Hello, Thanks for this great work!
I made the training of the synthesizer in 304.000 steps. During the training I get good result on the prediction of the spectogram and also good on audio generation.
for example:
But when i want to test the synthesize model, I get very bad spectogram, like this :
Hello, Thanks for this great work! I made the training of the synthesizer in 304.000 steps. During the training I get good result on the prediction of the spectogram and also good on audio generation.
for example:
But when i want to test the synthesize model, I get very bad spectogram, like this :
Any idea on what i make wrong please?