Vits2 prototype

[p0p4k]

continuing where i left earlier, adding some test cases.

• to do -> fix test cases to take in mel instead of linear spectrogram (if anywhere)

[erogol]

Looks like a bug (from CI)

======================================================================
ERROR: test_d_vector_forward (tests.tts_tests.test_vits2.TestVits2)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/runner/work/TTS/TTS/tests/tts_tests/test_vits2.py", line 214, in test_d_vector_forward
    model = Vits2.init_from_config(config, verbose=False).to(device)
  File "/home/runner/work/TTS/TTS/TTS/tts/models/vits2.py", line 1862, in init_from_config
    return Vits2(new_config, ap, tokenizer, speaker_manager, language_manager)
  File "/home/runner/work/TTS/TTS/TTS/tts/models/vits2.py", line 652, in __init__
    self.use_transformer_flow_layer = self.args.use_transformer_flow_layer
  File "/opt/hostedtoolcache/Python/3.9.18/x64/lib/python3.9/site-packages/coqpit/coqpit.py", line 626, in __getattribute__
    value = super().__getattribute__(arg)
AttributeError: 'Vits2Args' object has no attribute 'use_transformer_flow_layer'

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.

[Marioando]

Changing use_transformer_flow to use_transformer_flow_layer in Vits2Args get arround of this but the I get this error :
/opt/conda/lib/python3.10/site-packages/TTS/tts/layers/vits2/transformer.py:477, in ConditionalRelativePositionTransformer.init(self, in_channels, out_channels, hidden_channels, hidden_channels_ffn, num_heads, num_layers, kernel_size, dropout_p, rel_attn_window_size, input_length, layer_norm_type, cond_channels, cond_layer_idx)
475 super().init()
476 self.cond_channels = cond_channels
--> 477 if cond_layer_idx < 0 or cond_layer_idx >= num_layers:
478 raise ValueError(" [!] cond_layer_idx should be in [0, num_layers)")
479 self.cond_layer_idx = cond_layer_idx

TypeError: '<' not supported between instances of 'NoneType' and 'int'