训练声码器 的时候报错 CUFFT_INTERNAL_ERROR

[dushaofan]

CUDA_VISIBLE_DEVICES=0 python kantts/bin/train_hifigan.py --model_config speech_sambert-hifigan_tts_zh-cn_multisp_pretrain_16k/basemodel_16k/hifigan/config.yaml --resume_path speech_sambert-hifigan_tts_zh-cn_multisp_pretrain_16k/basemodel_16k/hifigan/ckpt/checkpoint_2000000.pth --root_dir newtest/training_stage/SSB0009_feats --stage_dir newtest/training_stage/SSB0009_hifigan_ckpt

cuFFT error: CUFFT_INTERNAL_ERROR
Traceback (most recent call last):
File "kantts/bin/train_hifigan.py", line 171, in train
trainer.train()
File "/home/dufei/git/KAN-TTS/kantts/train/trainer.py", line 199, in train
self.train_epoch()
File "/home/dufei/git/KAN-TTS/kantts/train/trainer.py", line 207, in train_epoch
self.train_step(batch)
File "/home/dufei/git/KAN-TTS/kantts/train/trainer.py", line 509, in train_step
mel_loss = self.criterion["mel_loss"](y_, y)
File "/root/anaconda3/envs/maasKan1/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "/home/dufei/git/KAN-TTS/kantts/train/loss.py", line 307, in forward
mel_hat = self.mel_spectrogram(y_hat)
File "/root/anaconda3/envs/maasKan1/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "/home/dufei/git/KAN-TTS/kantts/utils/audio_torch.py", line 175, in forward
x_stft = torch.stft(x, window=window, **self.stft_params)
File "/root/anaconda3/envs/maasKan1/lib/python3.7/site-packages/torch/functional.py", line 633, in stft
normalized, onesided, return_complex)
RuntimeError: cuFFT error: CUFFT_INTERNAL_ERROR
2023-08-17:16:52:02, INFO [train_hifigan.py:179] Successfully saved checkpoint @ 1steps.