Generative Creativity - lecture 9;
Soundscapes and sonification

Introduction

While generative methods are often used to produce note-based music, they are also used to produce more general forms of sonic experience, i.e,. `soundscapes'.

This can take the form of

Brian Eno on generative music

Key figure for ambient/generative music and creator of the seminal `Generative Music 1' album from 1996.

Views generative music as `the blending of several independent musical tracks, of varying sounds, length, and in some cases, silence. When each individual track concludes, it starts again mixing with the other tracks allowing the listener to hear an almost infinite combination'

Thursday Afternoon: http://www.youtube.com/watch?v=swQAwnusJVQ

Eno chats about generative music

Sonification of `Metamorphosis'

Mapping from text to soundscapes using low-level parameters (e.g., words and phrases) and higher-level constructs (e.g., plot developments) and including some of the original text.

Sonification of automata

Alice Eldridge (sussex DPhil student and BLIP contributor) has also experimented with sonification of neural nets and automata.

In an early work, she used a simulation of a cellular automaton combined with a homeostatic network.

Outputs from the homeostat were used to define pitch changes; the CA states are used to define when notes play.

For some pieces, she also used a 'melody' line, in which the timings were determined by a stochastic method

Output

See latest samples on http://www.ecila.org

Unstable homeostat

Simple wave

Homeostat controlling amplitudes of 10 sine tones in AudioMulch

Neural oscillators

In this work, Eldridge used a network of two artificial neurons arranged in mutual inhibition to produce oscillation.

For a range of parameter settings, the oscillation echoes the frequency of an input signal, enabling beat induction.

Output of 2 nodes with similar natural frequencies out of phase

Changing weights between 2 nodes

Sonification of prime numbers

Work by Daniel Cummerow from 1998 involves algorithm-mediated sonification of mathematical constructs, e.g., the prime numbers.

Begins by converting the sequence into base 3 notation.

The first 10 primes are then 2 10 12 21 102 111 122 201 212 1002.

Generation of bass part

The 3rd digits of the first 941 prime 9 (zero-padded) numbers were used.

The note values were determined by the number of consecutive prime numbers with the same 3rd digits.

This value defines both the number of tied sixteenth notes and number of scale steps relative to the previous tone.

0 is a pause, 1 means descend and 2 means ascend.

The initial pitch is F so the first tone is G#.

Generation of treble part

This was determined by the unit digits of the prime numbers.

For the 1st - 428th prime numbers the tones are determined the same way as the 1st part. It then pauses while the 3rd part executes.

When the 3rd part is done, the 2nd part does the same using the 569th - 1002nd prime numbers in the following way:

Start at F, 1 means descend a scale step, 2 means ascend a scale step (there are no zeros). When done, double the 3rd part to the end.

Combination procedure

For the 1st - 428th prime numbers, this part doubles the 2nd.

Then between the 429th - 463rd prime numbers, it uses the whole prime numbers (they have 8 digits in this interval) concatenated.

Starting at F, 0 and 2 means ascend a scale step, 1 means descend a scale step.

When it's done, it hooks up at the prime number where the 1st part is at the monent and uses the same algorithm as the 2nd part had at the beginning of the piece.

When the 2nd part is done, the pitch jumps back to the default F, and then continues as before to the end.

Output

Three Parts in A Harmonic Minor by Daniel Cummerow

Aesthetic issue

Cummerow has tried to map tones out of a mathematical structure. But why?

The assumption must be that the quality of a piece of music depends on its structural properties.

So find something non-musical with interesting structural properties and map it into music and you should get interesting music?

Fractal music

Fractal music follows the same general approach as Cummerow's maths-based compositions.

But in this case, the mathematical structure is a fractal, a generatively specified object which is infinitely detailed.

This means you can `zoom in forever'.

These objects are useful for graphics and image applications (they're a great way to generate images of trees) and feature extensively in the area of algorithmic art.

Generating music from fractal objects

For generation of musical output from fractals, data from the fractal construct are converted to musical parameters to create melodies, harmonies, rhythms, textures, etc.

This is most easily done by representing the factal in some sort of visual form. Then, we might...

As usual there is the problem that while fractals clearly have interesting structural properties, these may not translate easily into interesting musical structures.

Randomly selected fractal music site

http://www.myspace.com/moontoad

Summary

Exercises

Reading


Page created on: Wed Feb 11 15:33:56 GMT 2009
Feedback to Chris Thornton