啥报错都没有了，但就是不像，时而电流音

用的是Quick Start (Newbie) 快速开始 (新手友好版)教程
代码分支是tag0.0.1 模型是配对的那个 也解决了 https://github.com/babysor/MockingBird/issues/37 兼容性问题
还用过main分支 模型为中文完整教程里推荐的第一个
配好后都不再有报错
win10
nvcc -V 显示 cudatools 10.2
nvidia-smi 显示 cuda version 11.6
conda 4.9.2
pytorch 1.10.1
torch.cuda.is_available() 为 true
用的录音都在5-8秒左右
求助

Start by reproducing the distorted or electrical-noise output with the Quick Start (Newbie) tag0.0.1 setup and the main branch, using the reported Windows, CUDA, conda, PyTorch, model, and 5–8 second recording details. Compare the two configurations and review issue #37 for compatibility context. Done means identifying a reproducible cause and confirming normal speech output after the relevant environment or model setup is corrected.

Written by the indexing model from the issue text.