ivanvovk / WaveGrad

Implementation of WaveGrad high-fidelity vocoder from Google Brain in PyTorch.

ivanvovk/WaveGrad Issues

Evaluation tools
Updated 2 years ago
Audio quality improvements
Updated 2 years ago6
Poor Synthesis Quality on 44k Sample Rate
Updated 2 years ago1
Static Noise with f_max = 10000
Updated 2 years ago
Using NVIDIA RTX 3090 GPU?
Updated 2 years ago
slow training in single GPU
Updated 3 years ago1
Interpolation and Conv order in Upsample module
Updated 3 years ago1
The order of upsampling_dilations
Updated 3 years ago1
How to make it work with TPU?
Updated 3 years ago2
Training so slow
Updated 3 years ago3
TTS without Text?
Updated 3 years ago2
schedules model for other dataset and different sample rate
Closed 3 years ago4
What loss values are reasonable?
Closed 4 years ago3
best noise schedule
Closed 4 years ago1
predict_start_from_noise
Closed 4 years ago2
Exponents calculation in positional encoding
Closed 4 years ago9
Were your `generated_samples` generated using a model trained with AMP?
Closed 4 years ago2
inference.py seems not loading the specified checkpoint
Closed 4 years ago1
Matplotlib API change & NaNs for short clips & new hop_length
Closed 4 years ago27
ValueError: low >= high
Closed 4 years ago2
reduce noise with lower iteration
Closed 4 years ago3
Unable to load the pre-trained parameters for inference
Closed 4 years ago2
configuration for hop_size 256
Closed 4 years ago2
loss explode question
Closed 4 years ago7
Is it possible to load .npy spectrograms directly?
Closed 4 years ago3
On CPU performance
Closed 4 years ago2