使用paraformer-zh无法识别这一集中文音频

[zshnb]

Notice: In order to resolve issues more efficiently, please raise issue following the template.
（注意：为了更加高效率解决您遇到的问题，请按照模板提问，补充细节）

🐛 Bug

音频地址，使用如下模型代码识别

 AutoModel(
            model="paraformer-zh", model_revision="v2.0.4",
            vad_model="fsmn-vad", vad_model_revision="v2.0.4",
            punc_model="ct-punc-c", punc_model_revision="v2.0.4",
            device=self.device,
            disable_pbar=True,
            disable_update=True,
            log_level="NOTSET",
            disable_log=True,
        )

无法识别内容

To Reproduce

Steps to reproduce the behavior (always include the command you ran):

1. Run cmd '....'
2. See error

Code sample

Expected behavior

Environment

• OS (e.g., Linux):
• FunASR Version (e.g., 1.0.0): 2.0.4
• ModelScope Version (e.g., 1.11.0):
• PyTorch Version (e.g., 2.0.0):
• How you installed funasr (pip, source): pip
• Python version:
• GPU (e.g., V100M32) A10
• CUDA/cuDNN version (e.g., cuda11.7): 12.4
• Docker version (e.g., funasr-runtime-sdk-cpu-0.4.1)
• Any other relevant information:

Additional context