Prediction of trumpet performance descriptors using machine learning

Topical Issue - Musical Acoustics: Latest Advances in Analytical, Numerical and Experimental Methods Tackling Complex Phenomena in Musical Instruments

Open Access

Table 2

Mean Absolute Percentage Error (MAPE) and Standard deviation of the Absolute Percentage Error (StdAPE) in predicting descriptor values using various machine learning algorithms on the test set. The best performance for each descriptor is highlighted in bold.

	P_min1 (%)	P_min2 (%)	p_min (%)	p_max (%)	Slope (%)
Baseline	5.75 ± 0.01	4.75 ± 0.01	4.24 ± 0.01	1.56 ± 0.01	1.38 ± 0.01
LassoLars	0.57 ± 0.02	0.41 ± 0.03	0.76 ± 0.03	0.48 ± 0.02	0.8 ± 0.05
Linear regression	0.72 ± 0.06	0.67 ± 0.07	1.09 ± 0.07	0.66 ± 0.05	1.04 ± 0.07
SVR (linear)	0.71 ± 0.07	0.65 ± 0.05	1.1 ± 0.08	0.67 ± 0.05	1.06 ± 0.07
Ridge regression	0.72 ± 0.06	0.67 ± 0.07	1.08 ± 0.07	0.66 ± 0.05	1.04 ± 0.07
SVR (nonlinear)	1.21 ± 0.08	1.17 ± 0.06	1.43 ± 0.07	0.7 ± 0.04	0.98 ± 0.04
Nearest neighbours	4.62 ± 0.15	3.71 ± 0.12	3.49 ± 0.12	1.29 ± 0.04	1.13 ± 0.04
	f_0min (%)	f_0max (%)	$P_{f_{0 \min}}$ ${P}_{{f}_{0\mathrm{min}}}$	f_0 fold (%)	f_0H (%)
Baseline	0.26 ± 0.001	0.27 ± 0.001	3.81 ± 0.01	0.26 ± 0.001	0.25 ± 0.001
LassoLars	0.04 ± 0.002	0.04 ± 0.003	0.59 ± 0.03	0.03 ± 0.002	0.04 ± 0.002
Linear regression	0.05 ± 0.003	0.05 ± 0.004	0.85 ± 0.06	0.05 ± 0.003	0.06 ± 0.01
SVR (linear)	0.05 ± 0.005	0.06 ± 0.005	0.86 ± 0.06	0.05 ± 0.004	0.06 ± 0.01
Ridge regression	0.05 ± 0.003	0.05 ± 0.004	0.84 ± 0.06	0.05 ± 0.003	0.06 ± 0.01
SVR (nonlinear)	0.05 ± 0.003	0.06 ± 0.004	1.1 ± 0.06	0.05 ± 0.003	0.07 ± 0.01
Nearest neighbours	0.2 ± 0.01	0.2 ± 0.01	3.12 ± 0.1	0.2 ± 0.01	0.19 ± 0.01

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.