TABLE 2.

Exact parameters of the used deep convolutional network (FSDD data set).

Layer	Type	Input-output-dim	Activation	Characteristics
1	Convolution 2D	9,131, 30, 1; 9,102, 1, 32	ReLu
2	MaxPooling 2D	9,102, 1, 32; 4,551, 1, 32		Pool size: (2, 1)
3	DropOut	4,551, 1, 32; 4,551, 1, 32		Droupout: 0.2
4	Convolution 2D	4,551, 1, 32; 4,547, 1, 64	ReLu
5	MaxPooling 2D	4,547, 1, 64; 2,273, 1, 64		Pool size: (2, 1)
6	DropOut	2,273, 1, 64; 2,273, 1, 64		Dropout: 0.2
7	Convolution 2D	2,273, 1, 64; 2,272, 1, 32	ReLu
8	MaxPooling 2D	2,272, 1, 32; 1,136, 1, 32		Pool size: (2, 1)
9	DropOut	1,136, 1, 32; 1,136, 1, 32		Dropout: 0.2
10	Flatten	1,136, 1, 32; 36,352
11	Dense	36,352; 400	ReLu
12	DropOut	400; 400		Dropout: 0.2
13	Dense	400; 50	ReLu
14	Dense	50; 10	Softmax