular point in the parameter space of the SST without explicitly detailing how this sound changes and deforms from that frozen point. Such deformations of sound, or movements in parameter space, are necessary for mapping the sampled instrument to a keyboard and note scale, and implementing other transformations that add expressivity to the sound.

In order to map those dynamics from the original source of the sampled waveform, generally a large number of waveforms is needed.

As an absolute minimal requirement, at least three distinct waveforms would have to be used for each degree of freedom of the original sound source. For instance, if the source is a piano sound, the degrees of freedom of the piano would include: the key position, velocity, aftertouch and so on. In practice most acoustic instrument sounds do not vary in a linear fashion along their axes of freedom and far more than three samples would have to be used for each axis.

This can very easily result in a prohibitive number of samples, which places too high a computational demand on the EA. This can be a serious problem only if this technique is used to faithfully emulate an original sound source.

In contrast, if such fidelity is not required, then some interesting possibilities may begin to emerge. For instance, if the specific parameters are derived from a single waveform, then any deviation from these parameters will create sounds that are similar to the original but with deformation characteristics that depend on the particular SST used.

For example, if a piano sound is used to derive the parameters for a frequency modulation FM SST and a physical modelling SST, then the deformations afforded by the former would be unique to this particular implementation of FM and for the latter unique to the particular physical modelling used.

In effect there would be two instruments that would sound very similar at some performance configuration, but at the same time they would behave very differently in terms of sound deformation when the performance configuration changes. The second category of EA-based sound creation, that of developing new sounds, requires a somewhat different approach. This is partly because of the complexity and lack of transparency of SSTs and partly also because of the difficulty in modelling aesthetic judgements.

In this domain, the subjective usually rules over the objective. Although there are general problems such as maintain- ing a consistent judgement of quality, the time taken to evaluate a sound is usually considerably less than that for a composition.

This means that it is often feasible to run the algorithm for a reasonable number of cycles. The less constrained approach necessitated by the lack of formalised knowledge allows for a powerful exploration of sound space — the user is free to navigate a world of sonic possibilities, turning up interesting and unexpected new forms that can be put to good artistic use. Genophone Mandelis , is one such exploratory system, designed in part to allow a flexible exploration of sound spaces without the need for detailed understandings of SSTs.

We will now briefly describe aspects of the system, fo- cusing on general issues in the way evolutionary search is used. The system makes strong use of genetic recombination, which in biological systems is a creative pro- cess in itself.

A biological analogy would be the breeding of animals or plants, which humans have done for millennia. When pigeons are bred, for example, it is not normal at least not yet to employ gene level manipulations via genetic engi- neering. Instead, manipulations such as artificial insemination or pair choices are enough to manipulate the genome as a whole and consequently the resulting off- spring.

Genophone provides analogous macroevolutionary manipulations to those employed in organic breeding: parents can be selected by the user, particular traits can be encouraged and manipulated. In addition, via dataglove manipulations, it provides a local direct and interactive exploration that facilitates smaller changes when used as a performance tool.

This was achieved by evolving the particular parameter values that produce a desired sound along with a performance mapping scheme, where a subset of those parameters is mapped onto manipulation devices dataglove and keyboard controls for use in performance see Fig.

The option of locking individual genes, or even whole sections of the genotype, provides an added layer of control over the evolutionary process that helps bridge the gap between a totally free-form search and the tight regulation offered by a manual sound editor.

The inspiration for parameter locking came from the way genes are activated and deactivated in biological genomes, producing epigenetic evolutionary effects Singh and Krimbas An important difference between the way EAs are generally used in constrained searches towards fixed sound targets, on the one hand, and unconstrained explo- ration of sound spaces, on the other, is the choice of initial population.

In con- strained EAs, a population of random individuals is often used to jump-start the evolutionary process. This is partly to ensure no initial bias exists, which may di- rect the search away from the global maximum — the perfect match to the target. In the unconstrained exploratory case, this is not necessary; in fact experiments with Genophone have shown that it is not even desirable.

Exploration of sound and performance mapping spaces with Genophone after Mulder A large amount of knowledge is embedded in the parametric definitions of these sounds, information that ultimately encodes a set of aesthetic values, albeit in an implicit and not easily decipherable way.

By using such hand-designed sounds as points of departure for the evolutionary search, this embedded knowledge can be exploited. Experiments with Genophone also revealed that starting from hand-designed origins does not necessarily mean that the resulting offspring would sound very much like their parents. In fact sometimes they can sound surprisingly dissimilar, yet somehow still retain some of the original quality of the hand-designed parents.

The first one is in essence a convergent process, whereas the second one is diver- gent. In particular, it will be argued that adaptive systems can provide a rich interactive mechanism for performing as well as composing with the computer.

Musicians have always made use of, and arguably inspired, new technologies. The computer opens up an unimaginable scope for developing new sounds, new aesthetics and new composition and performance practices. The challenge, of course, is to make something that anyone actually wants to listen to. Eduardo R. Although lacking the hallmarks of any particular catalogued musical tradition, the organisational structures of the dy- namics of some evolutionary and adaptive systems bear strong similarities with the morphologies and structures that appear across all musical styles.

The behaviours of some models have an inherent liveliness that has been shown to effectively mimic certain musical phenomenon, and exhibit complex structural dynamics that have been shown to be musically effective at all levels, from timbral morphologies to long-term structure at the level of musical form. In addition, the responsive nature of some adaptive systems offers an appealing mechanism for interactive performances allowing us to integrate the aesthetically challenging possibilities of computer music within the traditions of human performance practice.

Generating Structure in Time Superficially, an evolving population of digital genes may seem to have little in common with our concept of musical form.

But this model of artificial evolution shares with music a very fundamental characteristic: it is a temporal process. That it exists in time is one of the few uncontroversially universal features of music, yet consideration of dynamic form is rarely a primary consideration in computer-assisted composition.

A common problem reported by practitioners of computer-based approaches such as rule-based systems and neural networks as well as some evolutionary systems is that despite successfully creating specific elements, there is a lack of overall musical energy or flow. But they are not exciting.

This makes the creation of long term or hierarchical structure a real difficulty. An Introduction to Evolutionary Computing for Musicians 21 principles from which many models are derived for a discussion of the temporal paradox in musicology see Cook There are myriad time-based models that could be used for generating music, and many composers have explored their possibilities. The fact that a process is formally defined as a function of time does not in any way ensure that the musical outcome will be engaging, nor even that the temporal dynamics can be appreciated by the listener.

Just as the application of EAs demands careful formulation of representation schemes, fitness functions and operators, this approach relies on the inspired selection and implementation of a suitable model and the definition of a meaningful mapping from numerical output to musical space.

The implementation of a model is often motivated by an intuition that it shares an organisational structure with a particular musical phenomenon or effect. The musical success of the approach is then dependent upon mapping the numerical output into a suitable musical domain in a way that preserves the desired structure. In Chapter 8 Miranda describes various implementations of cellular automata CA models for musical applications. In one of these, Chaosynth, a chemical oscillator CA is used to parameterise a granular synthesis engine Miranda The dynamics of the chemical oscillator CA rule, as it evolves from a random state to sustained oscillation, bear strong resemblance to the morphological evolution of sound in the voice and many acoustic instruments: their partials converge from a random distribution to a stable pattern of oscillation.

The mappings used to parameterise the granular synthesis engine preserve these characteristics and so the sounds produced similarly bear these morphological features, capturing the global spectral evolution of an acoustic note onset. Using a complex dynamic model allows the description of the changes in amplitude of multiple frequencies over time as well as the relations between them. These multiple levels of related dynamic structures are not peculiar to the timbral level, indeed almost all polyphonic music can be conceived as a complex of distinct, but interdependent voices weaving spatio-temporal forms at many levels.

The use of complex dynamic systems enables the generation of these sorts of rich spatio-temporal structures seen at all levels of musical organisation. Besides modelling musical form, specific musical phenomenon can be modelled using time-based systems, which would be difficult or impossible to capture using other approaches.

Blackwell and Young , p. The model is based on three simple principles: separation, alignment and cohesion. Separation means each bird must steer to avoid bumping into each other or any other object in the environment. Alignment keeps each individual moving in a similar path by taking the average heading of local flockmates.

Cohesion keeps the flock together as each bird steers towards the average position of local flockmates. Blackwell has employed a similar algorithm to parameterise a granular synthesis engine, creating an eerily lifelike movement of sound swarming through time.

Integrating the Interactive Machine In the Boids algorithm outlined above, note that the future position of each agent is described in terms of the current state of the other agents: the agents are sensitive to, and respond to changes in their environment.

In no in-game music mode, the thread that handles music in the game never gets created, so that may help with slow graphics or people crashing from incapatibilities. Installing this without ever owning the orginal CD-ROM may constitute a violation of Hasbro's copyrights and should not be done. Wiki Content. Explore Wikis Community Central.