Applications of Neural Networks and Genetic Programming in Computer Music and Art
- Are scaling properties of music related to aesthetic response of human listeners?
We extended results from Voss and Clarke (1975) and Zipf (1949) to measure the scaling properties of music (e.g., note pitch, note duration, harmonic and melodic intervals, etc.). These metrics are used to train artificial neural networks (ANNs) to classify music according to composer (authorship attribution), genre (style identification), and pleasantness (pleasantness prediction). ANNs achieve high success rates (~90%). Here are some early results
- Can we develop music search engines based on aesthetic similarity?
We are conducting experiments with a music search engine prototype which utilizes scaling properties of music to model aesthetic proximity.
- Can scaling properties of music be utilized in computer-assisted music composition?
We are conducting experiments with evolutionary algorithms to generate variations of existing music with prescribed scaling properties.
- Can we measure the effect of music on human biosignals?
We are analyzing biosignals of human listeners to explore if we can measure their aesthetic reaction to music. Preliminary results are very promising.
- Can scaling properties of text be used to predict language?
Preliminary studies on hundreds of e-texts from six natural languages (English, Esperanto, French, German, Italian, and Spanish) indicate that, under certain conditions, we may be able to predict language of unknown texts only from their scaling properties.
- Zipf, G.K. (1949), Human Behavior and the Principle of Least Effort, Addison-Wesley.
- Voss, R.F., and Clarke, J. (1975), "1/f Noise in Music and Speech", Nature 258, pp. 317-318.
- Spehar, B., C.W.G. Clifford, B.R. Newell, and R.P. Taylor. (2003). "Universal Aesthetic of Fractals." Computers & Graphics, vol. 27, pp. 813-820.
The following students have contributed to this exploration (since 2001, in reverse chronological order)
, Thomas Zalonis, William Epperson, Michael Miller, Luca Pellicoro, David Lyle, John Emerson, Brian Murtagh, Jason Trinklein, Hector Mojica, Brian Muller, William Daugherty, Nicholas Johnson, Dallas Vaughan, Chris Wagner, Tarsem Purewal
, Charles McCormick, Valerie Sessions, Yuliya Schmidt.