Table 2 Structure of the convolutional neural networks (CNNs) used for (a) effect classification and (b) parameter extraction. Each guitar effect investigated had two parameters resulting in an output dimension of two for the CNN for extraction of the guitar effect settings

(a) CNN structure for effect classification

Convolutional

3 \(\times\) 3

32

ReLU

-

Batch norm.

-

-

-

-

Max pooling

2 \(\times\) 2

-

-

-

Convolutional

3 \(\times\) 3

64

ReLU

0.3

Batch norm.

-

-

-

-

Max pooling

2 \(\times\) 2

-

-

-

Flatten

-

-

-

-

Dense

-

64

ReLU

0.3

Batch norm.

-

-

-

-

Dense

-

64

ReLU

0.3

Batch norm.

-

-

-

-

Dense (output)

-

11

Softmax

-

(b) CNN structure for effect parameter extraction

Convolutional

3 \(\times\) 3

6

ReLU

-

Batch normalization

-

-

-

-

Max pooling

2 \(\times\) 2

-

-

-

Convolutional

3 \(\times\) 3

12

ReLU

0,2

Batch normalization

-

-

-

-

Max pooling

2 \(\times\) 2

-

-

-

Flatten

-

-

-

-

Dense

-

64

ReLU

0,2

Batch normalization

-

-

-

-

Dense

-

64

ReLU

0,2

Batch normalization

-

-

-

-

Dense (output)

-

2

Sigmoid

-