首次训练报错：soundfile.LibsndfileError: <exception str() failed>

[hiatus233]

You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference. 0%| | 0/15 [00:10<?, ?it/s] Traceback (most recent call last): File "train_ms.py", line 840, in run() File "train_ms.py", line 361, in run train_and_evaluate( File "train_ms.py", line 423, in train_and_evaluate for batch_idx, ( File "E:\Bert-VITS2-2.3\Bert-VITS2-2.3\venv\lib\site-packages\tqdm\std.py", line 1178, in iter for obj in iterable: File "E:\Bert-VITS2-2.3\Bert-VITS2-2.3\venv\lib\site-packages\torch\utils\data\dataloader.py", line 630, in next data = self._next_data() File "E:\Bert-VITS2-2.3\Bert-VITS2-2.3\venv\lib\site-packages\torch\utils\data\dataloader.py", line 1345, in _next_data return self._process_data(data) File "E:\Bert-VITS2-2.3\Bert-VITS2-2.3\venv\lib\site-packages\torch\utils\data\dataloader.py", line 1371, in _process_data data.reraise() File "E:\Bert-VITS2-2.3\Bert-VITS2-2.3\venv\lib\site-packages\torch_utils.py", line 694, in reraise raise exception soundfile.LibsndfileError: <exception str() failed>

有无大佬知道这是为什么？打标的过程中出现繁中是影响吗？

[hiatus233]

初步感觉是数据集的问题，在减小数据集规模和去掉某些音频之后就没有这个报错了