Document Type

Theses, Ph.D

Rights

This item is available under a Creative Commons License for non-commercial use only

Abstract

Transcribing Irish traditional music is an open-field of research. The oral transmission of the music between generations explains the lack of transcription until recent times. The music can be played solo, which permits the player to exploit the variety of ornamentation types, in unison, and also with the accompaniment of a harmonic instrument. Different signal processing applications for transcribing Irish traditional music are presented in this thesis, including onset, ornamentation and pitch detection. An onset detection system which focuses on the characteristics of the tin whistle within Irish traditional music is first presented. The tin whistle is a good example of the features of Irish traditional music, and the detection of its onset encounters all the problems associated with onset detection identified in the literature review. An extension of this method is also implemented in an effort to detect the most common types of ornamentation, which has not been attempted to date. Existing onset detectors utilise energy and/or phase information to detect onsets. A novel onset detector, which focuses on the harmonicity of the signal to detect the onsets by using comb filters, is presented. This methods overcomes the difficulties encountered by existing onset detection approaches in respect of signal modulations and detection of slow onsets. Finally, a further comb filter based method is utilised to detect the triads played by a harmonic accompaniment. A set of results is presented for the four methods, followed by a commentary and explanation of the novel contributions.

Included in

Engineering Commons

Share

COinS