On the Use of Automatic Onset Detection for the Analysis of Maracatu de Baque Solto

In this work we investigate the use of automatic onset detection techniques to support the ethnomusicological analysis of microtiming in Maracatu de
Baque Solto. In order to lay the foundation for a robust and meaningful analysis
of Maracatu in terms of its temporal and rhythmic structure, we require extremely precise annotations of note onset positions. However, the nature of Maracatu performance, with percussionists in very close proximity to one another and playing very loud and at high tempo, makes manual annotation extremely challenging. To this end, we orient our work towards minimising the number of corrections necessary by a human annotator, and thus we incorporate explicit knowledge of Maracatu into a computational approach for onset detection. In our evaluation, we explore the divergence between the use of an “off the shelf” state-of-the-art onset detection technique which has been trained to generalise across a wide variety of musical material, with a bespoke approach which specifically targets the instrumentation in Maracatu. We then explore the impact of this approach when visualising microtiming profiles.