Exploring Music Spaces – Computer Science Research Journal


A few years ago when I was introduced to Borges I was moved by the idea of a complete library. His short story, Library of Babel, most importantly reminded me of the difference between the infinite and the very very large. It also forces one to understand what it means to explore a space of possibility.

A natural step to take is to apply this idea to other domains – such as Daniel Dennett’s Library of Mendel where all DNA information exists. If this idea is applied musically to the 12 tone scale, we arrive at a collection of all possible melodies, chords, harmonies…etc.

This metaphor is tempting, especially to Computer Scientists who analyze music or sound, since the idea of exploring [vector] spaces is very common. An Artificial Neural Network may partition a space into categories (for example it is possible to learn timber and other acoustic qualities quite easily this way), a Support Vector Machine maps a space into higher dimension as to make it linearly separable (for example to distinguish human voices), a Kohonen map performs multidimensional scaling for low-dimensional visualization. A Genetic Algorithm will explore a space using a random mutation process, a cellular automaton will blindly traverse a space based on initial rules (check out swarm music), while a Markov chain will bounce around predictably inside a space. This sort of space exploration is everywhere is modern computer music research literature. So, to step back and return to Library Of Babel it can be refreshing as a theoretical experiment to ask ourselves some questions.

Is there a fundamental difference between the expressive qualities of music and literature? Thinking specifically at both the structural rules (Grammer vs. Scales) and more abstractly in terms of human interpretation of these forms.

I find it interesting that an algorithm (check out Brian Eno’s work during the past 10 years – he uses a process called generative music) can generate expressive, coherent and meaningful music. While the task of generating expressive, coherent and meaningful literature is much more difficult. If you are in the mood to ponder the consequences for the conception of automated literature then you can read about it here..At the very least it’s important to imagine the creative process as the exploration of a possibility space (or with pop music, a predictable recombination of past discoveries, in a different rhythmic space…and a new outfit).

To avoid a long theoretical path, I will return to a related research problem I have read about in an article titled ‘A Flexible Music Composition Engine’ – this work is being done at UWO by Maia Hoeberechts, Ryan Demopoulos and Mike Katchabaw. They write:

Imagine an “emotional equalizer” which would allow a listener to alter the music’s mood as it is playing…it would operate exactly like a normal stereo equalizer, except that the sliders would be marked with emotional descriptors rather than frequencies

If you think about this dynamic in terms of the ‘music space’ of a 12-tone Library, then one could try and imagine coloring this music space according to these emotional states. So, adjusting these sliders would effectively push us along emotional gradients of this space. All of this sounds less crazy when we are working in the highly restricted musical domains that are commonly used (rather then the space of all possible music). Nevertheless, how is one to define an emotional space without restricting to elementary chord theory (minor=sad major=happy) or some slightly more complex version of this?

This I will look into next.

One Response to “Exploring Music Spaces – Computer Science Research Journal”

  1. britcruise Says:

    Thanks for your interest, glad someone see’s the humor in this.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: