about the Phonological and replearning problem 

thank you for your outstanding work! but i have some problems,
first about the phonological, i input my own wav flie(english and chinese) but get nothing in the pic, so i wonder to know how 
to fix it
second about the replearning, typeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.
sorry to bother you