Activity
Fix KL loss weight logging in distributed training
Fix KL loss weight logging in distributed training
Fix inference for single speaker/language on all utterances
Fix inference for single speaker/language on all utterances
Add TODOs and fixes of mysterious origin
Add TODOs and fixes of mysterious origin
Fix tensorboard logging of validation utterances
Fix tensorboard logging of validation utterances
Refactor effective batch size handling and validation
Refactor effective batch size handling and validation
Add TVC-GMM to mitigate over-smoothing mel spectrograms
Add TVC-GMM to mitigate over-smoothing mel spectrograms
Keep trailing spaces in meta text when adding speaker IDs
Keep trailing spaces in meta text when adding speaker IDs
Update README for added language embeddings
Update README for added language embeddings
Add additional conditioning vector, e.g. for language
Add additional conditioning vector, e.g. for language
Save only data for dumped mels in inference
Save only data for dumped mels in inference
Refactor data pipeline for parallel preprocessing
Refactor data pipeline for parallel preprocessing