VITS : stack expects each tensor to be equal size, but got [445852] at entry 0 and [445851] at entry 17

[possum-code]

Training VITS model

Training Recipe : egs/ljspeech/TTS

Tried 2 different datasets that was converted to Ljspeech dataset format, same approach has worked earlier

Original Traceback (most recent call last):
File "/opt/conda/lib/python3.11/site-packages/torch/utils/data/_utils/worker.py", line 309, in _worker_loop
data = fetcher.fetch(index) # type: ignore[possibly-undefined]
^^^^^^^^^^^^^^^^^^^^
File "/opt/conda/lib/python3.11/site-packages/torch/utils/data/_utils/fetch.py", line 54, in fetch
data = self.dataset[possibly_batched_index]
~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/conda/lib/python3.11/site-packages/lhotse/dataset/speech_synthesis.py", line 67, in getitem
audio, audio_lens = collate_audio(cuts)
^^^^^^^^^^^^^^^^^^^
File "/opt/conda/lib/python3.11/site-packages/lhotse/dataset/collation.py", line 202, in collate_audio
audios = torch.stack(audios)
^^^^^^^^^^^^^^^^^^^
RuntimeError: stack expects each tensor to be equal size, but got [445852] at entry 0 and [445851] at entry 2

[JinZr]

hi, seems like this is a RuntimeError happened within lhotse, do you mind sharing the command and dataset you used for training?