blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2026-07-11 01:31:12 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge 180d4a0010992a69e1ccd7c167a73e890fb533d7 into 968952fd2ab3db8ef228a79072f5456fe20bda52

Browse Source
This commit is contained in:
XXXXRT666 2025-06-01 17:21:44 +00:00 committed by GitHub
commit 63e060b63b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
17 changed files with 782 additions and 603 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
.github/build_windows_packages.ps1 vendored
View File

@ -162,7 +162,7 @@ Copy-Item -Path $curr -Destination $pkgName -Recurse
$7zPath = "$pkgName.7z"
$start = Get-Date
Write-Host "Compress Starting at $start"
& "C:\Program Files\7-Zip\7z.exe" a -t7z "$7zPath" "$pkgName" -m0=lzma2 -mx=9 -md=1g -ms=1g -mmc=500 -mfb=273 -mlc=0 -mlp=4 -mpb=4 -mc=8g -mmt=on -bsp1
& "C:\Program Files\7-Zip\7z.exe" a -t7z "$7zPath" "$pkgName" -m0=lzma2 -mx=9 -mmt=on -bsp1
$end = Get-Date
Write-Host "Elapsed time: $($end - $start)"
Get-ChildItem .

1
GPT_SoVITS/inference_webui.py
View File

@ -1295,5 +1295,6 @@ if __name__ == "__main__":
inbrowser=True,
share=is_share,
server_port=infer_ttswebui,
show_api=False,
# quiet=True,
)

1
GPT_SoVITS/inference_webui_fast.py
View File

@ -538,5 +538,6 @@ if __name__ == "__main__":
inbrowser=True,
share=is_share,
server_port=infer_ttswebui,
show_api=False,
# quiet=True,
)

5
GPT_SoVITS/text/g2pw/onnx_api.py
View File

@ -23,8 +23,11 @@ from .utils import load_config
onnxruntime.set_default_logger_severity(3)
try:
onnxruntime.preload_dlls()
except:
except AttributeError:
pass
except Exception as _:
traceback.print_exc()
warnings.filterwarnings("ignore")
model_version = "1.1"

57
README.md
View File

@ -1,3 +1,5 @@
#
<div align="center">
<h1>GPT-SoVITS-WebUI</h1>
@ -7,12 +9,21 @@ A Powerful Few-shot Voice Conversion and Text-to-Speech WebUI.<br><br>
<a href="https://trendshift.io/repositories/7033" target="_blank"><img src="https://trendshift.io/api/badge/repositories/7033" alt="RVC-Boss%2FGPT-SoVITS | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
<!-- img src="https://counter.seku.su/cmoe?name=gptsovits&theme=r34" /><br> -->
[![Train In Colab](https://img.shields.io/badge/Colab-Training-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-WebUI.ipynb)
[![Infer In Colab](https://img.shields.io/badge/Colab-Inference-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-Inference.ipynb)
[![Huggingface](https://img.shields.io/badge/HuggingFace-online%20demo-blue.svg?style=for-the-badge&logo=huggingface)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Open In Colab](https://img.shields.io/badge/Colab-F9AB00?style=for-the-badge&logo=googlecolab&color=525252)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/colab_webui.ipynb)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Huggingface](https://img.shields.io/badge/🤗%20-online%20demo-yellow.svg?style=for-the-badge)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Discord](https://img.shields.io/discord/1198701940511617164?color=%23738ADB&label=Discord&style=for-the-badge)](https://discord.gg/dnrgs5GHfG)
[![简体中文](https://img.shields.io/badge/简体中文-阅读文档-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e)
[![English](https://img.shields.io/badge/English-Read%20Docs-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://rentry.co/GPT-SoVITS-guide#/)
[![Change Log](https://img.shields.io/badge/Change%20Log-View%20Updates-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/docs/en/Changelog_EN.md)
[![GitHub release](https://img.shields.io/github/v/release/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/releases)
[![GitHub Stars](https://img.shields.io/github/stars/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/stargazers)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge&logo=opensourceinitiative)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Python](https://img.shields.io/badge/Python-3.9%2B-blue.svg?style=for-the-badge&logo=python)](https://www.python.org/downloads/)
[![PyTorch](https://img.shields.io/badge/PyTorch-2.5.1+-ee4c2c?style=for-the-badge&logo=pytorch)](https://pytorch.org/)
[![Image Size](https://img.shields.io/docker/image-size/xxxxrt666/gpt-sovits/latest?style=for-the-badge&logo=docker)](https://hub.docker.com/r/xxxxrt666/gpt-sovits)
**English** | [**中文简体**](./docs/cn/README.md) | [**日本語**](./docs/ja/README.md) | [**한국어**](./docs/ko/README.md) | [**Türkçe**](./docs/tr/README.md)
@ -20,7 +31,7 @@ A Powerful Few-shot Voice Conversion and Text-to-Speech WebUI.<br><br>
---
## Features:
## Features
1. **Zero-shot TTS:** Input a 5-second vocal sample and experience instant text-to-speech conversion.
@ -34,13 +45,13 @@ A Powerful Few-shot Voice Conversion and Text-to-Speech WebUI.<br><br>
Unseen speakers few-shot fine-tuning demo:
https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb
<https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb>
**User guide: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)**
<!-- **User guide: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)** -->
## Installation
For users in China, you can [click here](https://www.codewithgpu.com/i/RVC-Boss/GPT-SoVITS/GPT-SoVITS-Official) to use AutoDL Cloud Docker to experience the full functionality online.
For users in China, you can use [AutoDL Cloud Docker](https://www.codewithgpu.com/i/RVC-Boss/GPT-SoVITS/GPT-SoVITS-Official) to experience the full functionality online.
### Tested Environments
@ -171,7 +182,7 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
## Pretrained Models
**If `install.sh` runs successfully, you may skip No.1,2,3**
**If `install.sh` runs successfully, you may skip No.1, 2, 3**
**Users in China can [download all these models here](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e/dkxgpiy9zb96hob4#nVNhX).**
@ -193,10 +204,8 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
The TTS annotation .list file format:
```
```text
vocal_path|speaker_name|language|text
```
Language dictionary:
@ -209,10 +218,8 @@ Language dictionary:
Example:
```
```text
D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
```
## Finetune and inference
@ -222,7 +229,6 @@ D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
#### Integrated Package Users
Double-click `go-webui.bat`or use `go-webui.ps1`
if you want to switch to V1,then double-click`go-webui-v1.bat` or use `go-webui-v1.ps1`
#### Others
@ -230,14 +236,6 @@ if you want to switch to V1,then double-click`go-webui-v1.bat` or use `go-webui-
python webui.py <language(optional)>
```
if you want to switch to V1,then
```bash
python webui.py v1 <language(optional)>
```
Or maunally switch version in WebUI
### Finetune
#### Path Auto-filling is now supported
@ -253,7 +251,7 @@ Or maunally switch version in WebUI
#### Integrated Package Users
Double-click `go-webui-v2.bat` or use `go-webui-v2.ps1` ,then open the inference webui at `1-GPT-SoVITS-TTS/1C-inference`
Double-click `go-webui.bat` or use `go-webui.ps1` , then open the inference webui at `1-GPT-SoVITS-TTS/1C-inference`
#### Others
@ -356,11 +354,6 @@ Use the command line to open the WebUI for UVR5
python tools/uvr5/webui.py "<infer_device>" <is_half> <webui_port_uvr5>
```
<!-- If you can't open a browser, follow the format below for UVR processing,This is using mdxnet for audio processing
```
python mdxnet.py --model --input_root --output_vocal --output_ins --agg_level --format --device --is_half_precision
``` -->
This is how the audio segmentation of the dataset is done using the command line
```bash
@ -435,5 +428,5 @@ Thankful to @Naozumi520 for providing the Cantonese training set and for the gui
## Thanks to all contributors for their efforts
<a href="https://github.com/RVC-Boss/GPT-SoVITS/graphs/contributors" target="_blank">
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" />
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" alt="Contributors"/>
</a>

73
docker_build.ps1 Normal file
View File

@ -0,0 +1,73 @@
$ErrorActionPreference = "Stop"
$ScriptDir = Split-Path -Parent $MyInvocation.MyCommand.Definition
Set-Location $ScriptDir
if (-not (Get-Command "docker" -ErrorAction SilentlyContinue)) {
Write-Host "Docker Not Found"
exit 1
}
$Lite = $false
$CudaVersion = "12.6"
function Write-Help {
Write-Host @"
Usage: powershell -File docker_build.ps1 [OPTIONS]
Options:
--cuda 12.6|12.8 Specify the CUDA VERSION (REQUIRED)
--lite Build a Lite Image
-h, --help Show this help message and exit
Examples:
powershell -File docker_build.ps1 --cuda 12.6 --lite
"@
}
if ($args.Count -eq 0) {
Write-Help
exit 0
}
for ($i = 0; $i -lt $args.Count; $i++) {
switch ($args[$i]) {
'--cuda' {
$i++
$val = $args[$i]
if ($val -ne "12.6" -and $val -ne "12.8") {
Write-Host "Error: Invalid CUDA_VERSION: $val"
Write-Host "Choose From: [12.6, 12.8]"
exit 1
}
$CudaVersion = $val
}
'--lite' {
$Lite = $true
}
'-h' { Write-Help; exit 0 }
'--help' { Write-Help; exit 0 }
default {
Write-Host "Unknown Argument: $($args[$i])"
Write-Host "Use -h or --help to see available options."
exit 1
}
}
}
$arch = (Get-CimInstance Win32_Processor).Architecture
$TargetPlatform = if ($arch -eq 9) { "linux/amd64" } else { "linux/arm64" }
if ($Lite) {
$TorchBase = "lite"
} else {
$TorchBase = "full"
}
docker build `
--build-arg CUDA_VERSION=$CudaVersion `
--build-arg LITE=$Lite `
--build-arg TARGETPLATFORM=$TargetPlatform `
--build-arg TORCH_BASE=$TorchBase `
-t "$env:USERNAME/gpt-sovits:local" `
.

2
docker_build.sh
View File

@ -25,7 +25,7 @@ print_help() {
echo " -h, --help Show this help message and exit"
echo ""
echo "Examples:"
echo " bash docker_build.sh --cuda 12.6 --funasr --faster-whisper"
echo " bash docker_build.sh --cuda 12.6"
}
# Show help if no arguments provided

54
docs/cn/README.md
View File

@ -1,18 +1,29 @@
#
<div align="center">
<h1>GPT-SoVITS-WebUI</h1>
强大的少样本语音转换与语音合成Web用户界面.<br><br>
强大的少样本语音转换与语音合成Web用户界面. <br><br>
[![madewithlove](https://img.shields.io/badge/made_with-%E2%9D%A4-red?style=for-the-badge&labelColor=orange)](https://github.com/RVC-Boss/GPT-SoVITS)
<a href="https://trendshift.io/repositories/7033" target="_blank"><img src="https://trendshift.io/api/badge/repositories/7033" alt="RVC-Boss%2FGPT-SoVITS | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
<!-- img src="https://counter.seku.su/cmoe?name=gptsovits&theme=r34" /><br> -->
[![Train In Colab](https://img.shields.io/badge/Colab-Training-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-WebUI.ipynb)
[![Infer In Colab](https://img.shields.io/badge/Colab-Inference-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-Inference.ipynb)
[![Huggingface](https://img.shields.io/badge/HuggingFace-online%20demo-blue.svg?style=for-the-badge&logo=huggingface)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Open In Colab](https://img.shields.io/badge/Colab-F9AB00?style=for-the-badge&logo=googlecolab&color=525252)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/colab_webui.ipynb)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Huggingface](https://img.shields.io/badge/🤗%20-online%20demo-yellow.svg?style=for-the-badge)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Discord](https://img.shields.io/discord/1198701940511617164?color=%23738ADB&label=Discord&style=for-the-badge)](https://discord.gg/dnrgs5GHfG)
[![简体中文](https://img.shields.io/badge/简体中文-阅读文档-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e)
[![English](https://img.shields.io/badge/English-Read%20Docs-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://rentry.co/GPT-SoVITS-guide#/)
[![Change Log](https://img.shields.io/badge/更新日志-查看更新-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/docs/cn/Changelog_CN.md)
[![GitHub release](https://img.shields.io/github/v/release/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/releases)
[![GitHub Stars](https://img.shields.io/github/stars/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/stargazers)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge&logo=opensourceinitiative)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Python](https://img.shields.io/badge/Python-3.9%2B-blue.svg?style=for-the-badge&logo=python)](https://www.python.org/downloads/)
[![PyTorch](https://img.shields.io/badge/PyTorch-2.5.1+-ee4c2c?style=for-the-badge&logo=pytorch)](https://pytorch.org/)
[![Image Size](https://img.shields.io/docker/image-size/xxxxrt666/gpt-sovits/latest?style=for-the-badge&logo=docker)](https://hub.docker.com/r/xxxxrt666/gpt-sovits)
[**English**](../../README.md) | **中文简体** | [**日本語**](../ja/README.md) | [**한국어**](../ko/README.md) | [**Türkçe**](../tr/README.md)
@ -36,7 +47,7 @@
<https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb>
**用户手册: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)**
<!-- **用户手册: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)** -->
## 安装
@ -171,7 +182,7 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
## 预训练模型
**若成功运行`install.sh`可跳过 No.1,2,3**
**若成功运行`install.sh`可跳过 No.1, 2, 3**
**中国地区的用户可以[在此处下载这些模型](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e/dkxgpiy9zb96hob4#nVNhX).**
@ -181,19 +192,19 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
3. 对于 UVR5 (人声/伴奏分离和混响移除, 额外功能), 从 [UVR5 Weights](https://huggingface.co/lj1995/VoiceConversionWebUI/tree/main/uvr5_weights) 下载模型, 并将其放置在 `tools/uvr5/uvr5_weights` 目录中.
- 如果你在 UVR5 中使用 `bs_roformer` 或 `mel_band_roformer`模型, 你可以手动下载模型和相应的配置文件, 并将它们放在 `tools/UVR5/UVR5_weights` 中.**重命名模型文件和配置文件, 确保除后缀外**, 模型和配置文件具有相同且对应的名称.此外, 模型和配置文件名**必须包含"roformer"**, 才能被识别为 roformer 类的模型.
- 如果你在 UVR5 中使用 `bs_roformer` 或 `mel_band_roformer`模型, 你可以手动下载模型和相应的配置文件, 并将它们放在 `tools/UVR5/UVR5_weights` 中. **重命名模型文件和配置文件, 确保除后缀外**, 模型和配置文件具有相同且对应的名称. 此外, 模型和配置文件名**必须包含"roformer"**, 才能被识别为 roformer 类的模型.
- 建议在模型名称和配置文件名中**直接指定模型类型**, 例如`mel_mand_roformer`、`bs_roformer`.如果未指定, 将从配置文中比对特征, 以确定它是哪种类型的模型.例如, 模型`bs_roformer_ep_368_sdr_12.9628.ckpt` 和对应的配置文件`bs_roformer_ep_368_sdr_12.9628.yaml` 是一对.`kim_mel_band_roformer.ckpt` 和 `kim_mel_band_roformer.yaml` 也是一对.
- 建议在模型名称和配置文件名中**直接指定模型类型**, 例如`mel_mand_roformer`、`bs_roformer`. 如果未指定, 将从配置文中比对特征, 以确定它是哪种类型的模型. 例如, 模型`bs_roformer_ep_368_sdr_12.9628.ckpt` 和对应的配置文件`bs_roformer_ep_368_sdr_12.9628.yaml` 是一对.`kim_mel_band_roformer.ckpt` 和 `kim_mel_band_roformer.yaml` 也是一对.
4. 对于中文 ASR (额外功能), 从 [Damo ASR Model](https://modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/files)、[Damo VAD Model](https://modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-pytorch/files) 和 [Damo Punc Model](https://modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/files) 下载模型, 并将它们放置在 `tools/asr/models` 目录中.
5. 对于英语或日语 ASR (额外功能), 从 [Faster Whisper Large V3](https://huggingface.co/Systran/faster-whisper-large-v3) 下载模型, 并将其放置在 `tools/asr/models` 目录中.此外, [其他模型](https://huggingface.co/Systran) 可能具有类似效果且占用更少的磁盘空间.
5. 对于英语或日语 ASR (额外功能), 从 [Faster Whisper Large V3](https://huggingface.co/Systran/faster-whisper-large-v3) 下载模型, 并将其放置在 `tools/asr/models` 目录中. 此外, [其他模型](https://huggingface.co/Systran) 可能具有类似效果且占用更少的磁盘空间.
## 数据集格式
文本到语音 (TTS) 注释 .list 文件格式:
```
```text
vocal_path|speaker_name|language|text
```
@ -207,7 +218,7 @@ vocal_path|speaker_name|language|text
示例:
```
```text
D:\GPT-SoVITS\xxx/xxx.wav|xxx|zh|我爱玩原神.
```
@ -218,7 +229,6 @@ D:\GPT-SoVITS\xxx/xxx.wav|xxx|zh|我爱玩原神.
#### 整合包用户
双击`go-webui.bat`或者使用`go-webui.ps1`
若想使用 V1,则双击`go-webui-v1.bat`或者使用`go-webui-v1.ps1`
#### 其他
@ -226,14 +236,6 @@ D:\GPT-SoVITS\xxx/xxx.wav|xxx|zh|我爱玩原神.
python webui.py <language(optional)>
```
若想使用 V1,则
```bash
python webui.py v1 <language(optional)>
```
或者在 webUI 内动态切换
### 微调
#### 现已支持自动填充路径
@ -243,13 +245,13 @@ python webui.py v1 <language(optional)>
3. 进行降噪(可选)
4. 进行 ASR
5. 校对标注
6. 前往下一个窗口,点击训练
6. 前往下一个窗口, 点击训练
### 打开推理 WebUI
#### 整合包用户
双击 `go-webui.bat` 或者使用 `go-webui.ps1` ,然后在 `1-GPT-SoVITS-TTS/1C-推理` 中打开推理 webUI
双击 `go-webui.bat` 或者使用 `go-webui.ps1` , 然后在 `1-GPT-SoVITS-TTS/1C-推理` 中打开推理 webUI
#### 其他
@ -287,7 +289,7 @@ python webui.py
3. 需要从[huggingface](https://huggingface.co/lj1995/GPT-SoVITS/tree/main/gsv-v2final-pretrained) 下载预训练模型文件放到 GPT_SoVITS/pretrained_models/gsv-v2final-pretrained 下
中文额外需要下载[G2PWModel.zip(HF)](https://huggingface.co/XXXXRT/GPT-SoVITS-Pretrained/resolve/main/G2PWModel.zip)| [G2PWModel.zip(ModelScope)](https://www.modelscope.cn/models/XXXXRT/GPT-SoVITS-Pretrained/resolve/master/G2PWModel.zip) (下载 G2PW 模型,解压并重命名为`G2PWModel`,将其放到`GPT_SoVITS/text`目录下)
中文额外需要下载[G2PWModel.zip(HF)](https://huggingface.co/XXXXRT/GPT-SoVITS-Pretrained/resolve/main/G2PWModel.zip)| [G2PWModel.zip(ModelScope)](https://www.modelscope.cn/models/XXXXRT/GPT-SoVITS-Pretrained/resolve/master/G2PWModel.zip) (下载 G2PW 模型, 解压并重命名为`G2PWModel`, 将其放到`GPT_SoVITS/text`目录下)
## V3 更新说明
@ -416,5 +418,5 @@ python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p
## 感谢所有贡献者的努力
<a href="https://github.com/RVC-Boss/GPT-SoVITS/graphs/contributors" target="_blank">
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" />
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" alt="Contributors"/>
</a>

61
docs/ja/README.md
View File

@ -1,16 +1,27 @@
#
<div align="center">
<h1>GPT-SoVITS-WebUI</h1>
パワフルなFew-Shot音声変換・音声合成 WebUI.<br><br>
パワフルなFew-Shot音声変換・音声合成 WebUI. <br><br>
[![madewithlove](https://img.shields.io/badge/made_with-%E2%9D%A4-red?style=for-the-badge&labelColor=orange)](https://github.com/RVC-Boss/GPT-SoVITS)
<img src="https://counter.seku.su/cmoe?name=gptsovits&theme=r34" /><br>
[![Train In Colab](https://img.shields.io/badge/Colab-Training-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-WebUI.ipynb)
[![Infer In Colab](https://img.shields.io/badge/Colab-Inference-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-Inference.ipynb)
[![Huggingface](https://img.shields.io/badge/HuggingFace-online%20demo-blue.svg?style=for-the-badge&logo=huggingface)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Open In Colab](https://img.shields.io/badge/Colab-F9AB00?style=for-the-badge&logo=googlecolab&color=525252)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/colab_webui.ipynb)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Huggingface](https://img.shields.io/badge/🤗%20-online%20demo-yellow.svg?style=for-the-badge)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Discord](https://img.shields.io/discord/1198701940511617164?color=%23738ADB&label=Discord&style=for-the-badge)](https://discord.gg/dnrgs5GHfG)
[![简体中文](https://img.shields.io/badge/简体中文-阅读文档-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e)
[![English](https://img.shields.io/badge/English-Read%20Docs-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://rentry.co/GPT-SoVITS-guide#/)
[![Change Log](https://img.shields.io/badge/Change%20Log-View%20Updates-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/docs/ja/Changelog_JA.md)
[![GitHub release](https://img.shields.io/github/v/release/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/releases)
[![GitHub Stars](https://img.shields.io/github/stars/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/stargazers)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge&logo=opensourceinitiative)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Python](https://img.shields.io/badge/Python-3.9%2B-blue.svg?style=for-the-badge&logo=python)](https://www.python.org/downloads/)
[![PyTorch](https://img.shields.io/badge/PyTorch-2.5.1+-ee4c2c?style=for-the-badge&logo=pytorch)](https://pytorch.org/)
[![Image Size](https://img.shields.io/docker/image-size/xxxxrt666/gpt-sovits/latest?style=for-the-badge&logo=docker)](https://hub.docker.com/r/xxxxrt666/gpt-sovits)
[**English**](../../README.md) | [**中文简体**](../cn/README.md) | **日本語** | [**한국어**](../ko/README.md) | [**Türkçe**](../tr/README.md)
@ -18,7 +29,7 @@
---
## 機能:
## 機能
1. **Zero-Shot TTS:** たった 5 秒間の音声サンプルで、即座にテキストからその音声に変換できます.
@ -32,9 +43,9 @@
声の事前学習無しかつ Few-Shot でトレーニングされたモデルのデモ:
https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb
<https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb>
**ユーザーマニュアル: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)**
<!-- **ユーザーマニュアル: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)** -->
## インストール
@ -165,7 +176,7 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
## 事前訓練済みモデル
**`install.sh`が正常に実行された場合、No.1,2,3 はスキップしてかまいません.**
**`install.sh`が正常に実行された場合、No.1, 2, 3 はスキップしてかまいません.**
1. [GPT-SoVITS Models](https://huggingface.co/lj1995/GPT-SoVITS) から事前訓練済みモデルをダウンロードし、`GPT_SoVITS/pretrained_models` ディレクトリに配置してください.
@ -173,19 +184,19 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
3. UVR5 (ボーカル/伴奏 (BGM 等) 分離 & リバーブ除去の追加機能) の場合は、[UVR5 Weights](https://huggingface.co/lj1995/VoiceConversionWebUI/tree/main/uvr5_weights) からモデルをダウンロードし、`tools/uvr5/uvr5_weights` ディレクトリに配置してください.
- UVR5 で bs_roformer または mel_band_roformer モデルを使用する場合、モデルと対応する設定ファイルを手動でダウンロードし、`tools/UVR5/UVR5_weights`フォルダに配置することができます.**モデルファイルと設定ファイルの名前は、拡張子を除いて同じであることを確認してください**.さらに、モデルと設定ファイルの名前には**「roformer」が含まれている必要があります**.これにより、roformer クラスのモデルとして認識されます.
- UVR5 で bs_roformer または mel_band_roformer モデルを使用する場合、モデルと対応する設定ファイルを手動でダウンロードし、`tools/UVR5/UVR5_weights`フォルダに配置することができます. **モデルファイルと設定ファイルの名前は、拡張子を除いて同じであることを確認してください**. さらに、モデルと設定ファイルの名前には**「roformer」が含まれている必要があります**. これにより、roformer クラスのモデルとして認識されます.
- モデル名と設定ファイル名には、**直接モデルタイプを指定することをお勧めします**.例: mel_mand_roformer、bs_roformer.指定しない場合、設定文から特徴を照合して、モデルの種類を特定します.例えば、モデル`bs_roformer_ep_368_sdr_12.9628.ckpt`と対応する設定ファイル`bs_roformer_ep_368_sdr_12.9628.yaml`はペアです.同様に、`kim_mel_band_roformer.ckpt`と`kim_mel_band_roformer.yaml`もペアです.
- モデル名と設定ファイル名には、**直接モデルタイプを指定することをお勧めします**. 例: mel_mand_roformer、bs_roformer. 指定しない場合、設定文から特徴を照合して、モデルの種類を特定します. 例えば、モデル`bs_roformer_ep_368_sdr_12.9628.ckpt`と対応する設定ファイル`bs_roformer_ep_368_sdr_12.9628.yaml`はペアです. 同様に、`kim_mel_band_roformer.ckpt`と`kim_mel_band_roformer.yaml`もペアです.
4. 中国語 ASR (追加機能) の場合は、[Damo ASR Model](https://modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/files)、[Damo VAD Model](https://modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-pytorch/files)、および [Damo Punc Model](https://modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/files) からモデルをダウンロードし、`tools/asr/models` ディレクトリに配置してください.
5. 英語または日本語の ASR (追加機能) を使用する場合は、[Faster Whisper Large V3](https://huggingface.co/Systran/faster-whisper-large-v3) からモデルをダウンロードし、`tools/asr/models` ディレクトリに配置してください.また、[他のモデル](https://huggingface.co/Systran) は、より小さいサイズで高クオリティな可能性があります.
5. 英語または日本語の ASR (追加機能) を使用する場合は、[Faster Whisper Large V3](https://huggingface.co/Systran/faster-whisper-large-v3) からモデルをダウンロードし、`tools/asr/models` ディレクトリに配置してください. また、[他のモデル](https://huggingface.co/Systran) は、より小さいサイズで高クオリティな可能性があります.
## データセット形式
TTS アノテーション .list ファイル形式:
```
```text
vocal_path|speaker_name|language|text
```
@ -197,7 +208,7 @@ vocal_path|speaker_name|language|text
例:
```
```text
D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
```
@ -208,7 +219,6 @@ D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
#### 統合パッケージ利用者
`go-webui.bat`をダブルクリックするか、`go-webui.ps1`を使用します.
V1 に切り替えたい場合は、`go-webui-v1.bat`をダブルクリックするか、`go-webui-v1.ps1`を使用してください.
#### その他
@ -216,14 +226,6 @@ V1 に切り替えたい場合は、`go-webui-v1.bat`をダブルクリックす
python webui.py <言語(オプション)>
```
V1 に切り替えたい場合は
```bash
python webui.py v1 <言語(オプション)>
```
または WebUI で手動でバージョンを切り替えてください.
### 微調整
#### パス自動補完のサポート
@ -239,7 +241,7 @@ python webui.py v1 <言語(オプション)>
#### 統合パッケージ利用者
`go-webui-v2.bat`をダブルクリックするか、`go-webui-v2.ps1`を使用して、`1-GPT-SoVITS-TTS/1C-inference`で推論 webui を開きます.
`go-webui.bat`をダブルクリックするか、`go-webui.ps1`を使用して、`1-GPT-SoVITS-TTS/1C-inference`で推論 webui を開きます.
#### その他
@ -327,11 +329,6 @@ v2 環境から v3 を使用する方法:
python tools/uvr5/webui.py "<infer_device>" <is_half> <webui_port_uvr5>
```
<!-- ブラウザを開けない場合は、以下の形式に従って UVR 処理を行ってください.これはオーディオ処理に mdxnet を使用しています.
```
python mdxnet.py --model --input_root --output_vocal --output_ins --agg_level --format --device --is_half_precision
``` -->
コマンド ラインを使用してデータセットのオーディオ セグメンテーションを行う方法は次のとおりです.
```bash
@ -352,7 +349,7 @@ python tools/asr/funasr_asr.py -i <input> -o <output>
ASR 処理は Faster_Whisper を通じて実行されます(中国語を除く ASR マーキング)
(進行状況バーは表示されません.GPU のパフォーマンスにより時間遅延が発生する可能性があります)
(進行状況バーは表示されません. GPU のパフォーマンスにより時間遅延が発生する可能性があります)
```bash
python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p <precision>
@ -406,5 +403,5 @@ python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p
## すべてのコントリビューターに感謝します
<a href="https://github.com/RVC-Boss/GPT-SoVITS/graphs/contributors" target="_blank">
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" />
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" alt="Contributors"/>
</a>

52
docs/ko/README.md
View File

@ -1,3 +1,5 @@
#
<div align="center">
<h1>GPT-SoVITS-WebUI</h1>
@ -5,12 +7,21 @@
[![madewithlove](https://img.shields.io/badge/made_with-%E2%9D%A4-red?style=for-the-badge&labelColor=orange)](https://github.com/RVC-Boss/GPT-SoVITS)
<img src="https://counter.seku.su/cmoe?name=gptsovits&theme=r34" /><br>
[![Train In Colab](https://img.shields.io/badge/Colab-Training-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-WebUI.ipynb)
[![Infer In Colab](https://img.shields.io/badge/Colab-Inference-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-Inference.ipynb)
[![Huggingface](https://img.shields.io/badge/HuggingFace-online%20demo-blue.svg?style=for-the-badge&logo=huggingface)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Open In Colab](https://img.shields.io/badge/Colab-F9AB00?style=for-the-badge&logo=googlecolab&color=525252)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/colab_webui.ipynb)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Huggingface](https://img.shields.io/badge/🤗%20-online%20demo-yellow.svg?style=for-the-badge)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Discord](https://img.shields.io/discord/1198701940511617164?color=%23738ADB&label=Discord&style=for-the-badge)](https://discord.gg/dnrgs5GHfG)
[![简体中文](https://img.shields.io/badge/简体中文-阅读文档-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e)
[![English](https://img.shields.io/badge/English-Read%20Docs-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://rentry.co/GPT-SoVITS-guide#/)
[![Change Log](https://img.shields.io/badge/Change%20Log-View%20Updates-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/docs/ko/Changelog_KO.md)
[![GitHub release](https://img.shields.io/github/v/release/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/releases)
[![GitHub Stars](https://img.shields.io/github/stars/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/stargazers)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge&logo=opensourceinitiative)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Python](https://img.shields.io/badge/Python-3.9%2B-blue.svg?style=for-the-badge&logo=python)](https://www.python.org/downloads/)
[![PyTorch](https://img.shields.io/badge/PyTorch-2.5.1+-ee4c2c?style=for-the-badge&logo=pytorch)](https://pytorch.org/)
[![Image Size](https://img.shields.io/docker/image-size/xxxxrt666/gpt-sovits/latest?style=for-the-badge&logo=docker)](https://hub.docker.com/r/xxxxrt666/gpt-sovits)
[**English**](../../README.md) | [**中文简体**](../cn/README.md) | [**日本語**](../ja/README.md) | **한국어** | [**Türkçe**](../tr/README.md)
@ -18,7 +29,7 @@
---
## 기능:
## 기능
1. **제로샷 텍스트 음성 변환 (TTS):** 5초의 음성 샘플을 입력하면 즉시 텍스트를 음성으로 변환할 수 있습니다.
@ -32,9 +43,9 @@
보지 못한 발화자의 퓨샷(few-shot) 파인튜닝 데모:
https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb
<https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb>
**사용자 설명서: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)**
<!-- **사용자 설명서: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)** -->
## 설치
@ -165,7 +176,7 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
## 사전 학습된 모델
**`install.sh`가 성공적으로 실행되면 No.1,2,3 은 건너뛰어도 됩니다.**
**`install.sh`가 성공적으로 실행되면 No.1, 2, 3 은 건너뛰어도 됩니다.**
1. [GPT-SoVITS Models](https://huggingface.co/lj1995/GPT-SoVITS) 에서 사전 학습된 모델을 다운로드하고, `GPT_SoVITS/pretrained_models` 디렉토리에 배치하세요.
@ -185,7 +196,7 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
텍스트 음성 합성(TTS) 주석 .list 파일 형식:
```
```text
vocal_path|speaker_name|language|text
```
@ -197,7 +208,7 @@ vocal_path|speaker_name|language|text
예시:
```
```text
D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
```
@ -208,7 +219,6 @@ D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
#### 통합 패키지 사용자
`go-webui.bat`을 더블 클릭하거나 `go-webui.ps1`를 사용하십시오.
V1으로 전환하려면, `go-webui-v1.bat`을 더블 클릭하거나 `go-webui-v1.ps1`를 사용하십시오.
#### 기타
@ -216,14 +226,6 @@ V1으로 전환하려면, `go-webui-v1.bat`을 더블 클릭하거나 `go-webui-
python webui.py <언어(옵션)>
```
V1으로 전환하려면,
```bash
python webui.py v1 <언어(옵션)>
```
또는 WebUI에서 수동으로 버전을 전환하십시오.
### 미세 조정
#### 경로 자동 채우기가 지원됩니다
@ -239,7 +241,7 @@ python webui.py v1 <언어(옵션)>
#### 통합 패키지 사용자
`go-webui-v2.bat`을 더블 클릭하거나 `go-webui-v2.ps1`를 사용한 다음 `1-GPT-SoVITS-TTS/1C-inference`에서 추론 webui를 엽니다.
`go-webui.bat`을 더블 클릭하거나 `go-webui.ps1`를 사용한 다음 `1-GPT-SoVITS-TTS/1C-inference`에서 추론 webui를 엽니다.
#### 기타
@ -277,13 +279,13 @@ V1 환경에서 V2를 사용하려면:
3. [huggingface](https://huggingface.co/lj1995/GPT-SoVITS/tree/main/gsv-v2final-pretrained)에서 V2 사전 학습 모델을 다운로드하여 `GPT_SoVITS/pretrained_models/gsv-v2final-pretrained`에 넣으십시오.
중국어 V2 추가: [G2PWModel.zip(HF)](https://huggingface.co/XXXXRT/GPT-SoVITS-Pretrained/resolve/main/G2PWModel.zip)| [G2PWModel.zip(ModelScope)](https://www.modelscope.cn/models/XXXXRT/GPT-SoVITS-Pretrained/resolve/master/G2PWModel.zip) (G2PW 모델을 다운로드하여 압축을 풀고 `G2PWModel`로 이름을 변경한 다음 `GPT_SoVITS/text`에 배치합니다.)
중국어 V2 추가: [G2PWModel.zip(HF)](https://huggingface.co/XXXXRT/GPT-SoVITS-Pretrained/resolve/main/G2PWModel.zip)| [G2PWModel.zip(ModelScope)](https://www.modelscope.cn/models/XXXXRT/GPT-SoVITS-Pretrained/resolve/master/G2PWModel.zip) (G2PW 모델을 다운로드하여 압축을 풀고 `G2PWModel`로 이름을 변경한 다음 `GPT_SoVITS/text`에 배치합니다)
## V3 릴리스 노트
새로운 기능:
1. 음색 유사성이 더 높아져 목표 음성에 대한 학습 데이터가 적게 필요합니다. (기본 모델을 직접 사용하여 미세 조정 없이 음색 유사성이 크게 향상됩니다.)
1. 음색 유사성이 더 높아져 목표 음성에 대한 학습 데이터가 적게 필요합니다. (기본 모델을 직접 사용하여 미세 조정 없이 음색 유사성이 크게 향상됩니다)
2. GPT 모델이 더 안정적이며 반복 및 생략이 적고, 더 풍부한 감정 표현을 가진 음성을 생성하기가 더 쉽습니다.
@ -404,8 +406,8 @@ python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p
@Naozumi520 님께 감사드립니다. 광둥어 학습 자료를 제공해 주시고, 광둥어 관련 지식을 지도해 주셔서 감사합니다.
## 모든 기여자들에게 감사드립니다 ;)
## 모든 기여자들에게 감사드립니다
<a href="https://github.com/RVC-Boss/GPT-SoVITS/graphs/contributors" target="_blank">
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" />
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" alt="Contributors"/>
</a>

57
docs/tr/README.md
View File

@ -1,3 +1,5 @@
#
<div align="center">
<h1>GPT-SoVITS-WebUI</h1>
@ -7,12 +9,21 @@ Güçlü Birkaç Örnekli Ses Dönüştürme ve Metinden Konuşmaya Web Arayüz
<a href="https://trendshift.io/repositories/7033" target="_blank"><img src="https://trendshift.io/api/badge/repositories/7033" alt="RVC-Boss%2FGPT-SoVITS | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
<!-- img src="https://counter.seku.su/cmoe?name=gptsovits&theme=r34" /><br> -->
[![Train In Colab](https://img.shields.io/badge/Colab-Training-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-WebUI.ipynb)
[![Infer In Colab](https://img.shields.io/badge/Colab-Inference-F9AB00?style=for-the-badge&logo=googlecolab)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/Colab-Inference.ipynb)
[![Huggingface](https://img.shields.io/badge/HuggingFace-online%20demo-blue.svg?style=for-the-badge&logo=huggingface)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Open In Colab](https://img.shields.io/badge/Colab-F9AB00?style=for-the-badge&logo=googlecolab&color=525252)](https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/colab_webui.ipynb)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Huggingface](https://img.shields.io/badge/🤗%20-online%20demo-yellow.svg?style=for-the-badge)](https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2)
[![Discord](https://img.shields.io/discord/1198701940511617164?color=%23738ADB&label=Discord&style=for-the-badge)](https://discord.gg/dnrgs5GHfG)
[![简体中文](https://img.shields.io/badge/简体中文-阅读文档-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e)
[![English](https://img.shields.io/badge/English-Read%20Docs-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://rentry.co/GPT-SoVITS-guide#/)
[![Change Log](https://img.shields.io/badge/Change%20Log-View%20Updates-blue?style=for-the-badge&logo=googledocs&logoColor=white)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/docs/tr/Changelog_TR.md)
[![GitHub release](https://img.shields.io/github/v/release/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/releases)
[![GitHub Stars](https://img.shields.io/github/stars/RVC-Boss/gpt-sovits?style=for-the-badge&logo=github)](https://github.com/RVC-Boss/gpt-sovits/stargazers)
[![License](https://img.shields.io/badge/LICENSE-MIT-green.svg?style=for-the-badge&logo=opensourceinitiative)](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE)
[![Python](https://img.shields.io/badge/Python-3.9%2B-blue.svg?style=for-the-badge&logo=python)](https://www.python.org/downloads/)
[![PyTorch](https://img.shields.io/badge/PyTorch-2.5.1+-ee4c2c?style=for-the-badge&logo=pytorch)](https://pytorch.org/)
[![Image Size](https://img.shields.io/docker/image-size/xxxxrt666/gpt-sovits/latest?style=for-the-badge&logo=docker)](https://hub.docker.com/r/xxxxrt666/gpt-sovits)
[**English**](../../README.md) | [**中文简体**](../cn/README.md) | [**日本語**](../ja/README.md) | [**한국어**](../ko/README.md) | **Türkçe**
@ -20,7 +31,7 @@ Güçlü Birkaç Örnekli Ses Dönüştürme ve Metinden Konuşmaya Web Arayüz
---
## Özellikler:
## Özellikler
1. **Sıfır Örnekli Metinden Konuşmaya:** 5 saniyelik bir vokal örneği girin ve anında metinden konuşmaya dönüşümünü deneyimleyin.
@ -34,9 +45,9 @@ Güçlü Birkaç Örnekli Ses Dönüştürme ve Metinden Konuşmaya Web Arayüz
Görünmeyen konuşmacılar birkaç örnekli ince ayar demosu:
https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb
<https://github.com/RVC-Boss/GPT-SoVITS/assets/129054828/05bee1fa-bdd8-4d85-9350-80c060ab47fb>
**Kullanıcı Kılavuzu: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)**
<!-- **Kullanıcı Kılavuzu: [简体中文](https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e) | [English](https://rentry.co/GPT-SoVITS-guide#/)** -->
## Kurulum
@ -167,7 +178,7 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
## Önceden Eğitilmiş Modeller
**Eğer `install.sh` başarıyla çalıştırılırsa, No.1,2,3 adımını atlayabilirsiniz.**
**Eğer `install.sh` başarıyla çalıştırılırsa, No.1, 2, 3 adımını atlayabilirsiniz.**
1. [GPT-SoVITS Models](https://huggingface.co/lj1995/GPT-SoVITS) üzerinden önceden eğitilmiş modelleri indirip `GPT_SoVITS/pretrained_models` dizinine yerleştirin.
@ -187,7 +198,7 @@ docker exec -it <GPT-SoVITS-CU126-Lite|GPT-SoVITS-CU128-Lite|GPT-SoVITS-CU126|GP
TTS açıklama .list dosya formatı:
```
```text
vocal_path|speaker_name|language|text
```
@ -201,7 +212,7 @@ Dil sözlüğü:
Örnek:
```
```text
D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
```
@ -212,7 +223,6 @@ D:\GPT-SoVITS\xxx/xxx.wav|xxx|en|I like playing Genshin.
#### Entegre Paket Kullanıcıları
`go-webui.bat` dosyasına çift tıklayın veya `go-webui.ps1` kullanın.
V1'e geçmek istiyorsanız, `go-webui-v1.bat` dosyasına çift tıklayın veya `go-webui-v1.ps1` kullanın.
#### Diğerleri
@ -220,14 +230,6 @@ V1'e geçmek istiyorsanız, `go-webui-v1.bat` dosyasına çift tıklayın veya `
python webui.py <dil(isteğe bağlı)>
```
V1'e geçmek istiyorsanız,
```bash
python webui.py v1 <dil(isteğe bağlı)>
```
veya WebUI'de manuel olarak sürüm değiştirin.
### İnce Ayar
#### Yol Otomatik Doldurma artık destekleniyor
@ -243,7 +245,7 @@ veya WebUI'de manuel olarak sürüm değiştirin.
#### Entegre Paket Kullanıcıları
`go-webui-v2.bat` dosyasına çift tıklayın veya `go-webui-v2.ps1` kullanın, ardından çıkarım webui'sini `1-GPT-SoVITS-TTS/1C-inference` adresinde açın.
`go-webui.bat` dosyasına çift tıklayın veya `go-webui.ps1` kullanın, ardından çıkarım webui'sini `1-GPT-SoVITS-TTS/1C-inference` adresinde açın.
#### Diğerleri
@ -281,11 +283,11 @@ V1 ortamından V2'yi kullanmak için:
3. [huggingface](https://huggingface.co/lj1995/GPT-SoVITS/tree/main/gsv-v2final-pretrained) adresinden v2 önceden eğitilmiş modelleri indirin ve bunları `GPT_SoVITS/pretrained_models/gsv-v2final-pretrained` dizinine yerleştirin.
Ek olarak Çince V2: [G2PWModel.zip(HF)](https://huggingface.co/XXXXRT/GPT-SoVITS-Pretrained/resolve/main/G2PWModel.zip)| [G2PWModel.zip(ModelScope)](https://www.modelscope.cn/models/XXXXRT/GPT-SoVITS-Pretrained/resolve/master/G2PWModel.zip) (G2PW modellerini indirip, zipten çıkarıp, `G2PWModel` olarak yeniden adlandırıp `GPT_SoVITS/text` dizinine yerleştirin.)
Ek olarak Çince V2: [G2PWModel.zip(HF)](https://huggingface.co/XXXXRT/GPT-SoVITS-Pretrained/resolve/main/G2PWModel.zip)| [G2PWModel.zip(ModelScope)](https://www.modelscope.cn/models/XXXXRT/GPT-SoVITS-Pretrained/resolve/master/G2PWModel.zip) (G2PW modellerini indirip, zipten çıkarıp, `G2PWModel` olarak yeniden adlandırıp `GPT_SoVITS/text` dizinine yerleştirin)
## V3 Sürüm Notları
### Yeni Özellikler:
### Yeni Özellikler
1. **Tını benzerliği** daha yüksek olup, hedef konuşmacıyı yakınsamak için daha az eğitim verisi gerekmektedir (tını benzerliği, base model doğrudan kullanılacak şekilde fine-tuning yapılmadan önemli ölçüde iyileştirilmiştir).
@ -293,7 +295,7 @@ V1 ortamından V2'yi kullanmak için:
[daha fazla detay](<https://github.com/RVC-Boss/GPT-SoVITS/wiki/GPT%E2%80%90SoVITS%E2%80%90v3%E2%80%90features-(%E6%96%B0%E7%89%B9%E6%80%A7)>)
### v2 ortamında v3 kullanımı:
### v2 ortamında v3 kullanımı
1. `pip install -r requirements.txt` ile bazı paketleri güncelleyin.
@ -331,11 +333,6 @@ UVR5 için Web Arayüzünü açmak için komut satırını kullanın
python tools/uvr5/webui.py "<infer_device>" <is_half> <webui_port_uvr5>
```
<!-- Bir tarayıcı açamıyorsanız, UVR işleme için aşağıdaki formatı izleyin,Bu ses işleme için mdxnet kullanıyor
```
python mdxnet.py --model --input_root --output_vocal --output_ins --agg_level --format --device --is_half_precision
``` -->
Veri setinin ses segmentasyonu komut satırı kullanılarak bu şekilde yapılır
```bash
@ -410,5 +407,5 @@ python ./tools/asr/fasterwhisper_asr.py -i <girdi> -o <çıktı> -l <dil>
## Tüm katkıda bulunanlara çabaları için teşekkürler
<a href="https://github.com/RVC-Boss/GPT-SoVITS/graphs/contributors" target="_blank">
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" />
<img src="https://contrib.rocks/image?repo=RVC-Boss/GPT-SoVITS" alt="Contributors"/>
</a>

5
install.sh
View File

@ -34,8 +34,8 @@ print_help() {
echo " -h, --help Show this help message and exit"
echo ""
echo "Examples:"
echo " bash install.sh --source HF --download-uvr5"
echo " bash install.sh --source ModelScope"
echo " bash install.sh --device CU128 --source HF --download-uvr5"
echo " bash install.sh --device MPS --source ModelScope"
}
# Show help if no arguments provided
@ -149,7 +149,6 @@ else
echo "Installing，Please Wait..."
fi
done
conda install -c conda-forge -q -y
fi
echo "Installing ffmpeg and cmake..."

2
requirements.txt
View File

@ -8,7 +8,7 @@ pytorch-lightning>=2.4
gradio<5
ffmpeg-python
onnxruntime; platform_machine == "aarch64" or platform_machine == "arm64"
onnxruntime-gpu; platform_machine == "x86_64" or platform_machine == "AMD64"
onnxruntime-gpu; platform_machine == "x86_64" or platform_machine == "amd64"
tqdm
funasr==1.0.27
cn2an

544
tools/subfix.py Normal file
View File

@ -0,0 +1,544 @@
import datetime
import os
import threading
import traceback
from dataclasses import dataclass
from functools import partial
from typing import List
import click
import gradio as gr
import librosa
import numpy as np
import soundfile
from gradio.components.audio import WaveformOptions
from tools.i18n.i18n import I18nAuto
PARTIAL_EXIT = partial(os._exit, 0)
LANGUAGE_MAP: dict = {
"ZH": "ZH",
"zh": "ZH",
"JP": "JA",
"jp": "JA",
"JA": "JA",
"ja": "JA",
"EN": "EN",
"en": "EN",
"KO": "KO",
"ko": "KO",
"yue": "YUE",
"YUE": "YUE",
}
LOCK = threading.Lock()
IS_CLI = True
@dataclass
class SubfixErr:
error: Exception
tracebacks: str
class Subfix:
batch_size: int = 2
cur_idx: int = 0
list_path: str
textboxes: List[gr.Textbox] = []
audios: List[gr.Audio] = []
languages: List[gr.Dropdown] = []
selections: List[gr.Checkbox] = []
transcriptions_list: List[List[str]] = []
merge_audio_button: gr.Button
delete_audio_button: gr.Button
previous_index_button1: gr.Button
next_index_button1: gr.Button
previous_index_button2: gr.Button
next_index_button2: gr.Button
index_slider: gr.Slider
batch_size_slider: gr.Slider
close_button: gr.Button
def __init__(self, i18n: I18nAuto):
self.i18n = i18n
with gr.Row(equal_height=True):
with gr.Column(scale=2, min_width=160):
self.index_slider = gr.Slider(minimum=0, maximum=1, step=1, label=i18n("音频索引"))
with gr.Column(scale=1, min_width=160):
self.previous_index_button1 = gr.Button(value=i18n("上一页"), elem_id="btn_previous")
with gr.Column(scale=1, min_width=160):
self.next_index_button1 = gr.Button(value=i18n("下一页"), elem_id="btn_next")
with gr.Row(equal_height=True):
with gr.Column(scale=2, min_width=160):
self.batch_size_slider = gr.Slider(
minimum=4, maximum=20, step=2, value=self.batch_size, label=i18n("每页音频条数")
)
with gr.Column(scale=1, min_width=160):
self.merge_audio_button = gr.Button(value=i18n("合并选中音频"))
with gr.Column(scale=1, min_width=160):
self.delete_audio_button = gr.Button(value=i18n("删除选中音频"))
gr.render(
inputs=[self.index_slider, self.batch_size_slider],
triggers=[self.batch_size_slider.change],
)(self._render_text_area)
@property
def max_index(self):
return len(self.transcriptions_list) - 1
def load_list(self, list_path: str):
with open(list_path, mode="r", encoding="utf-8") as f:
list_data = f.readlines()
for idx, transcriptions in enumerate(list_data):
data = transcriptions.split("|")
if len(data) != 4:
print(f"Error Line {idx + 1}: {'|'.join(data)}")
continue
audio_name, audio_folder, text_language, text = data
self.transcriptions_list.append(
[
audio_name,
audio_folder,
LANGUAGE_MAP.get(text_language.upper(), text_language.upper()),
text.strip("\n").strip(),
]
)
self.list_path = list_path
def save_list(self):
data = []
for transcriptions in self.transcriptions_list:
data.append("|".join(transcriptions))
try:
with open(self.list_path, mode="w", encoding="utf-8") as f:
f.write("\n".join(data))
except Exception as e:
return SubfixErr(e, traceback.format_exc())
def change_index(self, index: int):
audios = []
texts = []
languages = []
checkboxs = []
with LOCK:
for i in range(index, index + self.batch_size):
if i <= self.max_index:
audios.append(gr.Audio(value=self.transcriptions_list[i][0]))
texts.append(gr.Textbox(value=self.transcriptions_list[i][3], label=self.i18n("Text") + f" {i}"))
languages.append(gr.Dropdown(value=self.transcriptions_list[i][2]))
else:
audios.append(gr.Audio(value=None, interactive=False))
texts.append(gr.Textbox(value=None, label=self.i18n("Text") + f" {i}", interactive=False))
languages.append(gr.Dropdown(value=None, interactive=False))
checkboxs = [gr.Checkbox(False) for i in range(self.batch_size)]
self.cur_idx = index
return *audios, *texts, *languages, *checkboxs
def next_page(self, index: int):
batch_size = self.batch_size
max_index = max(self.max_index - batch_size + 1, 0)
index = min(index + batch_size, max_index)
return gr.Slider(value=index), *self.change_index(index)
def previous_page(self, index: int):
batch_size = self.batch_size
index = max(index - batch_size, 0)
return gr.Slider(value=index), *self.change_index(index)
def delete_audio(self, index, *selected):
delete_index = [i + index for i, _ in enumerate(selected) if _]
delete_index = [i for i in delete_index if i < self.max_index]
for idx in delete_index[::-1]:
self.transcriptions_list.pop(idx)
self.save_list()
return gr.Slider(value=index, maximum=self.max_index), *self.change_index(index)
def submit(self, *input):
with LOCK:
index = self.cur_idx
batch_size = self.batch_size
texts = input[: len(input) // 2]
languages = input[len(input) // 2 :]
if texts is None or languages is None:
raise ValueError()
print(index, min(index + batch_size, self.max_index))
for idx in range(index, min(index + batch_size, self.max_index + 1)):
self.transcriptions_list[idx][3] = texts[idx - index].strip().strip("\n")
self.transcriptions_list[idx][2] = languages[idx - index]
result = self.save_list()
if isinstance(result, SubfixErr):
gr.Warning(str(result.error))
print(result.tracebacks)
def merge_audio(self, index, *selected):
batch_size = self.batch_size
merge_index = [i + index for i, _ in enumerate(selected) if _]
merge_index = [i for i in merge_index if i < self.max_index]
if len(merge_index) < 2:
return *(gr.skip() for _ in range(batch_size * 3 + 1)), *(gr.Checkbox(False) for _ in range(batch_size))
else:
merge_texts = []
merge_audios = []
first_itm_index = merge_index[0]
first_itm_path = f"{os.path.splitext(self.transcriptions_list[first_itm_index][0])[0]}_{str(datetime.datetime.now().strftime(r'%Y%m%d_%H%M%S'))}.wav"
final_audio_list = []
for idx in merge_index:
merge_texts.append(self.transcriptions_list[idx][3])
merge_audios.append(self.transcriptions_list[idx][0])
for idx in merge_index[:0:-1]:
self.transcriptions_list.pop(idx)
for audio_path in merge_audios:
final_audio_list.append(librosa.load(audio_path, sr=32000, mono=True)[0])
final_audio_list.append(np.zeros(int(32000 * 0.3)))
final_audio_list.pop()
final_audio = np.concatenate(final_audio_list)
soundfile.write(first_itm_path, final_audio, 32000)
self.transcriptions_list[first_itm_index][0] = first_itm_path
self.transcriptions_list[first_itm_index][3] = ",".join(merge_texts)
return gr.Slider(maximum=self.max_index), *self.change_index(index)
def _render_text_area(self, index, batch_size):
i18n = self.i18n
self.textboxes = []
self.audios = []
self.languages = []
self.selections = []
self.batch_size = batch_size
for i in range(index, index + batch_size):
with gr.Row(equal_height=True):
if i <= self.max_index:
with gr.Column(scale=2, min_width=160):
textbox_tmp = gr.Textbox(
value=self.transcriptions_list[i][3],
label=i18n("Text") + f" {i}",
lines=2,
max_lines=3,
interactive=True,
)
with gr.Column(scale=1, min_width=160):
audio_tmp = gr.Audio(
value=self.transcriptions_list[i][0],
show_label=False,
show_download_button=False,
editable=False,
waveform_options={"show_recording_waveform": False, "show_controls": False},
)
with gr.Column(scale=1, min_width=160):
with gr.Group():
with gr.Row():
language_tmp = gr.Dropdown(
choices=["ZH", "EN", "JA", "KO", "YUE"],
value=self.transcriptions_list[i][2],
allow_custom_value=True,
label=i18n("文本语言"),
interactive=True,
)
with gr.Row():
selection_tmp = gr.Checkbox(
label=i18n("选择音频"),
)
else:
with gr.Column(scale=2, min_width=160):
textbox_tmp = gr.Textbox(
label=i18n("Text") + f" {i}",
lines=2,
max_lines=3,
elem_id="subfix_textbox",
interactive=False,
)
with gr.Column(scale=1, min_width=160):
audio_tmp = gr.Audio(
streaming=True,
show_label=False,
show_download_button=False,
interactive=False,
waveform_options=WaveformOptions(show_recording_waveform=False, show_controls=False),
)
with gr.Column(scale=1, min_width=160):
with gr.Group():
with gr.Row():
language_tmp = gr.Dropdown(
choices=["ZH", "EN", "JA", "KO", "YUE"],
value=None,
allow_custom_value=True,
label=i18n("文本语言"),
interactive=False,
)
with gr.Row():
selection_tmp = gr.Checkbox(
label=i18n("选择音频"),
interactive=False,
)
self.textboxes.append(textbox_tmp)
self.audios.append(audio_tmp)
self.languages.append(language_tmp)
self.selections.append(selection_tmp)
with gr.Row(equal_height=True):
with gr.Column(scale=2, min_width=160):
self.close_button = gr.Button(value=i18n("保存并关闭打标WebUI"), variant="stop")
with gr.Column(scale=1, min_width=160):
self.previous_index_button2 = gr.Button(value=i18n("上一页"))
with gr.Column(scale=1, min_width=160):
self.next_index_button2 = gr.Button(value=i18n("下一页"))
# Event Trigger Binding
self.index_slider.release( # Change Index Button
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
).success(
fn=self.change_index,
inputs=[
self.index_slider,
],
outputs=[
*self.audios,
*self.textboxes,
*self.languages,
*self.selections,
],
max_batch_size=1,
trigger_mode="once",
)
self.next_index_button1.click( # Next Page Button on the Top
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
).success(
fn=self.next_page,
inputs=[
self.index_slider,
],
outputs=[
self.index_slider,
*self.audios,
*self.textboxes,
*self.languages,
*self.selections,
],
scroll_to_output=True,
trigger_mode="once",
)
self.next_index_button2.click( # Next Page Button on the Bottom, Binding to Next Page Button on the Top
lambda: None,
[],
[],
js="""
() => {
document.getElementById("btn_next").click();
}""",
trigger_mode="once",
)
self.previous_index_button1.click( # Previous Page Button on the Top
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
).success(
fn=self.previous_page,
inputs=[
self.index_slider,
],
outputs=[
self.index_slider,
*self.audios,
*self.textboxes,
*self.languages,
*self.selections,
],
scroll_to_output=True,
trigger_mode="once",
)
self.previous_index_button2.click( # Previous Page Button on the Bottom, Binding to Previous Page Button on the Top
lambda: None,
[],
[],
js="""
() => {
document.getElementById("btn_previous").click();
}""",
trigger_mode="once",
)
self.delete_audio_button.click( # Delete the Audio in the Transcription File
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
).success(
fn=self.delete_audio,
inputs=[
self.index_slider,
*self.selections,
],
outputs=[
self.index_slider,
*self.audios,
*self.textboxes,
*self.languages,
*self.selections,
],
scroll_to_output=True,
).success(
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
show_progress="hidden",
)
self.merge_audio_button.click( # Delete the Audio in the Transcription File
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
).success(
fn=self.merge_audio,
inputs=[
self.index_slider,
*self.selections,
],
outputs=[
self.index_slider,
*self.audios,
*self.textboxes,
*self.languages,
*self.selections,
],
scroll_to_output=True,
).success(
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
show_progress="hidden",
)
if not IS_CLI:
self.close_button.click( # Close the Subfix Tab, Binding to Close Button on Audio Processing Tab
fn=lambda: None,
inputs=[],
outputs=[],
js="""
() => {
document.getElementById("btn_close").click();
}""",
trigger_mode="once",
)
else:
self.close_button.click( # Close the Subfix Tab, Binding to Close Button on Audio Processing Tab
fn=self.submit,
inputs=[
*self.textboxes,
*self.languages,
],
outputs=[],
trigger_mode="once",
).then(
fn=PARTIAL_EXIT,
inputs=[],
outputs=[],
)
def render(self, list_path: str, batch_size: int = 10):
self.batch_size = batch_size
self.transcriptions_list = []
self.load_list(list_path=list_path)
@click.command(name="subfix")
@click.argument(
"list-path",
metavar="<Path>",
type=click.Path(exists=True, dir_okay=False, readable=True, writable=True),
required=True,
)
@click.option(
"--i18n-lang",
type=str,
default="Auto",
help="Languages for internationalisation",
show_default=True,
)
@click.option(
"--port",
type=int,
default="9871",
show_default=True,
)
@click.option(
"--share",
type=bool,
default=False,
show_default=True,
)
def main(list_path: str = "", i18n_lang="Auto", port=9871, share=False):
"""Web-Based audio subtitle editing and multilingual annotation Tool
Accept a transcription list path to launch a Gradio WebUI for text editing
"""
with gr.Blocks(analytics_enabled=False) as app:
subfix = Subfix(I18nAuto(i18n_lang))
subfix.render(list_path=list_path)
if subfix.max_index >= 0:
timer = gr.Timer(0.1)
timer.tick(
fn=lambda: (
gr.Slider(value=0, maximum=subfix.max_index, step=1),
gr.Slider(value=10),
gr.Timer(active=False),
),
inputs=[],
outputs=[
subfix.index_slider,
subfix.batch_size_slider,
timer,
],
)
else:
timer = gr.Timer(2)
timer.tick(
fn=lambda x: (_ for _ in ()).throw(gr.Error("Invalid List")) if x is None else None,
inputs=[],
outputs=[],
)
app.queue().launch(
server_name="0.0.0.0",
inbrowser=True,
share=share,
server_port=port,
quiet=False,
show_api=False,
)
if __name__ == "__main__":
main()

422
tools/subfix_webui.py
View File

@ -1,422 +0,0 @@
import sys
from tools.i18n.i18n import I18nAuto, scan_language_list
language = sys.argv[-1] if sys.argv[-1] in scan_language_list() else "Auto"
i18n = I18nAuto(language=language)
import argparse
import copy
import json
import os
import uuid
try:
import gradio.analytics as analytics
analytics.version_check = lambda: None
except:
...
import gradio as gr
import librosa
import numpy as np
import soundfile
g_json_key_text = ""
g_json_key_path = ""
g_load_file = ""
g_load_format = ""
g_max_json_index = 0
g_index = 0
g_batch = 10
g_text_list = []
g_audio_list = []
g_checkbox_list = []
g_data_json = []
def reload_data(index, batch):
global g_index
g_index = index
global g_batch
g_batch = batch
datas = g_data_json[index : index + batch]
output = []
for d in datas:
output.append({g_json_key_text: d[g_json_key_text], g_json_key_path: d[g_json_key_path]})
return output
def b_change_index(index, batch):
global g_index, g_batch
g_index, g_batch = index, batch
datas = reload_data(index, batch)
output = []
for i, _ in enumerate(datas):
output.append(
# gr.Textbox(
# label=f"Text {i+index}",
# value=_[g_json_key_text]#text
# )
{"__type__": "update", "label": f"Text {i + index}", "value": _[g_json_key_text]}
)
for _ in range(g_batch - len(datas)):
output.append(
# gr.Textbox(
# label=f"Text",
# value=""
# )
{"__type__": "update", "label": "Text", "value": ""}
)
for _ in datas:
output.append(_[g_json_key_path])
for _ in range(g_batch - len(datas)):
output.append(None)
for _ in range(g_batch):
output.append(False)
return output
def b_next_index(index, batch):
b_save_file()
if (index + batch) <= g_max_json_index:
return index + batch, *b_change_index(index + batch, batch)
else:
return index, *b_change_index(index, batch)
def b_previous_index(index, batch):
b_save_file()
if (index - batch) >= 0:
return index - batch, *b_change_index(index - batch, batch)
else:
return 0, *b_change_index(0, batch)
def b_submit_change(*text_list):
global g_data_json
change = False
for i, new_text in enumerate(text_list):
if g_index + i <= g_max_json_index:
new_text = new_text.strip() + " "
if g_data_json[g_index + i][g_json_key_text] != new_text:
g_data_json[g_index + i][g_json_key_text] = new_text
change = True
if change:
b_save_file()
return g_index, *b_change_index(g_index, g_batch)
def b_delete_audio(*checkbox_list):
global g_data_json, g_index, g_max_json_index
b_save_file()
change = False
for i, checkbox in reversed(list(enumerate(checkbox_list))):
if g_index + i < len(g_data_json):
if checkbox == True:
g_data_json.pop(g_index + i)
change = True
g_max_json_index = len(g_data_json) - 1
if g_index > g_max_json_index:
g_index = g_max_json_index
g_index = g_index if g_index >= 0 else 0
if change:
b_save_file()
# return gr.Slider(value=g_index, maximum=(g_max_json_index if g_max_json_index>=0 else 0)), *b_change_index(g_index, g_batch)
return {
"value": g_index,
"__type__": "update",
"maximum": (g_max_json_index if g_max_json_index >= 0 else 0),
}, *b_change_index(g_index, g_batch)
def b_invert_selection(*checkbox_list):
new_list = [not item if item is True else True for item in checkbox_list]
return new_list
def get_next_path(filename):
base_dir = os.path.dirname(filename)
base_name = os.path.splitext(os.path.basename(filename))[0]
for i in range(100):
new_path = os.path.join(base_dir, f"{base_name}_{str(i).zfill(2)}.wav")
if not os.path.exists(new_path):
return new_path
return os.path.join(base_dir, f"{str(uuid.uuid4())}.wav")
def b_audio_split(audio_breakpoint, *checkbox_list):
global g_data_json, g_max_json_index
checked_index = []
for i, checkbox in enumerate(checkbox_list):
if checkbox == True and g_index + i < len(g_data_json):
checked_index.append(g_index + i)
if len(checked_index) == 1:
index = checked_index[0]
audio_json = copy.deepcopy(g_data_json[index])
path = audio_json[g_json_key_path]
data, sample_rate = librosa.load(path, sr=None, mono=True)
audio_maxframe = len(data)
break_frame = int(audio_breakpoint * sample_rate)
if break_frame >= 1 and break_frame < audio_maxframe:
audio_first = data[0:break_frame]
audio_second = data[break_frame:]
nextpath = get_next_path(path)
soundfile.write(nextpath, audio_second, sample_rate)
soundfile.write(path, audio_first, sample_rate)
g_data_json.insert(index + 1, audio_json)
g_data_json[index + 1][g_json_key_path] = nextpath
b_save_file()
g_max_json_index = len(g_data_json) - 1
# return gr.Slider(value=g_index, maximum=g_max_json_index), *b_change_index(g_index, g_batch)
return {"value": g_index, "maximum": g_max_json_index, "__type__": "update"}, *b_change_index(g_index, g_batch)
def b_merge_audio(interval_r, *checkbox_list):
global g_data_json, g_max_json_index
b_save_file()
checked_index = []
audios_path = []
audios_text = []
for i, checkbox in enumerate(checkbox_list):
if checkbox == True and g_index + i < len(g_data_json):
checked_index.append(g_index + i)
if len(checked_index) > 1:
for i in checked_index:
audios_path.append(g_data_json[i][g_json_key_path])
audios_text.append(g_data_json[i][g_json_key_text])
for i in reversed(checked_index[1:]):
g_data_json.pop(i)
base_index = checked_index[0]
base_path = audios_path[0]
g_data_json[base_index][g_json_key_text] = "".join(audios_text)
audio_list = []
l_sample_rate = None
for i, path in enumerate(audios_path):
data, sample_rate = librosa.load(path, sr=l_sample_rate, mono=True)
l_sample_rate = sample_rate
if i > 0:
silence = np.zeros(int(l_sample_rate * interval_r))
audio_list.append(silence)
audio_list.append(data)
audio_concat = np.concatenate(audio_list)
soundfile.write(base_path, audio_concat, l_sample_rate)
b_save_file()
g_max_json_index = len(g_data_json) - 1
# return gr.Slider(value=g_index, maximum=g_max_json_index), *b_change_index(g_index, g_batch)
return {"value": g_index, "maximum": g_max_json_index, "__type__": "update"}, *b_change_index(g_index, g_batch)
def b_save_json():
with open(g_load_file, "w", encoding="utf-8") as file:
for data in g_data_json:
file.write(f"{json.dumps(data, ensure_ascii=False)}\n")
def b_save_list():
with open(g_load_file, "w", encoding="utf-8") as file:
for data in g_data_json:
wav_path = data["wav_path"]
speaker_name = data["speaker_name"]
language = data["language"]
text = data["text"]
file.write(f"{wav_path}|{speaker_name}|{language}|{text}".strip() + "\n")
def b_load_json():
global g_data_json, g_max_json_index
with open(g_load_file, "r", encoding="utf-8") as file:
g_data_json = file.readlines()
g_data_json = [json.loads(line) for line in g_data_json]
g_max_json_index = len(g_data_json) - 1
def b_load_list():
global g_data_json, g_max_json_index
with open(g_load_file, "r", encoding="utf-8") as source:
data_list = source.readlines()
for _ in data_list:
data = _.split("|")
if len(data) == 4:
wav_path, speaker_name, language, text = data
g_data_json.append(
{"wav_path": wav_path, "speaker_name": speaker_name, "language": language, "text": text.strip()}
)
else:
print("error line:", data)
g_max_json_index = len(g_data_json) - 1
def b_save_file():
if g_load_format == "json":
b_save_json()
elif g_load_format == "list":
b_save_list()
def b_load_file():
if g_load_format == "json":
b_load_json()
elif g_load_format == "list":
b_load_list()
def set_global(load_json, load_list, json_key_text, json_key_path, batch):
global g_json_key_text, g_json_key_path, g_load_file, g_load_format, g_batch
g_batch = int(batch)
if load_json != "None":
g_load_format = "json"
g_load_file = load_json
elif load_list != "None":
g_load_format = "list"
g_load_file = load_list
else:
g_load_format = "list"
g_load_file = "demo.list"
g_json_key_text = json_key_text
g_json_key_path = json_key_path
b_load_file()
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="Process some integers.")
parser.add_argument("--load_json", default="None", help="source file, like demo.json")
parser.add_argument("--is_share", default="False", help="whether webui is_share=True")
parser.add_argument("--load_list", default="None", help="source file, like demo.list")
parser.add_argument("--webui_port_subfix", default=9871, help="source file, like demo.list")
parser.add_argument("--json_key_text", default="text", help="the text key name in json, Default: text")
parser.add_argument("--json_key_path", default="wav_path", help="the path key name in json, Default: wav_path")
parser.add_argument("--g_batch", default=10, help="max number g_batch wav to display, Default: 10")
args = parser.parse_args()
set_global(args.load_json, args.load_list, args.json_key_text, args.json_key_path, args.g_batch)
with gr.Blocks(analytics_enabled=False) as demo:
gr.Markdown(
value=i18n("Submit Text: 将当前页所有文本框内容手工保存到内存和文件(翻页前后或者退出标注页面前如果没点这个按钮，你再翻回来就回滚了，白忙活。)")
)
with gr.Row():
btn_change_index = gr.Button("Change Index")
btn_submit_change = gr.Button("Submit Text")
btn_merge_audio = gr.Button("Merge Audio")
btn_delete_audio = gr.Button("Delete Audio")
btn_previous_index = gr.Button("Previous Index")
btn_next_index = gr.Button("Next Index")
with gr.Row():
index_slider = gr.Slider(minimum=0, maximum=g_max_json_index, value=g_index, step=1, label="Index", scale=3)
splitpoint_slider = gr.Slider(
minimum=0, maximum=120.0, value=0, step=0.1, label="Audio Split Point(s)", scale=3
)
btn_audio_split = gr.Button("Split Audio", scale=1)
btn_save_json = gr.Button("Save File", visible=True, scale=1)
btn_invert_selection = gr.Button("Invert Selection", scale=1)
with gr.Row():
with gr.Column():
for _ in range(0, g_batch):
with gr.Row():
text = gr.Textbox(label="Text", visible=True, scale=5)
audio_output = gr.Audio(label="Output Audio", visible=True, scale=5)
audio_check = gr.Checkbox(label="Yes", show_label=True, info="Choose Audio", scale=1)
g_text_list.append(text)
g_audio_list.append(audio_output)
g_checkbox_list.append(audio_check)
with gr.Row():
batchsize_slider = gr.Slider(
minimum=1, maximum=g_batch, value=g_batch, step=1, label="Batch Size", scale=3, interactive=False
)
interval_slider = gr.Slider(minimum=0, maximum=2, value=0, step=0.01, label="Interval", scale=3)
btn_theme_dark = gr.Button("Light Theme", link="?__theme=light", scale=1)
btn_theme_light = gr.Button("Dark Theme", link="?__theme=dark", scale=1)
btn_change_index.click(
b_change_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[*g_text_list, *g_audio_list, *g_checkbox_list],
)
btn_submit_change.click(
b_submit_change,
inputs=[
*g_text_list,
],
outputs=[index_slider, *g_text_list, *g_audio_list, *g_checkbox_list],
)
btn_previous_index.click(
b_previous_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[index_slider, *g_text_list, *g_audio_list, *g_checkbox_list],
)
btn_next_index.click(
b_next_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[index_slider, *g_text_list, *g_audio_list, *g_checkbox_list],
)
btn_delete_audio.click(
b_delete_audio,
inputs=[*g_checkbox_list],
outputs=[index_slider, *g_text_list, *g_audio_list, *g_checkbox_list],
)
btn_merge_audio.click(
b_merge_audio,
inputs=[interval_slider, *g_checkbox_list],
outputs=[index_slider, *g_text_list, *g_audio_list, *g_checkbox_list],
)
btn_audio_split.click(
b_audio_split,
inputs=[splitpoint_slider, *g_checkbox_list],
outputs=[index_slider, *g_text_list, *g_audio_list, *g_checkbox_list],
)
btn_invert_selection.click(b_invert_selection, inputs=[*g_checkbox_list], outputs=[*g_checkbox_list])
btn_save_json.click(b_save_file)
demo.load(
b_change_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[*g_text_list, *g_audio_list, *g_checkbox_list],
)
demo.launch(
server_name="0.0.0.0",
inbrowser=True,
# quiet=True,
share=eval(args.is_share),
server_port=int(args.webui_port_subfix),
)

21
tools/uvr5/webui.py
View File

@ -1,23 +1,22 @@
import logging
import os
import sys
import traceback
import gradio as gr
from tools.i18n.i18n import I18nAuto
from tools.my_utils import clean_path
i18n = I18nAuto()
logger = logging.getLogger(__name__)
import sys
import ffmpeg
import gradio as gr
import torch
from bsroformer import Roformer_Loader
from mdxnet import MDXNetDereverb
from vr import AudioPre, AudioPreDeEcho
from tools.i18n.i18n import I18nAuto
from tools.my_utils import clean_path, load_cudnn
i18n = I18nAuto()
logger = logging.getLogger(__name__)
weight_uvr5_root = "tools/uvr5/uvr5_weights"
uvr5_names = []
for name in os.listdir(weight_uvr5_root):
@ -44,6 +43,7 @@ def html_center(text, label="p"):
def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format0):
infos = []
load_cudnn()
try:
inp_root = clean_path(inp_root)
save_root_vocal = clean_path(save_root_vocal)
@ -220,5 +220,6 @@ app.queue().launch( # concurrency_count=511, max_size=1022
inbrowser=True,
share=is_share,
server_port=webui_port_uvr5,
show_api=False,
# quiet=True,
)

26
webui.py
View File

@ -62,8 +62,11 @@ for site_packages_root in site_packages_roots:
traceback.print_exc()
import shutil
import subprocess
from multiprocessing import cpu_count
from subprocess import Popen
import gradio as gr
from config import (
exp_root,
infer_device,
@ -77,22 +80,11 @@ from config import (
)
from tools import my_utils
from tools.i18n.i18n import I18nAuto, scan_language_list
from tools.my_utils import check_details, check_for_existance
language = sys.argv[-1] if sys.argv[-1] in scan_language_list() else "Auto"
os.environ["language"] = language
i18n = I18nAuto(language=language)
from multiprocessing import cpu_count
from tools.my_utils import check_details, check_for_existance
# os.environ['PYTORCH_ENABLE_MPS_FALLBACK'] = '1' # 当遇到mps不支持的步骤时使用cpu
try:
import gradio.analytics as analytics
analytics.version_check = lambda: None
except:
...
import gradio as gr
n_cpu = cpu_count()
@ -384,12 +376,7 @@ def change_label(path_list):
if p_label is None:
check_for_existance([path_list])
path_list = my_utils.clean_path(path_list)
cmd = '"%s" -s tools/subfix_webui.py --load_list "%s" --webui_port %s --is_share %s' % (
python_exec,
path_list,
webui_port_subfix,
is_share,
)
cmd = f'"{python_exec}" -s tools/subfix.py --i18n-lang {language} --port {webui_port_subfix} --share {is_share} "{path_list}"'
yield (
process_info(process_name_subfix, "opened"),
{"__type__": "update", "visible": False},
@ -1419,7 +1406,7 @@ with gr.Blocks(title="GPT-SoVITS WebUI", analytics_enabled=False) as app:
value=process_info(process_name_slice, "close"), variant="primary", visible=False
)
gr.Markdown(value="0bb-" + i18n("语音降噪工具")+i18n("(不稳定，先别用，可能劣化模型效果！)"))
gr.Markdown(value="0bb-" + i18n("语音降噪工具") + i18n("(不稳定，先别用，可能劣化模型效果！)"))
with gr.Row():
with gr.Column(scale=3):
with gr.Row():
@ -1981,5 +1968,6 @@ with gr.Blocks(title="GPT-SoVITS WebUI", analytics_enabled=False) as app:
inbrowser=True,
share=is_share,
server_port=webui_port_main,
show_api=False,
# quiet=True,
)