Audio Model Blocks
WaveGrad UBlock
Introduced by Chen et al. in WaveGrad: Estimating Gradients for Waveform GenerationThe WaveGrad UBlock is used for upsampling in WaveGrad. Neural audio generation models often use large receptive field. Dilation factors of four convolutional layers are 1, 2, 1, 2 for the first two UBlocks and 1, 2, 4, 8 for the rest. Orthogonal initialization is used.
Source: WaveGrad: Estimating Gradients for Waveform Generation
Papers
| Paper | Code | Results | Date | Stars |
|---|
Tasks
| Task | Papers | Share |
|---|---|---|
| Speech Synthesis | 5 | 45.45% |
| Image Generation | 2 | 18.18% |
| Denoising | 2 | 18.18% |
| Text-To-Speech Synthesis | 2 | 18.18% |
Usage Over Time
Components
| Component | Type |
|
|---|---|---|
1x1 Convolution
|
Convolutions | |
Convolution
|
Convolutions | |
Leaky ReLU
|
Activation Functions | |
Residual Connection
|
Skip Connections |
