gpu换大的后碰到这个问题，是什么原因呢？

:6: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.
Traceback (most recent call last):
File "gen_audio_from_srt.py", line 430, in
Path("vocoder/saved_models/pretrained/g_hifigan.pt"), fpath, gen_materials
File "gen_audio_from_srt.py", line 144, in generate_wav
gen_one_wav(synthesizer, embed, processed_texts, file_name, hint_txt)
File "gen_audio_from_srt.py", line 74, in gen_one_wav
generated_wav = encoder.preprocess_wav(generated_wav)
File "/home/evers/MyGithub/MockingBird/encoder/audio.py", line 46, in preprocess_wav
wav = normalize_volume(wav, audio_norm_target_dBFS, increase_only=True)
File "/home/evers/MyGithub/MockingBird/encoder/audio.py", line 115, in normalize_volume
if (dBFS_change < 0 and increase_only) or (dBFS_change > 0 and decrease_only):
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Start with the traceback in gen_audio_from_srt.py, especially generate_wav and gen_one_wav, then inspect encoder/audio.py at preprocess_wav and normalize_volume. Reproduce the failure with the reported generation flow and determine why normalize_volume receives an ambiguous value. Done should include a confirmed cause and a verified correction, but the issue does not specify an expected behavior or test location.

Written by the indexing model from the issue text.