Windows errors.
FutureWarning: torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.
WeightNorm.apply(module, name, dim)
Missing tensor: mel_transformer.spectrogram.window
Missing tensor: mel_transformer.mel_scale.fb
Where are these and how can we fix them?
I have same problem happened to me, any progress going on? Has anyone encountered and resolved this issue before? Any guidance on how to fix it or ensure the required tensors are properly loaded would be greatly appreciated!
Found my own simple fix and runs great now:
Rename 'Spark-TTS-main' to 'Spark-TTS'
This still persists and I am not sure what you mean with the "fix" being renaming "Spark-TTS-main" to "Spark-TTS".
Where did you rename this exactly?
Downloaded zip produces "Spark-TTS-main", and the application is looking for "Spark-TTS" and you will have your own issues if your file structure is incorrect. Check and Move up double deep folders to see if that fixes it. Each file is asking for other files from different folders. Python can be fickle if you do not provide correct import paths.
The errors come up even when the application runs correctly, so no need to look into depths to see if that is the issue. Ensure files are in the correct spot.
Spark-TTS\pretrained_models
Spark-TTS\pretrained_models\Spark-TTS-0.5B
Spark-TTS\pretrained_models\Spark-TTS-0.5B\LLM
Ahh, you cloned that way - that problem does not persist if you clone with git CLI. Thanks though :)
Everything else loads fine and everythings seems to work fine without the MelScale and Spectogram Window parameters, so not sure how relevant they are for our research purposes.