Home arrow All arrow Cullen, Charlie
Cullen, Charlie Profile Page

Thesis 1

Thesis Title: The Sonic Representation of Mathematical Data
Degree Type: PhD
Degree Specialism: Music Technology
Supervisor(s): Dr Eugene Coyle, Dr Noel Russell
Thesis Status: Accepted
Date Submitted / Accepted: 2006
Institution Submitting / Submitted To: Dublin Institute of Technology
Thesis Location / Link: http://arrow.dit.ie/appadoc/2
Abstract: Conveying data and information using non-speech audio is an ever growing field of research. Existing work has been performed investigating sonfication and its applications, and this research seeks to build upon these ideas while also suggesting new areas of potential. In this research, initial work focused on the sonification of DNA and RNA nucleotide base sequences for analysis. A case study was undertaken into the potential of rhythmic parsing of such data sequences, with test results indicating that a more effective method of representing data in a sonification was required. Sonification of complex data such as DNA and RNA was found to require more verbose methods than pitch to parameter mappings, and so investigation was made into musical pattern sonifcation. Existing low level pattern design methods were next evaluated in an experiment concerning the use of musical patterns to represent data. This experiment suggested that while a musical pattern may be made distinct, it does not necessarily follow that it is memorable. This experiment also suggested that concurrent pattern representation was difficult to process, and so improved methods were required. Improvements to pattern design were made with the introduction of contour icons, which allow detectable and memorable musical patterns to be designed using simple shapes. Testing showed contour icons to be significantly more effective than low level patterns in a sonfication, and as such form the basis of the novel contributions of this thesis. Improvements in concurrent representation were considered by the use of harmonic combination, a method of defining intersections in a data set as harmonies of a single common musical pattern. Significant improvements were observed over non-harmonic concurrent representation, although limitations were observed due to constraints in the number of combinations available using a specific value. Harmonic combination has potential for further development, and is a novel contribution of this thesis. The organisation and grouping of data in a sonfication using rhythmic parsing was also investigated. Rhythmic parsing uses rest notes within a musical framework to define sub-groupings in a data sonfication. Tests showed rhythmic parsing significantly improved the comparison of values and intersections between groups in a data sonification, and is another novel contribution of this thesis

Thesis 2

Thesis Title: ScoGen software development project
Degree Type: MA
Degree Specialism: Music Technology
Thesis Status: Accepted
Date Submitted / Accepted: 2000
Institution Submitting / Submitted To: Dublin Institute of Technology
No. of Volumes (no. of pages): 1(62pp.)
Thesis Location / Link: DIT Conservatory of Music & Drama Library
Abstract: The aim of this project is to construct a piece of software to automate the task of writing Csound Score files.
One of the main stumbling blocks to mass acceptance of the Csound synthesis software is its labor intensive nature-often coupled with an obscura of technical definitions which seek to confuse all bu the most sonically literate. The Csound compiler has yet to claim its rightful place within the hierarchy of modern synthesis due to the mass of data it requires to produce even the most simple tones.
To this end, several applications have already been constructed to circumvent the construction process-notably Winsound (by Barry Vercoe of MIT) that automates the traditional command line compiler of Csound with a Windows-based GUI, and a Visorc (by Dave Perry) that provides an object driven GUI for Orchestra (.orc) file design.
Both applications have proved of great benefit to those who would seek to compose using the Csound interface-Winsound allows for rapid generation of output sound files various audio formats at the click of a few buttons, while Visorc not only simplifies the construction of virtual orchestras, but also gives them a much needed visual and modular definition with aids in their design.
The aim of ScoGen software was to automate the third (and usually most laborious) element of Csound- the generation of svore (.sco) files. Atypical Csound score file can contain anywhere from 4 to 10 variables per sonic instance, and coupled with the intricacy of the data associated with these variables it can rapidly mushroom into a very large file producing often very limited output in relation.
Many ways already exist to circumvent this particular drudgery- from the omittance of non-essential parameters to the use of Copy/Paste functions in order to create greater output from less user input. However, both succeed only in hampering or even crippling the overall composition process, whereby many pieces descend into the temptation for repetition rather than become bogged down in the minutiae of often seemingly abstract data.
ScoGen seeks to provide an alternative means of automation, based around three aspects of .sco file generation which have become apparent to the author after many hours spent in Csound composition over the last few years.
a) Many Parameters in Csound .sco files are often static, e.g. time/duration values or noise ratios or such.
b) Many parameters in Csound make use of a small range of data, and as such the same values can be seen littered throughout a .sco file.
c) Further, even supposedly random values that are so often the beloved preserve of Csound composers are frequently defined by the parameters they control- e.g. a random pitch still has to fall within the bounds of defined pitch, and therefore is not truly a random value, merely a random instance of an existing defined data set.
To this end, any code which could achieve both automation and a shift in compositional emphasis would allow a far greater degree of control over Csound generated output than existing labour-intensive processes. If a GUI driven routine were made available whereby any and all parameters could be user-defined as required within the framework of a repetitive .sco generator without the need for endless typing and editing it is conceivable that the Csound compiler could finally be judged on its merit as a software synthesizer rather than avoided due to its compositional monotony.

Site by TF Design