Skip to main content

Table 3 \(95\,\%\) confidence intervals for the classification accuracy of the convolutional neural network (CNN) using the listed time-frequency representations as well as the accuracy of the baseline support vector machine (SVM) using the method by Stein et al. [4]. As subsequent parameter extraction is not necessarily compromised by a confusion of slapback delay (SD) and feedback delay (FD), the respective accuracies, when SD and FD are treated as the same effect, are given as well

From: Convolutional neural networks for the classification of guitar effects and extraction of the parameter settings of single and multi-guitar effects from instrument mixes

Method

GEC-GIM

GEC-GIM (SD = FD)

IDMT-SMT

SVM

85.0 % ± 0.44 %

89.4 % ± 0.32 %

96.1 % ± 0.3 %

CNN + spectrogram

90.0 % ± 0.57 %

95.8 % ± 0.52 %

97.4 % ± 0.7 %

CNN + MFCCs

87.7 % ± 0.52 %

93.4 % ± 0.32 %

96.5 % ± 0.13 %

CNN + GFCCs

24.7 % ± 27.3 %

28.7 % ± 30.47 %

97.4 % ± 0.3 %

CNN + chromagram

87.0 % ± 0.64 %

92.9 % ± 0.12 %

86.2 % ± 0.2 %