In October 2011 I carried out my second year review for my PhD. This involved submitting a report on what I've done so far and my plans for the next year and giving a presentation on it.
Report
The report presents a review of the programme of work of my PhD as it stands now. It is in three distinct sections.
Chapter 2 is a current draft of a journal paper that I'm working on with my supervisor. It presents much of what we've done to date, including experimental results using the jazz grammar with the parser.
Chapter 3 is a document on the tonal space semantics associated with the grammar. The details of this were skated over in the paper (chapter 2), but are presented in full here. We plan to use this chapter as an appendix to the paper.
Chapter 4 concerns my current work on extending the models to the harder problem of parsing harmony directly from MIDI. See the presentation (below) for a motivation of this task.
Presentation
Abstract:
An important part of our perception of Western music is our ability to interpret harmony and its structure. Among other things, this underlies our perception of the similarity of two pieces of music. Tasks in the field of music information retrieval rely on metrics of musical similarity and require a way of analyzing the structure of harmony.
We have used a musical adaptation of combinatory categorial grammar and associated parsing techniques to perform this analysis automatically. Up to now, we have only considered the problem of parsing chord sequences - the abstract representation of harmony used by musicians when improvising. We now turn our attention to the harder problem of parsing streams of musical notes (in the form of MIDI files).
In the presentation, I omitted much of the subject matter of chapters 2 and 3, focusing instead on chapter 4 - the basis of our ideas for future work. I motivated whole problem of harmonic parsing and described very briefly what we've done up to now. I then presented a related model and my ideas for adapting some aspects of this model to our task.
This was a talk, not a film: it is therefore meaningless to read the slides on their own. I have written an article consisting of the slides and a prose form of my presentation. In case you wish to refer to the slides themselves, you can download them here as well.