In this PhD, I propose that a useful approach to designing user interface for music composition software is to try and remove restrictions to how the composition can be represented. In particular, this PhD focuses on allowing users to customise the external representations of their composition, so that composers can evolve how they are working over time as the nature of the work changes.

I have been studying pro-amateur composers, working in any genre, using a "digital audio workstation" like Cubase, Garageband, Logic or Pro Tools. This software allows a musician to arrange audio recordings and MIDI notes along a timeline in order to compose a piece of music, and provides the facilities of a recording studio like effects and a mixing facilities. While composers often work collaboratively, this PhD looks at them in isolation.

Studies of composers have found they make considerable use of their environment, and that composing a work frequently involves lots of paper with custom notations on it, arranged spatially in their work environment. Some authors have suggested that switching to software reduces the amount composers can do this, reducing their creativity.

My PhD investigates how supporting evolution of representations in music software affect how composers work.

What have I done so far?

Initially, I carried out a literature review to find out what features for customisation there are in existing music software systems or which have been proposed by other authors. These suggest that relatively unexplored areas are:

• changing the preciseness which composers can enter their composition into the computer
• switching the metaphors composers use to view and edit the composition
• supporting evolving the semantics of what the composers enter over time

I also reviewed the existing literature of composers working with software, identifying what user groups had been studied, what methodologies had been used, and how the studies varied in quality and settings.

In my first study, I've studied a number of pro-amateur composers by watching them composing using a variety of music software packages. The composers were asked to work on a project they would normally be working on at that time, using their normal tools, in the environment they would normally work in. The study methodology was based around interaction analysis and involved collecting data about the setting the work took place in, objects the composers used and created as part of their work, and the composer's different activities over time. Some patterns stood out about these composers:

• The composers appear to do significant work away from the DAW, which has a different and complementary character to that done at the DAW. This suggests that some composers may require a second type of interface with a very different set of qualities, designed for use at different times to those dedicated to "productive work".

• The composers were observed trying to minimise the time they spent exploring the space of possibilities available. This suggests that these composers would benefit from software that supported their strategies for focusing their time on the areas of interest to them: using habitual strategies and non-musical conceptual frameworks to restrict their work on aspects of the work that are less important to them, and selecting and focusing on areas that are important.

• The composers made significant use of instruments and tools in order to work in less abstract and more directly experienced ways. This allowed the composers to listen to incomplete forms of the composition, even when the representations of it did not currently completely specify how it should sound. This suggests it would be helpful for DAW software to provide functionality for playing incompletely specified compositions - or at least to support coordinated use of musical instruments with the software to do this.

In parallel with this study, I have also worked on some prototype implementation of features that would be useful for building prototypes to test with composers.

• A framework for communicating between Ableton Live and a web page to allow real time control of a commercial digital audio station by custom JavaScript interfaces.

• A prototype digital audio workstation using the Web Audio API.

• A framework to coordinate interacting with a branching version history of a composition and allowing a composer to edit this across multiple synchronised web pages open on multiple devices.