Read Me

This section contains most of my algorithmic music. In addition to being algorithmic in composition they are all based upon a Harmonic structure. Four of the works exhibit microtonal aspects: Humanesque, Planet Earth, The Cross and Dance with the 21st Century Revolution. All of the works are microtemporal.

From Computer Music Journal, Spring 1995.

From the first issue of Computer Music Journal I have taken great delight in and still at many times have been baffled by the 'state of the art'. In this relatively short time the variety of computer music systems and the ingenuity of those both designing and using these systems coupled with the business savvy of manufacturers both great and small has exploded into a new market for computer music products. I am very thankful to all of these people for the tools that are currently available.

I have not strayed far from my humble beginnings with computers in 1974 as a composer studying BASIC at the University of Cologne in Germany. Although I did not acquire my first computer until 1984 I still remember nine years earlier purchasing a Texas Instruments Scientific Calculator with memory for three numbers, also a random generator, and applying it with musical proportion algorithms to write a trio for piano, flute and cello. It was a time for changing my musical awareness. Those were my days with paper computer, monochord, compiling mounds of data with colored pencils and filling in the score - by hand!

I started to develop my musical ear to ratios. Harmonic proportions examining isolated musics were used in tuning the ear to poetry, geology, astronomy, painting & color and weights & measures. The one thing that I discovered is this. Whistling a friends phone number is much more satisfying than whistling numbers out of the phone book. Here the musicality of the algorithm depends upon a model for perception namely, the numbers to the tones and the representational kinship with the resulting melos. On this we can all agree.

There are times when writing programs to generate music that I get up and start moving my arms and body mimicking rhythms which I perceive relative to harmonic time ordered variables. The excitement builds as the computer first uses the program and I get to hear dimensional harmonic transpositions of this same material. I just want to emphasize the musical human emotive side in algorithmic composition to the extent that the abstraction is only as incoherent as the composers understanding and organization of the musical material.

When composers sit down to write they may take a tuning fork to ear or when hard pressed sit at a piano for anchoring. A similar perceptual anchoring should be the basis of every good algorithm and more so musically structured for the end user. This puts a demand on the programmer to create algorithms that can be perceived musically. Of course there are situations in which one may randomly punch-up variables to wait and see what happens. This usage has tendencies toward a cut and paste improvisational music. Trying to justify the means and the end is difficult especially with today’s existentialism, but in practice it can be assumed that algorithmic composition combines the best of both improvisational and perceived musics. The learning curve tends toward perception because a composer gains in musical strength with each new algorithmic composition by becoming better at answering the question, what happens if? 

In my work with ratios a flow chart for the algorithmic composition process goes something like this: 1. Object-idea-inspiration. This first step examines extrinsic elements and concepts usually by making a drawing for a model, charts harmonic space as the model or uses music inspiration as a model. 2. Adoption of a standard of measurement with which to examine the model including a choice of time and frequency set at (1/1) or other ratios applicable to the model and compiling the actual data which represents the model. 3. Listening to the initial representation of the model projected onto the chosen harmonic space at the piano or by representation. 4. Selection of variables to represent and harmonic tables with which to manipulate the model. 5. Designing the overall algorithm for harmonic distribution throughout the selected time. 6. Running the program, debugging and the generation of MIDI data (timestamps and midibytes). 7. Transfer to a sequencer where further listening evokes changes and preparation of tracks for scoring. 8. Transfer to a score program, making appropriate changes and score print out.

For the past 16 years on a scale from paper & calculator to a full computer implementation this has been my compositional process. Harmonics (the science of ratios) fits neatly into the prescribed musical practice of division. The divisions of string length & frequency and time & space with other reciprocals make it easy to jump between astronomy and poetics while maintaining a firm grasp on time/tonal, capable of being perceived, aural representations. This process is musically successful at automatic generation of music under algorithmic control. Again the learning curve is proportionate to the frequency of usage, within a persons sentient musical capability, in relationship to the algorithmic abstraction which covers the gamut between improvised versus written music and seemingly tying the past to the present sensibility.

Two years ago I decided to write mainly for small acoustic ensembles with accompanying taped or live computer parts. Up to this time most of my work was exclusively for tape and there was really no need for a score. Most of the work consisted of graphics, harmonic charting, programs in BASIC to accomplish the rather large task of number crunching and neat stacks of computer printout for the musical parameters. The printout would then be used to enter the music data by hand at synthesizers and other electronic equipment using a click track as a time guide.

On my first attempt at generating the score material for a string quartet and MIDI part, I ended up with over 100 pages of printout detailing note, start & end times and metric locations for score placement. After scoring 5 pages I realized that the task would take months and I definitely was not enjoying the process. Even if the scoring could be completed there would still loom the prospect of having to enter the music data at the synthesizers to make a simulation.

I realized that there must be a way to siphon the music data into a MIDI data stream and thus automate the entire compositional process. A year later a modified version of the original program for the string quartet was up and running in MS Basic with MidiBasic version3 from Altech Systems (Jim Chandler, helpful people out there!) firmly attached in the resource fork. The wait was worth it!

I have noted that the amount of time it takes to compose and realize a work has decreased by a factor of six. I find that I am spending more time making aesthetic decisions by previewing the music and readjusting the program. A new set of problems have arisen which are directly related to understanding the limitations of available hardware and the programs used for sequencing and scoring, but adjustments can be made to work within these limitations. eg. My commercially available sequencer is limited to midibyte 111.

As a composer on the home front I am ecstatic for the ability to automate my compositional techniques. I am convinced that current research leading to the future development of new products will certainly aide growing numbers of composers in the development and understanding of musically perceived algorithmic processes. The innovative work of soft- and hardware designers has allotted more time for what it is really all about, which is making music.