Skip to content

Multi speaker training error #69

@kumdori88

Description

@kumdori88

Hi, I am trying to train a multi-speaker, but when I run "train.py" I get the following error:

INFO:korean_base_2:{'train': {'use_cuda': True, 'log_interval': 20, 'seed': 1234, 'epochs': 10000, 'learning_rate': 1.0, 'betas': [0.9, 0.98], 'eps': 1e-09, 'warmup_steps': 4000, 'scheduler': 'noam', 'batch_size': 64, 'ddi': True, 'fp16_run': True, 'save_epoch': 10}, 'data': {'load_mel_from_disk': True, 'training_files': 'filelists/korean_audio_text_train_filelist_suffle.txt', 'validation_files': 'filelists/korean_audio_text_val_filelist_suffle.txt', 'text_cleaners': ['korean_cleaners'], 'max_wav_value': 32768.0, 'sampling_rate': 22050, 'filter_length': 1024, 'hop_length': 256, 'win_length': 1024, 'n_mel_channels': 80, 'mel_fmin': 0.0, 'mel_fmax': 8000.0, 'add_noise': False, 'add_blank': True}, 'model': {'hidden_channels': 192, 'filter_channels': 768, 'filter_channels_dp': 256, 'kernel_size': 3, 'p_dropout': 0.1, 'n_blocks_dec': 12, 'n_layers_enc': 6, 'n_heads': 2, 'p_dropout_dec': 0.05, 'dilation_rate': 1, 'kernel_size_dec': 5, 'n_block_layers': 4, 'n_sqz': 2, 'prenet': True, 'mean_only': True, 'n_speakers': 12, 'gin_channels': 12, 'hidden_channels_enc': 192, 'hidden_channels_dec': 192, 'window_size': 4}, 'model_dir': './logs/korean_base_2'}
WARNING:korean_base_2:/data/tts/glow-tts is not a git repository, therefore hash value comparison will be ignored.
Traceback (most recent call last):
  File "init.py", line 58, in <module>
    main()
  File "init.py", line 51, in main
    _ = generator(x, x_lengths, y, y_lengths, gen=False)
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/data/tts/glow-tts/models.py", line 280, in forward
    x_m, x_logs, logw, x_mask = self.encoder(x, x_lengths, g=g)
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/data/tts/glow-tts/models.py", line 117, in forward
    logw = self.proj_w(x_dp, x_mask)
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/data/tts/glow-tts/models.py", line 29, in forward
    x = self.conv_1(x * x_mask)
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 259, in forward
    self.padding, self.dilation, self.groups)
RuntimeError: Given groups=1, weight of size [256, 204, 3], expected input[64, 192, 1] to have 204 channels, but got 192 channels instead

Single-speaker training works fine, but when I add "gin_channels" and start multi-speaker training, I get an error.
computer specifications are
GPU: RTX 3090 x2
CUDA: 11.1
Pytorch: 1.7.1+cu110

How can I solve this?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions