Create README_CN.md
Browse files- README_CN.md +52 -0
README_CN.md
ADDED
|
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
tags:
|
| 3 |
+
- DiffSVC
|
| 4 |
+
- pre-trained_model
|
| 5 |
+
- basemodel
|
| 6 |
+
- diff-svc
|
| 7 |
+
license: "gpl"
|
| 8 |
+
datasets:
|
| 9 |
+
- 512rc_50k
|
| 10 |
+
- 512rc_80k
|
| 11 |
+
- 512rc_100k
|
| 12 |
+
---
|
| 13 |
+
[English](./README.md) | **简体中文**
|
| 14 |
+
# DiffSVCBaseModel
|
| 15 |
+
|
| 16 |
+
任何类型的音色都能用的大型底模!
|
| 17 |
+
|
| 18 |
+
## 这玩意咋用?
|
| 19 |
+
|
| 20 |
+
1. 按照你的数据挑选一个zip包,保存到你的电脑上
|
| 21 |
+
|
| 22 |
+
2. 填好你的配置，把数据集丢到```(diffsvc 根目录)/data/raw/{角色名称}/```
|
| 23 |
+
|
| 24 |
+
3. 把底模(仅丢 .ckpt 文件)丢到 ```(diffsvc 根目录)/checkpoints/{角色名称}```
|
| 25 |
+
|
| 26 |
+
4. 向原仓库一样预处理和训练数据集,如果不是从step1开始,那就成功了
|
| 27 |
+
|
| 28 |
+
## 我用了多少数据集?
|
| 29 |
+
|
| 30 |
+
两份开源数据集(opencpop ,m4singer),40多个小时的音频
|
| 31 |
+
|
| 32 |
+
## 下载专区
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
| 版本 | 下载传送门 |
|
| 36 |
+
| -------------- | ---------------------------------- |
|
| 37 |
+
| 384rc,50k_step | [传送](./384rc_50k_step.zip) |
|
| 38 |
+
| 384rc,80k_step | [传送](./384rc_80k_step.zip) |
|
| 39 |
+
| 384rc,100k_step | [传送](./384rc_100k_step.zip) |
|
| 40 |
+
|
| 41 |
+
更多版本还在路上
|
| 42 |
+
|
| 43 |
+
> rc: 就是配置里的 residual_channels
|
| 44 |
+
|
| 45 |
+
## 涉及仓库
|
| 46 |
+
|
| 47 |
+
| 仓库 | 传送门 |
|
| 48 |
+
| --------------- | ---------------------------------------------------- |
|
| 49 |
+
| Diff-SVC | [传送](https://github.com/prophesier/diff-svc) |
|
| 50 |
+
| 44.1KHz声码器 | [传送](https://openvpi.github.io/vocoders) |
|
| 51 |
+
| M4Singer | [传送](https://github.com/M4Singer/M4Singer) |
|
| 52 |
+
| OpenCPOP | [传送](https://github.com/wenet-e2e/opencpop) |
|