CriDora

> **Describe the bug** 命令行使用`git clone -c core.symlinks=true https://github.com/wenet-e2e/wetts.git`将repo克隆至本地后，尝试使用Android Studio编译，cmake报错找不到依赖 > > **To Reproduce** Steps to reproduce the behavior: > > 1. 使用Android Studio打开wetts\runtime\android目录 > 2. 点击Build->Rebuild Project > 3....

> 试试直接把目录拷贝过去呢？ 是的，我昨天把#include "fst/fstlib.h" 等文件替换成了绝对路径，这个错误就没有了

你好，这个问题解决了吗，我也遇到了合成音频开头回头滴的声音，而且克隆出来的声音跟原声音相比音量小很多

**This is the full output：** [ 1%] Building CXX object _deps/googletest-build/googletest/CMakeFiles/gtest.dir/src/gtest-all.cc.o [ 1%] Building CXX object _deps/gloo-build/gloo/CMakeFiles/gloo.dir/allgatherv.cc.o [ 1%] Building CXX object _deps/gloo-build/gloo/CMakeFiles/gloo.dir/algorithm.cc.o [ 1%] Building CXX object _deps/gloo-build/gloo/CMakeFiles/gloo.dir/allreduce.cc.o [...

> Hi@CriDora, thanks for using our code! From my experience, the nccl timeout problem is usually related to a problem in the dataset loading, like it takes too long time...

> ## Problem Overview > calc_metrics.py: error: unrecognized arguments > > ## Steps Taken > (Detail your attempts to resolve the issue, including any relevant steps or processes.) > >...

> do you mean `--checkpoint`? I've tried that, but it doesn't work. I think the LR scheduler being different makes the training loss go kind of crazy and in the...

> I think the author will update the code for this issue, however I have a temporary solution you can refer to. > > 1. When starting a new training...

> I thought the above code would make `step` go back to `current_step` when starting a new epoch. I modified it a bit. > > ``` > # Save init...

> Yes, you need to modify the `current_epoch` and `current_step` values ​​in `cosyvoice.fromscratch.yaml` every time you resume training from a checkpoint. > > You can see `current_epoch` and `current_step` in...