Dataset partitions

Hello,

I've seen there are VCTK model (48kHz) and LibriTTS model (24kHz), but I couldn't find what version of LibriTTS is actually used for training the model. However, I found this line:
https://github.com/facebookresearch/AudioDec/blob/main/config/autoencoder/symAD_libritts_24000_hop300.yaml#L17

I've seen similar nomenclatures for `train-clean-100.tar.gz + train-clean-360.tar.gz`. Is this the case?

Thanks,
Esteban

Start with config/autoencoder/symAD_libritts_24000_hop300.yaml at line 17 and trace the referenced LibriTTS training data. Compare the configured archive names with the LibriTTS partitions mentioned in the issue; done means confirming the dataset version and whether train-clean-100 and train-clean-360 are both used, then recording that clarification.

Written by the indexing model from the issue text.