search

chrisdonahue / wavegan

WaveGAN: Learn to synthesize raw audio with generative adversarial networks
MIT License
1.32k stars 282 forks source link

Invalid argument: You must feed a value for placeholder tensor 'ngl' with dtype int32 #101

Open mikemech83 opened 3 years ago

mikemech83 commented 3 years ago 
tensorflow.python.framework.errors_impl.InvalidArgumentError: 2 root error(s) found.
  (0) Invalid argument: You must feed a value for placeholder tensor 'ngl' with dtype int32
     [[{{node ngl}}]]
     [[G_z/_53]]
  (1) Invalid argument: You must feed a value for placeholder tensor 'ngl' with dtype int32
     [[{{node ngl}}]]

Hi, I am trying to generate samples using the models generated by the SpecGAN component. The script I have works great for the wavegan part but fails with the above error when trying to produce sounds using the specgan models. Any suggestions?

markhanslip commented 3 years ago

I have the exact same issue here. Tried defining z, G_z etc as tf.placeholder(tf.int32) before passing to graph.get_tensor_by_name() as per this issue but still no joy.

nhattruongpham commented 1 year ago

@mikemech83 @markhanslip @chrisdonahue Has anyone fixed this issue yet? Actually, I have faced the same problem when trying to generate speech with the trained SpecGAN model. Thanks,