tensorflow-wavenet
tensorflow-wavenet copied to clipboard
ibab
A TensorFlow implementation of DeepMind's WaveNet paper
How are division corpus data VCTK for training, testing and validation? For example 80%, 10%, 10%
https://github.com/ibab/tensorflow-wavenet/blob/3c973c038c8c2c20fef0039f111cb04139ff594b/wavenet/model.py#L309 I would like to know why do we have this slice operation over the skip connection!!! Any help is much appreciated!!
I'm confused as to this parameter listed in wavenet_params.json "dilations": [1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512,...
I have a mistake, it seems a bit troublesome I use `Python train.py --data_dir=corpus --gc_channels=32` Train I use `Python generate.py --samples 16000 logdir/train/2017-02-xxxxx-xx-xx/model.ckpt-xxx` Generate There is no problem with this...
There is an option for adding, or not adding, biases. The wavenet_params.json file says to use biases, the arguments in the WaveNetModel says not to use biases. Can anyone please...
Is anyone maybe aware of an implementation of this paper? https://arxiv.org/pdf/1806.10474.pdf It seems to rely on much of the same stuff, except now it has the ability to generate sort...
This implementation creates an extra set of weights in the last residual block that do not get trained because the output from these weights never gets used for the model's...
