Noise Estimation For Denoising Diffusion Models
Here are some samples from our model for you to listen:
- Ground Truth - original sample from LJ dataset
- Ours - Generated with 6 denoising steps using our proposed method
- WaveGrad - sample generated with network and noise schedule (grid search) for 6 denosing steps provided in the Git
LJ Speech samples
Ground Truth |
Ours |
WaveGrad |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|