Yesiler, FurkanBozkurt, BarisSerra, Xavier2019-03-062019-03-062018Yesiler F, Bozkurt B, Serra X. Makam recognition using extended pitch distribution features and multi-layer perceptrons. In: Georgaki A, Andreopoulou A, editors. Proceedings of the 15th Sound and Music Computing Conference (SMC2018). Sonic crossing; 2018 Jul 4-7; Limassol, Xipre. Limassol: Cyprus University of Technology; 2018. p.249-53. DOI: 10.5281/zenodo.1254253http://hdl.handle.net/10230/36756Comunicació presentada a: 15th Sound and Music Computing Conference (SMC2018). Sonic crossing, celebrat a Limassol, Xipre, del 4 al 7 de juliol de 2018.This work focuses on automatic makam recognition taskfor Turkish makam music (TMM) using pitch distributionsthat are widely used in mode recognition tasks for variousmusic traditions. Here, we aim to improve the performanceof previous works by extending distribution features andperforming parameter optimization for the classifier. Mostmusic theory resources specifically highlight two aspectsof the TMM makam concept: use of microtonal intervalsand an overall melodic direction that refers to the designof melodic contour on the song/musical piece level. Previous studies for makam recognition task already utilizethe microtonal aspect via making use of high resolutionhistograms (using much finer bin width than one 12th ofan octave). This work considers extending the distributionfeature by including distributions of different portions ofa performance to reflect the long-term characteristics re-ferred in theory for melodic contour, more specifically forintroduction and finalis. Our design involves a Multi-LayerPerceptron classifier using an input feature vector com-posed of pitch distributions of the first and the last sectionstogether with the overall distribution, and the mean accu-racy of 10 iterations is 0.756. The resources used in thiswork are shared for facilitating further research in this direction.application/pdfeng© 2018 Furkan Yesiler et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 Unported License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.info:eu-repo/semantics/conferenceObjecthttp://dx.doi.org/10.5281/zenodo.1254253info:eu-repo/semantics/openAccess