Research Article
Application of Generative Adversarial Nets (GANs) in Active Sound Production System of Electric Automobiles
Table 4
Discriminator structure of the spectrogram GAN model.
| | Layer | Type | Length and width of convolution kernels | Number of convolution kernels |
| | G(z) | Input layer | — | — | | Conv2D (stride = 2) | Convolutional layer | 5∗5 | 64 | | LReLU (α = 0.2) | Activation layer | — | — | | Conv2D (stride = 2) | Convolutional layer | 5∗5 | 128 | | LReLU (α = 0.2) | Activation layer | — | — | | Conv2D (stride = 2) | Convolutional layer | 5∗5 | 256 | | LReLU (α = 0.2) | Activation layer | — | — | | Conv2D (stride = 2) | Convolutional layer | 5∗5 | 512 | | LReLU (α = 0.2) | Activation layer | — | — | | Conv2D (stride = 2) | Convolutional layer | 5∗5 | 1024 | | LReLU (α = 0.2) | Activation layer | — | — | | Reshape | Adjustment layer | — | — | | Fully connected layer | Fully connected layer | 16384∗1 | 64 |
|
|
