Flexatone HFP
Icon
  • System Detail

System: MP1

Authors

Description

The computer-assisted composition program MP1 is based on the idea that the computer is a composer's collaborator. While the composer controls the overall outlook of the music and its abstract structure, the computer provides the details of the piece and ensures a certain degree of randomness unencumbered by cultural conditionings. The composer sets in motion a well-defined process by supplying a set of rules as well as the initial conditions and then does not interfere with the process or its end result. The composer accepts the output as long as it is consistent with the logic of the program and the input data. As the particular details of a piece depend on random occurrences, the same code and data can produce an unlimited number of compositions belonging to the same "equivalence class" or manifold composition. The members of a manifold composition are variants of the same piece; they share the same structure and are the result of the same process, but differ in the way specific events are arranged in time. A manifold composition is somewhat similar to the serigraphs produced by a visual artist, except that its individual members may be more distinct from one another. The global effect depends on the coherence with which the control parameters are chosen. For example, the members of a manifold composition may share certain fixed sections, while other sections diverge in different degrees. At present, MP1 running on an IBM RS/6000 can generate a chamber music work in a little less time than its actual duration: a 12- minute piece takes about 10 minutes of CPU time. MP1, written by Sever Tipei, was first used in 1973. Although its underlying structure remained the same over the years, MP1 has been evolving continuously, a frame-work in progress to which new features are added. Based on the proposition that the sounds of a composition can be described as entities in a multidimenssional vector space, the program has three interlocked layers: stochastic distributions, sieves or logical filters, and Markov chains. Any number of two or more sound parameters (sound characteristics) can be defined for an arbitrary number of parts (streams of events); by default, the first parameter is assumed to be time. For each parameter, equivalence relations mod m of the form: s = km + n are described where n is the equivalence class and m is the size of the equivalence class. This way, control over details is exerted through n while the broader aspects (such as octave or register in the case of pitch) are controlled through k. The main loop of the program assigns values for each sound parameter by matching the probability of various elements on a list of possibilities with a random number. A choice is completed when all such values are assigned for an event. MP1 is a comprehensive program in the sense that once the computations start, the user can not intervene until the piece is finished and then, either accepts or rejects the result. This way, the composition is treated as a process which, once triggered, is capable of continuing on its own, like a natural phenomenon. In the author's mind, this approach 1. elevates the act of composing with computers to a higher or meta-level, 2. insures its integrity of the process, and 3. uses the computer as a collaborator in stead of a dumb slave. Stochastic layer: A list of possibilities is created starting with preferred (mean) values for both n and k and their deviations (variances) which are defined by the user. The range of the instrument or sound source for that particular parameter is also taken into account along with preferred intervals between adjacent values and their deviations. Next, coherence factors between successive values and/or intervals both at the same part and between simultaneous parts are considered. Eventually, each element on the list is assigned a probability (along a normal-like distribution curve) by comparing it with the preferences specified by the user. Sieves: Logical expressions containing modulo m terms, sieves are used to extract subsets from a continuum. An example in the pitch domain would be the selection of a certain scale out of the total chromatic. Besides simply retaining or discarding elements, sieves can, at the same time, assign weights to the elements allowed to pass through. A powerful tool, sieves are also used in MP1 to correlate various sound parameters and establish an equivalent to orchestration constraints: values at one parameter which may not coexist with certain values at another one are rejected. For example, a ten second, very soft sound can not occur in the high register of the piano. Markov chains: MP1 can handle patterns or strings of elements defined at one or more parameters. They are treated as Markov chains and they refer to intervals between consecutive values (e.g. durations, not absolute timings; pitch intervals, not absolute pitches). Sequences of arbitrary length can be constructed if, in stead of considering the probability of an interval following another, the probability of an interval to follow a certain number of others which have already occurred in a certain order is considered. When such patterns are defined simultaneously for most or all parameters, motives or themes in the traditional sense could result. The 100% option: In MP1 the user has the possibility to specify the accuracy with which patterns are reproduced. When 100% accuracy is desired, all other options are short-circuited and the pattern is copied without change. However, this operation might not be possible due to range restrictions and/or to the sieves in effect at that moment. In order to avoid such situations, MP1 finds out before starting to copy the pattern if its exact reproduction is possible. All transpositions of the pattern are tried and if none of them is successful, the option is abandoned. Chords: Vertical aggregates can be created in MP1 in two ways: 1. by having an arbitrary number of parts contribute one sound each, similar to a contrapuntal texture where chords are the sum of simultaneous independent lines; 2. by declaring them as one of the sound parameters , giving values to various chord-sizes, and choosing particular chords form a library. The second option was used in Aga Matter, for piano and tape (see reference below). Entering data: All information contributing to a choice is controlled by the user and it can vary during the piece. This results in large quantities of data if care for detail is desired. Two auxiliary programs, PREAUX and AUXIL allow the user to enter and/or edit the data used by MP1.

References

“Composers are now able, as never before, to satisfy the dictates of that inner ear of the imagination. They are also lucky so far in not being hampered by esthetic codification -- at least not yet! But I am afraid it will not be long before some musical mortician begins embalming electronic music in rules.”

(Full citation)

“The use of computers is the logical outcome of a historical development. It by no means heralds a new musical epoch; it simply offers a fast, reliable and versatile means of solving problems that already demanded solution. The person who writes the computer programme must bear the development of musical language up to the present in mind, and try to advance a stage further.”

(Full citation)

“... and the hope of an extraordinary aesthetic success based on extraordinary technology is a cruel deceit.”

(Full citation)

“The computing machine is a marvelous invention and seems almost superhuman. But in reality it is as limited as the mind of the individual who feeds it material. Like the computer, the machines we use for making music can only give back what we put into them.”

(Full citation)

“... the individual and the society are deprived of the formidable power of free imagination that musical composition offers them. We are able to tear down this iron curtain, thanks to the technology of computers...”

(Full citation)

“Music is then no longer primarily conceived as a guide for premeditated emotions, but as the density of the possible relationships which first become actuality during production under the influence of chance, and which during performance are presented to the listener as sounds beyond any environmental associatiations, independent of bodily actions required to produce sounds...”

(Full citation)

“The characteristics of every sound depend on the way in which the sound was produced. Each art-form exploits its special production methods in order to endow the phenomena with unmistakable characteristics. Artistic economy demands that the means be appropriate to the end, and that the exploitation of the means be an end in itself.”

(Full citation)

“With the development of electronic and computer music, multidemnsionality of sound representation turned out to be both natural and useful. But music goes beyond multidimensionality -- it is even more complex.”

(Full citation)

“... the use of numerical machines no longer stands in need of justification. It is not a mystery. If there is a mystery, it is in the mental structures of music and not in the computers, which are only tools, extensions of the hand and the slide rule.”

(Full citation)

“... but beware, technique can submerge the user: We must defend ourselves; it is good to use techniques, but we have to dominate them, to stay alert.”

(Full citation)

“The danger is great of letting oneself be trapped by the tools and of becoming stuck in the sands of technology that has come like an intruder into the relatively calm waters of the thought in instrumental music.”

(Full citation)