blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-04-05 12:38:35 +08:00
Code Issues Packages Projects Releases Wiki Activity
664 Commits 3 Branches 2 Tags
Commit Graph

664 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
ChasonJiang
5dfce9a3f0
[fast_inference] 兼容torch.jit.script (#1489) 
* 放弃了在t2s模型中使用@torch.jit.script，确保pytorch环境之间的兼容性

* 优化@torch.jit.script策略
 2024-08-16 16:55:36 +08:00
ChasonJiang
089636424b
放弃了在t2s模型中使用@torch.jit.script，确保pytorch环境之间的兼容性 (#1488) 2024-08-16 16:26:18 +08:00
ChasonJiang
f5a5f1890f
[fast_inference] 优化batch inference的mask策略 (#1477) 
* 优化了batch inference的mask策略，使音频合成的质量更加稳定；改善了一些代码逻辑。

* 删除无用代码
 2024-08-16 10:49:53 +08:00
undefined
7c43b41e6d
fix: 修复x一x类型分词合并时可能出现的吞字问题; (#1362) 2024-07-30 10:38:33 +08:00
undefined
668631e752
fix: 修复整数+量词克发音不准确问题; (#1339) 2024-07-23 18:40:56 +08:00
XXXXRT666
65d754965e
DS_STORE (#1267) 2024-07-01 23:14:13 +08:00
RVC-Boss
1ff74f001a
Update s1_train.py 2024-06-29 22:54:12 +08:00
SetoKaiba
836bfec1fb
fix #1240 (#1241) 2024-06-27 22:53:53 +08:00
CyberWon
8dd7cfab93
Optimize： 推理完成后自动触发gc回收 (#1221) 
在推理完成之后，使用gc控制器对内存进行回收。
#1218
 2024-06-26 22:26:32 +08:00
bstr9
afbcf4007a
fix: 修复prompt_cache里面没有cache住ref_audio_path，导致每次TTS推理都需要加载一遍reference_audio的bug【首次音频帧从3.2s -> 1.7s】 (#1235) 2024-06-26 22:26:20 +08:00
RVC-Boss
e4b17c40bf
Update webui.py 2024-06-13 16:44:50 +08:00
CyberWon
aee85fd98f
fix: 修复参考音频泄露问题 (#1176) 
* Update TextPreprocessor.py

* fix: 修复参考音频泄露的问题。

1. 之前变量命名有错误
2. 还存在参考音频泄露问题。

* 调整一下判断纯符号所处的函数，更符号逻辑
 2024-06-11 20:45:35 +08:00
XXXXRT666
277b258360
关于标点符号导致参考泄漏的问题 (#1169) 
* punctuation

* update

* update
 2024-06-10 16:18:35 +08:00
RVC-Boss
a3da8e87b5
Add files via upload 2024-06-10 16:14:36 +08:00
RVC-Boss
347dcd7d76
微调未读取bert文件重大bug修复 2024-06-06 16:40:02 +08:00
RVC-Boss
b57f92582f
Delete api.py 2024-06-04 22:33:14 +08:00
XXXXRT666
952c1e0e37
Sync (#1135) 
* Add files via upload (#978)

* Update api.py (#979)

修复SynthesizerTrn kwargs 传递异常。

---------

Co-authored-by: Enes Çakırcalı <31443145+enescakircali@users.noreply.github.com>
Co-authored-by: Hayo <632244510@qq.com>
 2024-05-28 00:06:59 +08:00
XXXXRT666
3e1288bdd6
Sync with main branch (#1134) 
* api接口，修复文本切分符号设定中，中文分号错写为英文分号的问题 (#1001)

* 一些小问题修复 (#1021)

* fix import error. It may happen when calling
 api.py

* Update README.md

* Update gpt-sovits_kaggle.ipynb

* Update gpt-sovits_kaggle.ipynb

* fix path error
delete useless line wraps

* 删除重复的 COPY 指令 (#1073)

* [优化] 1Aa-文本获取 (#1102)

* Filter unsupported languages

* add feedback

* simplify modification

* fix detail

* Update english.py (#1106)

copy but not ref the phones list becoz it will be extend later, if not do so,it will affect the self.cmu dict values.

* Update models.py

* modify freeze_quantizer mode, avoid quantizer's codebook updating (#953)

---------

Co-authored-by: FengQingYunDan <pingdengjia0liu@163.com>
Co-authored-by: Kenn Zhang <breakstring@hotmail.com>
Co-authored-by: 蓝梦实 <36986837+SapphireLab@users.noreply.github.com>
Co-authored-by: lyris <lyris@users.noreply.github.com>
Co-authored-by: hcwu1993 <15855138469@163.com>
 2024-05-27 11:24:24 +08:00
XXXXRT666
8fc1e34f58
name2go fix for fast_inference (#1133) 2024-05-27 00:45:20 +08:00
刘悦
f822b9588f
Update api_v2.py (#1104) 2024-05-24 18:32:44 +08:00
Kevin Zhang
50c3664496
chore: add the ability of lru cache for api v3 to improve the inference speed when exchange model weights (#1058) 
* chore: add the ability of lru cache for api v3 to improve the inference speed when exchange model weights

* chore: Dockerfile start api v3

* chore: api default port from 127.0.0.1 to 0.0.0.0

* chore: make gpu happy when do tts

* chore: rollback Dockerfile

* chore: fix

* chore: fix

---------

Co-authored-by: kevin.zhang <kevin.zhang@cardinfolink.com>
 2024-05-19 17:15:56 +08:00
ChasonJiang
2cafde159c
修复报错 TypeError:'type' object is not subscriptable. (#1087) 2024-05-19 16:40:13 +08:00
ChasonJiang
35e755427d
修复了一些bug (#994) 2024-04-21 20:28:33 +08:00
ChasonJiang
29f22115fb
[fast_inference] 回退策略，减少padding影响，开放选项，同步代码 (#986) 
* Update README

* Optimize-English-G2P

* docs: change akward expression

* docs: update Changelog_KO.md

* Fix CN punc in EN,add 's match

* Adjust normalize and g2p logic

* Update zh_CN.json

* Update README (#827)

Update README.md
Update some outdated file paths and commands

* 修复英文多音字,调整字典热加载,新增姓名匹配 (#869)

* Fix homograph dict

* Add JSON in dict

* Adjust hot dict to hot reload

* Add English name dict

* Adjust get name dict logic

* Make API Great Again (#894)

* Add zh/jp/en mix

* Optimize code readability and formatted output.

* Try OGG streaming

* Add stream mode arg

* Add media type arg

* Add cut punc arg

* Eliminate punc risk

* Update README (#895)

* Update README

* Update README

* update README

* update README

* fix typo s/Licence /License (#904)

* fix reformat cmd (#917)

Co-authored-by: starylan <starylan@outlook.com>

* Update README.md

* Normalize chinese arithmetic operations (#947)

* 改变训练和推理时的mask策略，以修复当batch_size>1时，产生的复读现象

* 同步main分支代码，增加“保持随机”选项

* 在colab中运行colab_webui.ipynb发生的uvr5模型缺失问题 (#968)

在colab中使用git下载uvr5模型时报错：
fatal: destination path 'uvr5_weights' already exists and is not an empty directory.
通过在下载前将原本从本仓库下载的uvr5_weights文件夹删除可以解决问题。

* [ASR] 修复FasterWhisper遍历输入路径失败 (#956)

* remove glob

* rename

* reset mirror pos

* 回退mask策略；
回退pad策略；
在T2SBlock中添加padding_mask，以减少pad的影响；
开放repetition_penalty参数，让用户自行调整重复惩罚的强度；
增加parallel_infer参数，用于开启或关闭并行推理，关闭时与0307版本保持一致；
在webui中增加“保持随机”选项；
同步main分支代码。

* 删除无用注释

---------

Co-authored-by: Lion <drain.daters.0p@icloud.com>
Co-authored-by: RVC-Boss <129054828+RVC-Boss@users.noreply.github.com>
Co-authored-by: KamioRinn <snowsdream@live.com>
Co-authored-by: Pengoose <pengoose_dev@naver.com>
Co-authored-by: Yuan-Man <68322456+Yuan-ManX@users.noreply.github.com>
Co-authored-by: XXXXRT666 <157766680+XXXXRT666@users.noreply.github.com>
Co-authored-by: KamioRinn <63162909+KamioRinn@users.noreply.github.com>
Co-authored-by: Lion-Wu <130235128+Lion-Wu@users.noreply.github.com>
Co-authored-by: digger yu <digger-yu@outlook.com>
Co-authored-by: SapphireLab <36986837+SapphireLab@users.noreply.github.com>
Co-authored-by: starylan <starylan@outlook.com>
Co-authored-by: shadow01a <141255649+shadow01a@users.noreply.github.com>
 2024-04-19 14:35:28 +08:00
ChasonJiang
959269b5ae
改变训练和推理时的mask策略，以修复当batch_size>1时，产生的复读现象 (#966) 2024-04-12 18:00:50 +08:00
箱庭XTer
3706ad1b8b
【fast_inference_】通过简单添加torch.no_grad()修饰器，可能会使速度加快 (#930) 
* 添加with torch.no_grad()，速度快一大截

* 恢复先前缩进

* 恢复make batch的位置

* 改用修饰器

* 去除没必要的增加的空行

---------

Co-authored-by: XTer <xxoy1234@outlook.com>
 2024-04-07 18:36:09 +08:00
ChasonJiang
ec7647e08d
[fast_inference] Make API Great Again And Again (#923) 
* 为fast_inference分支适配新的WebAPI

* modified:   api_v2.py

* 修改文档细节

* modified:   api_v2.py

* 优化了代码逻辑，提升健壮性

* 修改文档细节

* 修改文档细节

* 优化代码逻辑

* 修改注释细节
 2024-04-06 21:07:59 +08:00
SapphireLab
72c0eca0a2
spellcheck (#916) 
Co-authored-by: starylan <starylan@outlook.com>
 2024-04-03 17:42:23 +08:00
RVC-Boss
ed75ecdd6d
Merge pull request #820 from ChasonJiang/main 
[fast_inference] 优化代码结构
 2024-03-19 22:41:57 +08:00
chasonjiang
c3ac108ed4 优化代码结构 2024-03-19 22:29:07 +08:00
chasonjiang
df22a4fc04 Merge branch 'fast_inference_' of https://github.com/RVC-Boss/GPT-SoVITS 2024-03-19 21:55:43 +08:00
RVC-Boss
8fa344b83d
Merge pull request #815 from zkd8907/fast_inference_ 
在推理按钮下方增加推理使用参数
 2024-03-19 21:34:41 +08:00
Bobby
a44d4a8b91 在推理按钮下方增加推理使用参数 2024-03-19 09:36:29 +00:00
RVC-Boss
5106fbc62a
Merge pull request #778 from jmaple12/patch-2 
Update text_segmentation_method.py
 2024-03-17 14:40:11 +08:00
chasonjiang
8b548e1956 Merge branch 'fast_inference_' of https://github.com/RVC-Boss/GPT-SoVITS 2024-03-16 22:00:20 +08:00
RVC-Boss
7a112b804a
Merge pull request #777 from KamioRinn/Fix-chinese-ellipsis-fast 
Fix chinese ellipsis for fast_inference_
 2024-03-16 21:33:57 +08:00
RVC-Boss
c2ac07bc3d
Merge pull request #775 from KamioRinn/Fix-chinese-ellipsis 
Fix chinese ellipsis
 2024-03-16 21:33:56 +08:00
RVC-Boss
099059de93
Merge pull request #758 from ChasonJiang/fast_inference_ 
[fast inference] 推理功能增强和改进
 2024-03-16 21:16:43 +08:00
chasonjiang
864a148d75 缓解了batch_size>1时的复读问题，缓解方法是：在T2S模型中，先对phones进行embedding、对bert_features进行project，再pad到相同长度。 2024-03-16 21:04:49 +08:00
chasonjiang
3c78539c44 增加健壮性 2024-03-15 14:58:25 +08:00
chasonjiang
a2f2a5f4a7 添加了自定义修改随机数种子，方便复现结果。 2024-03-15 14:34:10 +08:00
KamioRinn
fabb00af79 Fix chinese ellipsis 2024-03-15 03:13:09 +08:00
KamioRinn
bcb9914b60 Fix chinese ellipsis 2024-03-15 02:53:14 +08:00
jmaple12
160dfbdd2c
Update text_segmentation_method.py 
第85行，修改cut1函数，如果文本的句数为11句，则原函数会把句子切分为 4/7而不是4/4/3。模型处理长句子容易出现漏字现象。
第137行函数cut5中的punds的取值做出修改：删除重复的";"，新增":；"
 2024-03-14 18:14:42 +08:00
chasonjiang
b8ce03fd1b 增加健壮性，防止在cpu推理时设置半精度报错 2024-03-14 11:24:10 +08:00
chasonjiang
85dea3008b 同步合并main分支到fast_inference_分支 2024-03-14 11:18:39 +08:00
chasonjiang
49eb242b17 Merge branch 'main' of https://github.com/RVC-Boss/GPT-SoVITS into fast_inference_ 2024-03-14 11:18:34 +08:00
chasonjiang
8698c28c90 优化代码 2024-03-14 11:15:09 +08:00
RVC-Boss
c287430ecc
Update models.py 2024-03-14 00:42:15 +08:00
RVC-Boss
3dfbcbcd82
Merge pull request #761 from Lion-Wu/fix/cpu 
Support CPU training, use CPU on macOS
 2024-03-13 23:59:33 +08:00