Borasky Research Journal

Download the AlgoCompSynth Virtual Appliance

Follow @AlgoCompSynth On Twitter!

Algorithmic Composition: Paradigms of Automated Music Generation

If you want to get started in algorithmic composition, this book is the best place to start. The author, Gerhard Niehaus, teaches algorithmic composition at the Institute of Electronic Music (IEM) at the University of Music and Dramatic Arts in Graz, Austria. His goal was to create a “detailed overview of prominent procedures of algorithmic composition in a pragmatic way,” and I think he has succeeded admirably. As you may know, algorithmic composition has a long history, featuring some names you might recognize – Kirnberger, Mozart and C.P.E. Bach. Niehaus recaps the history of algorithmic composition and algorithms in general in chapter 2. Following this introduction, Niehaus devotes a chapter to each of the major algorithmic composition paradigms:

• Markov Models
• Generative Grammars
• Transition Networks
• Chaos and Self-Similarity
• Genetic Algorithms
• Cellular Automata
• Artificial Neural Networks
• Artificial Intelligence

A final synopsis completes the book. My own compositions have mostly used Markov models, primarily as inspired by Xenakis’ Formalized Music. I don’t think I’m unique in staying with a single paradigm – most other algorithmic composers that I’m aware of have tended to focus on a single one of these paradigms, choosing to exploit other means to keep their music interesting. But I’ve certainly grown tired of the limitations of Markov models, and found several other techniques I can use in this book. I would have liked to see more on sonification and on “found music”, which I regard as algorithmic composition paradigms on an equal footing with the ones Niehaus covers. My personal favorite piece of my own was in fact “found music” – recordings taken from the sounds computers used to make as a by-product of� earning their keep. And I think fuzzy logic should have had more coverage as well – Peter Elsea has done quite a bit of research in this technique that deserves to be better known. Algorithmic composition certainly isn’t for every composer – it requires a disposition towards music theory that some composers can do without. But if you’re a musician / composer with a theoretical bent, I encourage you to try it, and this book is the best place to start.

Algorithmic Composition: Paradigms of Automated Music Generation by Gerhard Nierhaus

AlgoCompSynth Reading List

Download the AlgoCompSynth Virtual Appliance

Follow @AlgoCompSynth On Twitter!

The story behind “When Harry Met Iannis”

Iannis Xenakis passed away on February 4, 2001. When he arrived in Heaven, he sought out his mentor, Olivier Messiaen. Messiaen said, “Iannis, there’s someone here I want you to meet.” The two walked over to a small sidewalk cafe, and there sat Harry Partch. And so from this fanciful meeting in Heaven, “When Harry Met Iannis” was born. From Partch, the piece inherits the 43-tone just scale, tuned to G = 392 Hz. Also from Partch, the piece inherits the need to build instruments to play in the Partch scale. The harmonies in “When Harry Met Iannis” are derived from the Tonality Diamond. Now we have the harmonic structure – the Tonality Diamond. This is where Xenakis comes in. Xenakis was a pioneer in stochastic music – music composed algorithmically by using chance elements. The Perl script that composed “When Harry Met Iannis” was given an initial state – a chord on the Tonality Diamond – and a duration for the piece. All the other elements of the score are derived stochastically by a random walk. The duration for each note is chosen at random. Whether the note is sounded or silent is chosen at random. And the progression of chords around the Tonality Diamond is chosen at random. At each transition, one of four chord transformations is chosen at random:

1. The tonality can change from an Otonality to the corresponding Utonality or vice versa.
2. The chord structure can stay the same but be shifted up or down an octave.
3. A new Numerary Nexus can be chosen for the chord, with the rest of the structure remaining the same.
4. An Odentity or Udentity can be added or deleted, with the rest of the structure remaining the same.

Because some of these transitions can take the piece out of a practical performance range, if the piece goes out of range, the operation is reversed. For example, if the chosen transformation is to go up an octave and the resulting chord would be pitched too high, the next chord is instead chosen by going down one octave. If removing an Odentity or Udentity would leave an empty chord, one is added instead. Once the score has been computed, the piece is rendered. For this piece, I used a little-known digital synthesizer called sfront. Sfront is similar in principle to the much-better-known CSound digital synthesizer, but sfront code is vastly easier to read than CSound code. The synthesis consists of sung vowels with reverberation, and the position of the chord from left to right between the two stereo channels is chosen at random.

Iannis Xenakis

Formalized Music by Iannis Xenakis

Music and Architecture: Architectural Projects, Texts, and Realizations by Iannis Xenakis

Harry Partch

Genesis of a Music: An Account of a Creative Work, Its Roots and Its Fulfillments (Da Capo Paperback) by Harry Partch