in
1 with torch.no_grad():
2 # get rhythm (alignment map) using tacotron 2
----> 3 mel_outputs, mel_outputs_postnet, gate_outputs, rhythm = mellotron.forward(x)
4 rhythm = rhythm.permute(1, 0, 2)
/datas/mellotron/model.py in forward(self, inputs)
600
601 embedded_inputs = self.embedding(inputs).transpose(1, 2)
--> 602 embedded_text = self.encoder(embedded_inputs, input_lengths)
603 embedded_speakers = self.speaker_embedding(speaker_ids)[:, None]
604 embedded_gst = self.gst(targets, output_lengths)
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in __call__(self, *input, **kwargs)
545 result = self._slow_forward(*input, **kwargs)
546 else:
--> 547 result = self.forward(*input, **kwargs)
548 for hook in self._forward_hooks.values():
549 hook_result = hook(self, input, result)
/datas/mellotron/model.py in forward(self, x, input_lengths)
186 def forward(self, x, input_lengths):
187 for conv in self.convolutions:
--> 188 x = F.dropout(F.relu(conv(x)), drop_rate, self.training)
189
190 x = x.transpose(1, 2)
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in __call__(self, *input, **kwargs)
545 result = self._slow_forward(*input, **kwargs)
546 else:
--> 547 result = self.forward(*input, **kwargs)
548 for hook in self._forward_hooks.values():
549 hook_result = hook(self, input, result)
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/container.py in forward(self, input)
90 def forward(self, input):
91 for module in self._modules.values():
---> 92 input = module(input)
93 return input
94
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in __call__(self, *input, **kwargs)
545 result = self._slow_forward(*input, **kwargs)
546 else:
--> 547 result = self.forward(*input, **kwargs)
548 for hook in self._forward_hooks.values():
549 hook_result = hook(self, input, result)
/datas/mellotron/layers.py in forward(self, signal)
34
35 def forward(self, signal):
---> 36 conv_signal = self.conv(signal)
37 return conv_signal
38
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in __call__(self, *input, **kwargs)
545 result = self._slow_forward(*input, **kwargs)
546 else:
--> 547 result = self.forward(*input, **kwargs)
548 for hook in self._forward_hooks.values():
549 hook_result = hook(self, input, result)
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/conv.py in forward(self, input)
198 _single(0), self.dilation, self.groups)
199 return F.conv1d(input, self.weight, self.bias, self.stride,
--> 200 self.padding, self.dilation, self.groups)
201
202
RuntimeError: Calculated padded input size per channel: (4). Kernel size: (5). Kernel size can't be greater than actual input size
This seems to be self.conv(signal) wrong in the mellotron/layers.py, but I didn't change anything before I run the code.
Can you give me some suggestions?
And my dataset is thchs30, it is a mardarin corpus.
BabaiLi
commented
3 years ago
When I run inference.ipynb, I got a error in the In[16]:
This seems to be
self.conv(signal)wrong in the mellotron/layers.py, but I didn't change anything before I run the code. Can you give me some suggestions? And my dataset is thchs30, it is a mardarin corpus.