Export Kokoro 1.0 to sherpa-onnx

[csukuangfj]

Need to add C++ code to support it.

[ahadjawaid]

I found some issues when trying to run your code.

1. Additional dependencies
   I had to install these dependencies:
   pip install pypinyin ordered_set jieba cn2an

2. Missing script call in run.sh
   The script ./add_meta_data.py is not called in run.sh. You can add this before running ./test.py.

Suggestion: Save voice .pt files in a separate directory

To keep the directory organized, consider storing voice files in a voices/ directory. Here are the necessary changes:

.gitignore

   voices/  

add_meta_data.py (line 18)

   style = torch.load("./voices/af_alloy.pt", weights_only=True, map_location="cpu")  

generate_voices_bin.py (line 74)

   f"voices/{speaker}.pt",  

run.sh (line 80)

   mkdir -p voices  
   for v in "${voices[@]}"; do  
     if [ ! -f voices/$v.pt ]; then  
       wget -O voices/$v.pt https://huggingface.co/hexgrad/Kokoro-82M/resolve/main/voices/$v.pt