Jerry Chang
Jerry Chang
I am not able to get the results with the same quality as the paper either.
I'm also interested in this process. @jixinya Do you have a plan to release the audio disentanglement module?
Yes you can remove all the rank checks in the train.py.
btw, what's the version (?x, g=?) of this shufflenet?
I'm now converting tf mobilenet-v2 to caffe model. I use the protrotxt here (https://github.com/shicai/MobileNet-Caffe) and have converted all the params correctly, but cannot get the same accuracy Do I miss...
@bobauditore Thanks for your advice. I still want to convert movilenet-v2 ckpt to caffe model Except for the different "depthwise convolution"in that repository(https://github.com/shicai/MobileNet-Caffe), I found another problem during conversion. The...
I would recommend training the vocoder first before you train the converter. Normally it takes a few days for the vocoder to output decent voices. You can check that before...
Looks like your model only learns to predict Mel from previous step. You can check the intermediate tensors in the model to verify this. A possible solution to this is...
I've found that problem too. Since I revised the prototxt file from other's, I found that mobilenet graphs are different in that layer you mentioned. Tensorflow version does not have...
The function is used to check the square error of a layer's output. Since I use pad=2 and 2 slice layers to solve the padding problem, these 2 layers' shape...