Title: Research Student
Research Group: Centre for Digital Music
Supervisor: Simon Dixon
Research Topic: Linking music-related information and audio data
The research in this project is focussed on taking the meta-data associated with digital audio and presenting this in such a way that it is usable to consumers of music and providers of musical content. Working in collaboration with the OMRAS 2 project, this research will look into automatically extracting user-relevant information from digital audio and aligning this information with existing meta-data such as musical scores, lyrics and MIDI. Such meta-data can be useful to students of music, musical database searching and recommendation software.
As more people turn to internet based music content such as mp3's and streaming, there arises a need for users to filter through the increasingly vast on-line repertoire of music in a way that is meaningful to the end user. It is hoped that the work done in this project could lead towards applications in musical education, music searching and identification as well as tools for providing and handling music meta-data in on-line resources to meet the growing requirements of a growing digital music industry.
The following link is to an early draft version of the first chapters of my thesis. For more information on the work I am doing please look here (PhD Draft.pdf).
I presented a demo of my piano tutoring game "NoteScroller" at the 8th International Conference on New Interfaces for Musical Expression (NIME 08) from the 5th-7th June 2008 in Genova, Italy. Click here (NoteScoller.zip) for the application. The poster is available here (MacraeDixon08-nime-poster.pdf) and the paper accompaniment here (MacraeDixon08-nime.pdf).
I presented a demo of my audio-score alignment software "MidiMatch" at OMRAS 2 Workshop 1 on the 17th December 2008. This application uses Online Dynamic Warping to synchronise midi to audio in real-time. I hope to release this software shortly.
In order to evaluate audio and score synchronisation methods I am building a test data set and I hope to publish the non-copyright element (the reference files linking the audio and MIDI events) here as well.
Here are some sample Java and Matlab source codes:
MidiText.java: An example of how to interpret MIDI data in Java.
MidiRealTime.java: How to get the real time in ms of MIDI events in Java.
Simtrix.m: A Matlab script for producing a similarity matrix and optionally finding the DTW path through this (as in the image shown).
I am a teaching assistant at QMUL for following modules:
ELE201: Introductory Java Programming (2nd Semester)
ELE502: Digital Signal Processing (1st Semester)
ELEM020: Fundamentals of DSP (1st Semester)