Colab进入URV5 WebUI不能识别文件路径以及Faster Whisper报错无法执行

[cresstoo]

1. 在URV5 WebUI中复制粘贴文件路径“/content/GPT-SoVITS/input/abc.wav”无法识别，只能上传文件进行提取干声。

2. ASR打标这一步，选择Faster Whisper模型后，报错：

"/usr/local/bin/python" tools/asr/fasterwhisper_asr.py -i "output/slicer_opt" -o "output/asr_opt" -s large-v3 -l ja -p float16
loading faster whisper model: large-v3 large-v3
model.bin: 0% 0.00/3.09G [00:00<?, ?B/s]
config.json: 2.39kB [00:00, 293kB/s]

preprocessor_config.json: 100% 340/340 [00:00<00:00, 50.7kB/s]

tokenizer.json: 0.00B [00:00, ?B/s]
model.bin: 0% 10.5M/3.09G [00:00<01:18, 39.1MB/s]

vocabulary.json: 0.00B [00:00, ?B/s]
tokenizer.json: 602kB [00:00, 1.96MB/s]

model.bin: 1% 21.0M/3.09G [00:00<01:10, 43.6MB/s]
tokenizer.json: 1.04MB [00:00, 2.53MB/s]

vocabulary.json: 284kB [00:00, 1.00MB/s]
tokenizer.json: 2.48MB [00:00, 4.54MB/s]
model.bin: 1% 31.5M/3.09G [00:00<01:13, 41.4MB/s]

vocabulary.json: 1.07MB [00:00, 2.54MB/s]
model.bin: 100% 3.09G/3.09G [01:13<00:00, 41.7MB/s]
0% 0/3 [00:00<?, ?it/s]Unable to load any of {libcudnn_ops.so.9.1.0, libcudnn_ops.so.9.1, libcudnn_ops.so.9, libcudnn_ops.so}
Invalid handle. Cannot load symbol cudnnCreateTensorDescriptor
Aborted (core dumped)

Whisper-WebUI之前类似报错给出的方案是：降级 ctranslate2 == 4.4.0
不知道是不是这个原因。