VITS2 Backbone with bert

• 1.若干时长的高质量语音文本
• 2.音频切割工具slicer-gui
• 3.数据处理和标注工具，来自b站up主领航员未鸟【Bert-VITS2/VITS】更快更准确！自动标注一键包来啦】 https://www.bilibili.com/video/BV1dr4y1X7RL/?share_source=copy_web&vd_source=0850903c2c3f5ed7b220c9bda4e285f6

+!git clone https://github.com/MengXing15646/Bert-VITS2

1.原始数据音频必须是.wav格式，使用slicer-gui将文件切割为短于20s的音频 2.按照【Bert-VITS2/VITS】更快更准确！自动标注一键包来啦】 https://www.bilibili.com/video/BV1dr4y1X7RL/?share_source=copy_web&vd_source=0850903c2c3f5ed7b220c9bda4e285f6教程处理数据 3.将处理后的clean_barbara.list文件上传到filelists文件夹，并改名成【自己需要的名字】 4.将preprocess_text文件中的名字改为【自己需要的名字】 5.创建raw文件夹和它的子文件夹【自己需要的名字】，将处理后的音频上传到【自己需要的名字】文件夹

• %cd /content/Bert-VITS2/
• !pip install -r requirements.txt

• hugging face不提供直链下载，使用二次分发下载模型`
• !wget -P bert/chinese-roberta-wwm-ext-large/ https://huggingface.co/hfl/chinese-roberta-wwm-ext-large/resolve/main/flax_model.msgpack
• !wget -P bert/chinese-roberta-wwm-ext-large/ https://huggingface.co/hfl/chinese-roberta-wwm-ext-large/resolve/main/pytorch_model.bin
• !wget -P bert/chinese-roberta-wwm-ext-large/ https://huggingface.co/hfl/chinese-roberta-wwm-ext-large/resolve/main/tf_model.h5

• !wget -P logs/{datasetname}/ https://huggingface.co/Erythrocyte/bert-vits2_base_model/resolve/main/DUR_0.pth
• !wget -P logs/{datasetname}/ https://huggingface.co/Erythrocyte/bert-vits2_base_model/resolve/main/D_0.pth
• !wget -P logs/{datasetname}/ https://huggingface.co/Erythrocyte/bert-vits2_base_model/resolve/main/G_0.pth

• !python resample.py

• !python preprocess_text.py

• !python bert_gen.py

• !python train_ms.py -m {datasetname} -c configs/config.json

• !python webui.py