search

leimao / Singing-Voice-Separation-RNN

Singing-Voice Separation From Monaural Recordings Using Deep Recurrent Neural Networks
MIT License
60 stars 20 forks source link

assert len #1

Open Tortoise17 opened 5 years ago

Tortoise17 commented 5 years ago

Dear Friend.

I want to ask a question from you. !! in model, why is this necessary?

assert len(num_hidden_units) == num_rnn_layer

and why can't we change the number of layers with this exception?

Please guide..

Greetings.

leimao commented 5 years ago

Because we are using multi-layer RNN, we have to specify the number of hidden units for each layer. Therefore, there is such assertion.

Tortoise17 commented 5 years ago

Dear Leimao.

Thank you so much for kind reply. Yes, I understood this. But, the hidden units are by default 256 and layers 3, and if layers are increased, the error comes. Is it possible to increase layers here? I mean how it is fixed with that ? If you can guide a bit.

Greetings .

leimao commented 5 years ago

Did you mean if you use, say, num_rnn_layer = 4, num_hidden_units = [256, 256, 256, 256], the program will be broken?

Tortoise17 commented 5 years ago

Yes something like this.