Soundtracking the different dissertation sections

At my last meeting with Sian I proposed that that I would create a different piece of music to represent and accompany different sections of my dissertation. Sian seemed enthusiastic about this idea and felt that I would be capable of this. I'm not so sure about the capable, but am keen to have a go nevertheless. I'm going to write some notes here then on how this could work.

I have to be realistic about what's achievable, both in terms of time and also my own ability. Perhaps that means that I shouldn't try to create an original track for every single part of the dissertation? Maybe I could create some form of ambient music to accompany the more functional parts of the dissertation, for instance the front cover, contents, appendix, bibliography. 

Taking this approach would free up more time to create music that is specifically representative of the different areas of discussion. I don't exactly know what this means or whether it would be feasible. For instance, what type of music could be seen as representative of a literature review? Or maybe I can channel the spirit of Brian Eno and think about composing Music for Methodology.

I'm assuming here that music needs to go beyond simply being pleasant accompanying sound? Or is it justifiable to include an audio element solely on the basis that it is intended to create an aural space that the listener might find conducive to reading or interpreting the words on the screen? I expect that I will need to offer a more convincing rationale for the inclusion of sound  in this way - and within that, how I decided that a particular sound would be conducive to reading?

A way around this might be to use music that is itself the product of what is written in the text. Is there a way of feeding the text from a particular section into a piece of software that in turn will create music? I could always edit it afterwards without losing its original form. 

In fact, does each piece of music have to represent what's contained in the text of for that part of the dissertation? Could I instead just create a number of pieces that are representative of the wider dissertation, rather than specific sections? This would mean that I could focus on the general subject matter - books, screen, colour, visual image, film learning - rather than the content of particular sections. If I did take this approach, maybe I could instead use the artefacts objects as a way of creating sound? For instance, perhaps there's a way of using one of the Brian Eno apps to create shapes that are representative of objects (e.g. a book) that in turn create music. In this way, the visual representation of the book creates its own sound.

If I create a particular sound for the different artefacts, these could then form the basis of different soundtracks. Taking the Brian Eno approach again, they would be used repeatedly in different sequences so that no two tracks were the same.

To any of the tracks I could add additional layers of sound that are in some way representative of the subject, for instance it could be the sounds of someone flicking through a book, a film reel or effects that fit with the constellation metaphor.