Skip to main content

Table 4 \(95\,\%\) confidence interval of the mean absolute error of the CNN across the fivefold cross-validation of the parameter extraction and all volumes for the individual effects. Results are given for Jürgens et al. [12] as well. The lowest errors are highlighted using bold font

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

 

Distortion

Tremolo

Slapback delay

 

Gain [\(10^{-2}\)]

Tone [\(10^{-2}\)]

Depth [\(10^{-2}\)]

Freq. [\(10^{-2}\)]

Time [\(10^{-2}\)]

Mix [\(10^{-2}\)]

CNN + chromagram

3.2 ± 0.15

5.1 ± 0.48

6.3 ± 0.17

6.1 ± 0.38

6.2 ± 0.25

5.4 ± 0.3

CNN + GFCC

1.6 ± 0.25

1.7 ± 0.15

3.4 ± 0.2

6.3 ± 0.33

3.8 ± 0.07

2.7 ± 0.14

CNN + MFCC

1.4 ± 0.09

1.6 ± 0.09

4.9 ± 0.45

6.5 ± 0.15

3.9 ± 0.1

3.1 ± 0.22

CNN + spectrogram

2.1 ± 0.63

2.3 ± 0.18

3.9 ± 0.25

5.2 ± 0.59

4.9 ± 0.09

3.7 ± 0.47

Jürgens et al.

4.8 ± 0.82

7.1 ± 2.4

12.9 ± 3.4

11.8 ± 2.08

22.8 ± 7.6

20.07 ± 6.1