[2noise/ChatTTS]使用自己的音频文件克隆音色，生成新的声音文件时报错 “_lzma.LZMAError: Corrupt input data”

使用自己的音频文件(约6秒，标准英式native speaker发音, 正常语速)克隆音色，生成新的声音文件时报错 “_lzma.LZMAError: Corrupt input data” 请参见如下 "运行结果" 和 "测试代码"

no GPU or NPU found, use CPU instead

found invalid characters: {'1', '0', '-'}
text:   0%|▏                                                                            | 1/384(max) [00:00,  3.05it/s]`get_max_cache()` is deprecated for all Cache classes. Use `get_max_cache_shape()` instead. Calling `get_max_cache()` will raise error from v4.48
text:  10%|███████▋                                                                    | 39/384(max) [00:04,  9.69it/s]
code:  13%|█████████▌                                                                | 265/2048(max) [00:21, 12.32it/s]
text:   8%|██████▎                                                                     | 32/384(max) [00:03,  8.12it/s]
Traceback (most recent call last):
  File "C:\code\projs\ChatTTS\test_with_upload.py", line 41, in <module>
    wavs = chat.infer(
           ^^^^^^^^^^^
  File "C:\code\projs\ChatTTS\ChatTTS\core.py", line 261, in infer
    for wavs in res_gen:
  File "C:\code\projs\ChatTTS\ChatTTS\core.py", line 436, in _infer
    self._infer_code(
  File "C:\Programs\Anaconda\envs\chattts\Lib\site-packages\torch\utils\_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "C:\code\projs\ChatTTS\ChatTTS\core.py", line 631, in _infer_code
    self.speaker.apply(
  File "C:\Programs\Anaconda\envs\chattts\Lib\site-packages\torch\utils\_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "C:\code\projs\ChatTTS\ChatTTS\model\speaker.py", line 32, in apply
    spk_emb_tensor = torch.from_numpy(self._decode(spk_emb))
                                      ^^^^^^^^^^^^^^^^^^^^^
  File "C:\code\projs\ChatTTS\ChatTTS\model\speaker.py", line 148, in _decode
    lzma.decompress(
  File "C:\Programs\Anaconda\envs\chattts\Lib\lzma.py", line 343, in decompress
    res = decomp.decompress(data)
          ^^^^^^^^^^^^^^^^^^^^^^^
_lzma.LZMAError: Corrupt input data

import ChatTTS
import torch
import scipy
from typing import Optional
import torchaudio
from tools.audio import load_audio

chat = ChatTTS.Chat()
chat.load(compile=False)

def on_upload_sample(sample_audio_input: Optional[str]) -> str:
    sample_audio = torch.tensor(load_audio(sample_audio_input, 24000)).to('cpu')
    spk_smp = chat.sample_audio_speaker(sample_audio)
    del sample_audio
    return spk_smp

spk_smb = on_upload_sample(r"C:\Users\admin\Desktop\2.wav")

# 这里的文字就是音频 2.wav 中的文字
smp_txt = ["Our eco-friendly packaging is made from 100 percent biodegradable materials, including recycled paper and plant"]
reftext = chat.infer(smp_txt, refine_text_only=False)

params_infer_code = ChatTTS.Chat.InferCodeParams(
    txt_smp=reftext,
    spk_emb=spk_smb,
    temperature=0.8,
    top_P=0.4, 
    top_K=7,
)

texts = ["The output indicates that Torchaudio has successfully detected the soundfile backend.", "This is a valid backend for audio processing,", "but it does not rely on Sox."]

wavs = chat.infer(
    texts,
    params_infer_code=params_infer_code,
)

torchaudio.save("3.wav", torch.from_numpy(wavs[0]).unsqueeze(0), 24000)