About Level3
Search archives
- June 2007
- August 2006
- May 2005
- June 2004
- November 2003
DIT Home

Read postings about this article   |   Post a comment about this article  |  print this article [pdf]

The DiTME Project

Interdisciplinary research in music technology


Author - Eugene Coyle, Dan Barry, Mikel Gainza, David Dorran, Charlie Pritchard, John Feeley and Derry Fitzgerald

[<<previous][ next>>]

Show/ hide article menu (click icons opposite)

1.1 Technological Strand 3 research application

Following an application to the Department of Education Technological Research Sector Strand 3 scheme in April 2001, the emerging audio group at DIT was successful in its application for an interdisciplinary project titled Digital Tools for Music Education (DiTME). The project proposed an integrated array of research objectives in music technology, with development of a toolkit to run on a standard multimedia PC, and a with a number of novel features which would be of benefit to both teachers and students of musicianship at all levels. These included

  • a slow-down/speed-up facility which would not affect the pitch of the recorded music
  • an instrument separation facility to ‘comb out’ a lead instrument from a piece of recorded music
  • a music transcription facility to convert recorded music into music notation.

            It is often beneficial for students to play along with an accompaniment whilst practising. A live accompaniment is not always available and a recording may be used instead. However, this accompaniment will have been recorded at a certain fixed tempo. Time-scale modification algorithms may be used to enable independent control of the playback rate (without change of key) to suit a student’s current learning cycle. The desirability of such a facility for music teaching and learning had been ratified by a number of music teaching professionals in the conservatory of music at DIT.

            The task of extracting individual sound sources from a number of recorded mixtures of those sound sources is often referred to as sound source separation. Audio source separation is a complex problem, however significant benefits and possibilities present if an audio mixture can be separated into signals that are perceptually close to the original before mixing. For example in the study of musicianship, from the most elementary stages through to virtuoso performance, the service of a competent accompanist during practice is highly desirable though not always feasible. Further, much music is scored for orchestral accompaniment but few aspiring instrumental or vocal musicians have the regular opportunity to rehearse with a professional orchestra. Music Minus One (MMO; see http://www.musicminusone.com) recognized this dilemma over 50 years ago and has recorded a library of over 400 CDs containing the most requested accompaniments (orchestral as well as piano) for a wide range of music including classical, jazz, rock ‘n’ roll and country and western. However, the accompaniments are recorded by professional orchestras and accompanists playing with virtuoso soloists, so the trainee musician needs to have reached a very advanced level in order to use an MMO accompaniment. If the lead instrument (or voice) could be ‘combed out’ of any ensemble recording then any audio CD could be transformed into an MMO format. Such a facility would be useful for both the trainee lead-part musician and the trainee accompanist.

            A third highly desirable feature of the proposed music teaching and learning ‘toolkit’ suggested by the DIT target users is a music transcription facility. Music transcription refers to the process of converting recorded music into music notation. Existing automatic transcription systems are limited to simple monophonic (one note at a time) music. For polyphonic (more than one note at a time) the only reliable means of transcription is a very tedious manual process involving repeatedly listening to short segments of the music and comparing them to known tones. For fast music such as Irish traditional music this is often impossible. If such music can be slowed down and the lead instrument separated from the ensemble recording, then this will help to develop an automatic transcription algorithm.

[<<previous][ next>>]