Exploring Dízi Performance Parameters with Machine Learning
Abstract
In musical performance, three questions are often asked: what is the contribution of the instrument? What is the contribution of the player? Which musical exercise was performed? Here, we attempt to objectively quantify and compare the effects of the player, instrument and performed exercise by analyzing audio samples collected from a series of musical exercises performed on the traditional and modernized Chinese transverse flute, Dízi, for 5 expert players. Feature extraction is employed on the samples and a machine learning classifier algorithm is applied to the same dataset to confidently identify distinct populations: firstly, separating traditional vs modernized Dízi, secondly, separating player identity, and finally identifying the musical exercise performed all based purely on the Dízi performance’s acoustic output alone.
How to cite
Kai Ren Teo, Teck Seng Ng, Balamurali B T, and Jer-Ming Chen. 2018. “Exploring Dízi Performance Parameters with Machine Learning.” In Proceedings of the Australasian Computer Music Conference 2018, edited by Lindsay Vickery, 95–100. Edith Cowan University, Perth: Australasian Computer Music Association. https://computermusic.org.au/proceedings/ACMC2018-proceedings/teo-exploring-dizi-performance.html
BibTeX
@inproceedings{teo2018exploring,
author = {Kai Ren Teo and Teck Seng Ng and Balamurali B T and Jer-Ming Chen},
title = {{Exploring Dízi Performance Parameters with Machine Learning}},
booktitle = {Proceedings of the Australasian Computer Music Conference 2018},
editor = {Lindsay Vickery},
pages = {95--100},
year = {2018},
publisher = {Australasian Computer Music Association},
address = {Edith Cowan University, Perth},
issn = {2206-5296},
url = {https://computermusic.org.au/proceedings/ACMC2018-proceedings/13Teo.pdf}
}