Dr. Bryan Duggan's Pages @ DIT

Bryan Duggan, Brendan O'Shea, Mikel Gainza, Padraig Cunningham, "The Annotation of Traditional Irish Dance Music using MATT2 and TANSEY", The 8th Annual Information Technology & Telecommunication Conference, Galway Mayo Institute of Technology, Galway, Ireland, October 2008. (2008) |pdf| |abstract|

Currents estimates put the cannon of traditional Irish dance tunes at least 7,000 compositions. Given this diversity, a common problem faced by musicians and ethnomusicologists is identifying tunes from recordings. This is evident even in the number of commercial recordings whose title is gan ainm (without name). This work attempts to solve this problem by developing a Content Based Music Information Retrieval (CBMIR) System adapted to the characteristics of traditional Irish music. A system is presented called MATT2 (Machine Annotation of Traditional Tunes) whose primary goal is to annotate recordings of traditional Irish dance music with useful meta-data including tune names. MATT2 incorporates a number of novel algorithms for transcription of traditional music and for adapting melodic similarity measures to the creativity and style present in the playing of traditional music. It incorporates a new algorithm for filtering ornamentation notes and accommodating "the long note" in traditional music called Ornamentation Filtering using Adaptive Histograms (OFAH). A new algorithm is presented called TANSEY (Turn ANnotation from SEts using SimilaritY profiles) that annotates sets of tunes played segue as is the custom in traditional Irish dance music. The work presented is validated in experiments using 130 real-world field recordings of traditional music from sessions, classes, concerts and commercial recordings. Test audio includes solo and ensemble playing on a variety of instruments recorded in real-world settings such as noisy public sessions. Results are reported using standard measure from the field of information retrieval (IR) including accuracy, error, precision and recall and the system is compared to alternative approaches for CBMIR common in the literature.
[hide]

B. Duggan, B. O'Shea, and P. Cunningham, "A System for Automatically Annotating Traditional Irish Music Field Recordings", The Sixth International Workshop on Content-Based Multimedia Indexing, Queen Mary University of London, UK, Jun. 2008. (2008) |pdf| |abstract|

This paper presents MATT2 (Machine Annotation of Traditional Tunes). MATT2 is a novel system which can automatically annotate field recordings of traditional Irish music with useful metadata such as tune name, key signature, time signature, composer and discography. MATT2 works by using a number of algorithms to automatically transcribe digital audio to be annotated to the ABC music notation language. It then compares these transcriptions against a corpus of 860 human made transcriptions in ABC using a variation of the edit distance algorithm. Results using MATT2 to annotate fifty recordings of flute and fiddle tunes demonstrate a high success rate at annotating recordings made by different musicians. Additionally, several of the recordings successfully annotated in testing MATT2 were recorded in imperfect conditions, with badly degraded audio.
[hide]

B. Duggan, B. O'Shea, Mikel Gainza and P. Cunningham, "Machine Annotation of Sets of Traditional Irish Dance Tunes", The Ninth International Conference on Music Information Retrieval (ISMIR), Drexel University, Philadelphia, USA, September 2008. (2008) |pdf| |abstract|

A set in traditional Irish music is a sequence of two or more dance tunes in the same time signature, where each tune is repeated an arbitrary number of times. A turn in a set represents the point at which either a tune repeats or a new tune is introduced. Tunes in sets are played in a segue (without a pause) and so detecting the turn is a significant challenge. This paper presents the MATS algorithm, a novel algorithm for identifying turns in sets of traditional Irish music. MATS works on digitised audio files of monophonic flute and tin-whistle music. Previous work on machine annotation of traditional music is summarised and experimental results validating the MATS algorithm are presented.
[hide]

Research

2009

2008

2007

2006

2005

2003