Generative music

Music automatically generated by processes

• Wikipedia article
• Harley, J. (1995). Generative Processes in Algorithmic Composition: Chaos and Music. Leonardo 28:3, 221-24.

What is Generative music?

If we look at the dictionary term for Generative grammar and simply replace the word grammar with music, we get a great simple definition to help us understand a complex idea with many different aspects of particular focus. The dictionary term for generative grammar on the mac os x dictionary program is as follows

• A type of grammar (or music) that describes a language in terms of a set of logical rules formulated so as to be capable of generating the infinate number of possible sentances (or in our case musical phrases) of that language and providing them with the correct structural desciption.

This definition provides a great base point for understanding what we mean when we say generative music. It is important now to look at the rationale behind the generative music movement and what artists who explore this type of musical composition aim to achieve.

A Very Brief History

Charles Babbage.

Charles Babbage is best known for his experiments with his Analytical engine. His early model in the 1830's was the first machine that followed modern computer architecture in that it had a central processor where data was processed and delieved in a sequences that would achieve a solution. Babbage even thought of areas whereby these operations could be stored and reused and manipulated at will depending upon the results required. The daughter of Lord Byron, Lady Lovelace "foresaw that it might be possible for a machine to write music. She predicted that if it were possible to define "The fundamental relations of pitched sounds in the science of harmony and of musical composition...then...the engine might compose elaborate and scientific pieces of music of any degree of complexity or extent." One hundred years later, the first experiments using computers to write music were initiated"(Holtzman 1994, p.131-159).

Generative Music and Human Psychology

• In the 1950's computers were used mainly for diagnostics and data analysis.
• It was also a time where extensive research had gone into the understanding of the psychological processes involved in the creation of semantic meaning.
• Key texts were Charles Osgood's "The Measurment of Meaning", and Ithiel Sola Pool's "Trends in Content Analysis".

The reason why I mention these two texts is because it gives an understanding of acedemic thought at this time, which was that all semantic meaning could be analysed, in short the human generative processes when creating music could be logically mapped and analysed with the help of computers. The generative process was analysed in a hope that;

• A, we would gain a better understanding of how our thought processes create music and
• B, through this understanding implement computer aided musical composition whereby the computer can for example help a composer to complete sections he or she is stuck on. The way in which analysed data of musical compositions and their generative process is read by a computer is through the use of algorithms

Algorithms

As we have discussed computers were being used to give us a set of logical processes that were used to create a particular piece of music, or vice versa we could tell the computer a set of logical processes in order to create a musical work.

So what is an algorithm?

• Our trusty dictionary states: "A process or set of rules to be followed in calculations or other problem solving operations".

So essentially an algorithm expresses a set of logical rules that must be followed in order to reach a solution. In a musical sense for example, after chord A is completed, then move to chord B, then to chord C, then once c is reached move back to chord A. This of course is an extremely simple example as many variables are used to create a much more interesting progression, such as phrase length, dynamics, overall structure etc. As stated before by using computers to analyse a score we could create data that could create an algorithm to reproduce that piece through a series of logical operations understood by the computer. However this could even be taken a step further as cognitive models could be analysed and mapped. One such composer who looked extensively into this area is Lejaren Hiller who created cognitive models based on musical ideas such as the rules of counterpoint composition so that the computer would generate music on its own that would be in a counterpoint style.

Lejaren Hiller

• Lejaren Hiller is often regarded as one of the first to use computers in creating cognitive models for music composition by computers.
• Sets of mathematical processes and operations understood by these early computers were used to apply to musical composistion by Hiller. These are very complex and can be found and more closely understood in a book on him by James Matthew Bohn called "The music of American composer Lejaren Hiller".
• As stated earlier Hillers manipulation of computer operations allowed him to experiment with cognitive models in particular the rules of musical counterpoint.

Other Composition Models

• Interactive and Knowledge Based Systems

This model look at developing the computer-composer interation so that they both form (as Harley puts it) a 'feedback loop' meaning that the actions of both the computer and the composer is recognised by both parties so that a conversation can occur.

• Non-Linear Functions

All of the processes that have been mentioned so far has been either based on random or indeterminate functions, or composer specified determinate functions. Non-Linear uses the chaos theory the results achieved are apparently random however oreder and structure can be found as it is based on a mathematical process, so elements of form are apparent in this style. (Harly 1995, pg.222)

John Cage

Fontana Mix (1958)

• A program written by Karlheinz Essl in max/msp enables a live realisation of Cage's Fontana Mix. Essl converted the score into a set of algorithmic operations in order to create the piece.
• "The 'score' supplies a number of transient foils with dots, lines and graphs which have to be overlapped in random positions and placed over a page that contains 6 curved and intermingled lines. By measuring positions of crossing lines one will obtain 6 parameter values which are needed to determine a single sound event. The choice of parameters is free eg, panning, dynamics, oscilator frequency, phrase length. For a detailed description and piture of the score go to the following link,http://www.medienkunstnetz.de/works/fontana-mix/images/1/

References

Bohn, J, M. (1970) "The Music of American Composer Lejaren Hiller And an Examination of His Early Works Involving Technology" The Edwin Miller Press:New York.

Essl, K. (2003) "Fontana Mix Max Msp Program" adaptation for real time realisation by Karlheinz Essl, Original score by John Cage (1958).

Harley, J. (1995). Generative Processes in Algorithmic Composition: Chaos and Music. Leonardo 28:3, 221-24.

Holtzman S,R. (1995) "Digital Mantras: The languages of abstarct and visual worlds" The MIT Press:Cambridge.

Sloboda, J, A. (1988) "Generative Processes in Music" Oxford University Press:New York.

Roads, C. (1996) "The Computer Music Tutorial" The MIT Press:Massachusetts.