Signal Analysis for the Discovery of Traditional Music Repertories

SIGMUS was a project coordinated by the MTG and funded by the Spanish Ministry of Economy and Competitiveness (Subprograma de Proyectos de Investigación Fundamental no Orientada, TIN2012-36650). SIGMUS started on February 1st, 2013 and finished on December 31st 2015.

The SIGMUS project focused on the development of information technologies in the field of music. We worked on two specific aspects: in the creation and annotation of specific musical repertoires and on the development of signal analysis techniques and machine learning approaches suitable for the particular repertoires treated.
Regarding the musical repertoires treated, we worked on Flamenco and on Arab Andalusian music. The research focused on developing signal processing methods and semantic analysis approaches for the melodic and rhythmic characterization, for computing musical similarity, and for interacting with music collections. The primary objective was to develop new computational tools that can help us discover and understand the specific characteristics of these musics, their similarities and differences, and offer solutions to broader technological challenges.
Regarding the results obtained we should highlight the creation of datasets relevant to the computational study of Flamenco and the Arab Andalusian music, most of the data available under open licenses, and emphasize the improving of a software library, Essentia, for analyzing recorded music, also available with an open license. It is also important to note the studies that have been made of Flamenco especially the melodic aspects of this music.