{"id":112,"date":"2026-01-21T09:47:44","date_gmt":"2026-01-21T01:47:44","guid":{"rendered":"https:\/\/www.starverse-ai.com\/guide\/?p=112"},"modified":"2026-01-21T09:49:25","modified_gmt":"2026-01-21T01:49:25","slug":"gptsovits%e6%80%8e%e4%b9%88%e5%ae%89%e8%a3%85%e9%83%a8%e7%bd%b2ai%e8%af%ad%e9%9f%b3%e7%94%9f%e6%88%90","status":"publish","type":"post","link":"https:\/\/www.starverse-ai.com\/guide\/archives\/112","title":{"rendered":"gptsovits\u600e\u4e48\u5b89\u88c5\u90e8\u7f72(ai\u8bed\u97f3\u751f\u6210)"},"content":{"rendered":"\n

\u60f3\u8981\u5b8c\u6574\u7684 GPT-SoVITS \u5b89\u88c5\u90e8\u7f72\u6d41\u7a0b,\u9700\u8981\u4ece\u73af\u5883\u51c6\u5907\u3001\u5b89\u88c5\u6b65\u9aa4\u3001\u8fd0\u884c\u6d4b\u8bd5\u5230\u5e38\u89c1\u95ee\u9898\uff0c\u4e3a\u4f60\u63d0\u4f9b\u4e00\u4efd\u8be6\u7ec6\u4e14\u9002\u5408\u65b0\u624b\u7684\u64cd\u4f5c\u6307\u5357\u3002<\/p>\n\n\n\n

\u4e00\u3001GPT-SoVITS \u5b89\u88c5\u90e8\u7f72\u5168\u6d41\u7a0b<\/p>\n\n\n\n

1. \u524d\u7f6e\u6761\u4ef6\u8bf4\u660e<\/p>\n\n\n\n

GPT-SoVITS \u662f\u4e00\u6b3e\u7ed3\u5408 GPT \u548c SoVITS \u7684\u8bed\u97f3\u5408\u6210\u5de5\u5177\uff0c\u5bf9\u786c\u4ef6\u6709\u57fa\u7840\u8981\u6c42\uff1a<\/p>\n\n\n\n

\u663e\u5361\uff1a\u5efa\u8bae NVIDIA \u663e\u5361(\u663e\u5b58 \u2265 8GB\uff0c12GB+ \u6700\u4f73)\uff0c\u9700\u652f\u6301 CUDA(\u7b97\u529b \u2265 7.0);\u65e0\u72ec\u663e\u53ef\u4f7f\u7528 CPU \u8fd0\u884c\uff0c\u4f46\u901f\u5ea6\u6781\u6162\uff0c\u4ec5\u7528\u4e8e\u6d4b\u8bd5\u3002<\/p>\n\n\n\n

\u7cfb\u7edf\uff1aWindows 10\/11(\u63a8\u8350)\u3001Linux(Ubuntu 20.04+);macOS \u4ec5\u652f\u6301 CPU \u6a21\u5f0f\u3002<\/p>\n\n\n\n

Python\uff1a\u7248\u672c\u9700\u4e3a 3.10(\u63a8\u8350\uff0c3.9\/3.11 \u4e5f\u53ef\u517c\u5bb9\uff0c\u4f46 3.10 \u7a33\u5b9a\u6027\u6700\u4f73)\u3002<\/p>\n\n\n

\n
\"\"<\/figure>\n<\/div>\n\n\n

2. \u73af\u5883\u51c6\u5907<\/p>\n\n\n\n

(1)\u5b89\u88c5 Python \u4e0e Git<\/p>\n\n\n\n

\u4e0b\u8f7d Python 3.10\uff1a\u4ece Python \u5b98\u7f51 \u4e0b\u8f7d\uff0c\u5b89\u88c5\u65f6\u52fe\u9009\u300cAdd Python 3.10 to PATH\u300d\u3002<\/p>\n\n\n\n

\u4e0b\u8f7d Git\uff1a\u4ece Git \u5b98\u7f51 \u5b89\u88c5\uff0c\u9ed8\u8ba4\u9009\u9879\u5373\u53ef\u3002<\/p>\n\n\n\n

(2)\u5b89\u88c5 CUDA \u4e0e cuDNN(\u663e\u5361\u7528\u6237\u5fc5\u88c5)<\/p>\n\n\n\n

\u67e5\u770b\u663e\u5361\u652f\u6301\u7684 CUDA \u7248\u672c\uff1a\u6253\u5f00 NVIDIA \u63a7\u5236\u9762\u677f \u2192 \u5e2e\u52a9 \u2192 \u7cfb\u7edf\u4fe1\u606f \u2192 \u7ec4\u4ef6\uff0c\u67e5\u770b\u300cNVCUDA.DLL\u300d\u5bf9\u5e94\u7684 CUDA \u7248\u672c(\u5982 12.1)\u3002<\/p>\n\n\n\n

\u4e0b\u8f7d\u5bf9\u5e94\u7248\u672c\u7684 CUDA Toolkit\uff1a\u4ece NVIDIA \u5b98\u7f51 \u4e0b\u8f7d\uff0c\u5b89\u88c5\u65f6\u9009\u62e9\u300c\u81ea\u5b9a\u4e49\u5b89\u88c5\u300d\uff0c\u4ec5\u52fe\u9009\u300cCUDA \u2192 \u6838\u5fc3\u7ec4\u4ef6\u300d\u3002<\/p>\n\n\n\n

\u4e0b\u8f7d cuDNN\uff1a\u9700\u6ce8\u518c NVIDIA \u8d26\u53f7\uff0c\u4e0b\u8f7d\u4e0e CUDA \u7248\u672c\u5339\u914d\u7684 cuDNN\uff0c\u89e3\u538b\u540e\u5c06\u6587\u4ef6\u5939\u5185\u7684 bin\u3001include\u3001lib \u590d\u5236\u5230 CUDA \u5b89\u88c5\u76ee\u5f55(\u5982 C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v12.1)\u3002<\/p>\n\n\n\n

\u661f\u5b87\u667a\u7b97\u5b98\u7f51\u652f\u6301\u514d\u8d39\u90e8\u7f72\u7acb\u9a6c\u5b89\u88c5\uff01\u70b9\u51fb\u4e0b\u65b9\u7acb\u5373\u8bd5\u7528!<\/strong><\/p>\n\n\n

\n
\"\"<\/a><\/figure>\n<\/div>\n\n\n

3. \u5b89\u88c5\u90e8\u7f72\u6b65\u9aa4<\/p>\n\n\n\n

(1)\u514b\u9686\u9879\u76ee\u4ed3\u5e93<\/p>\n\n\n\n

\u6253\u5f00\u547d\u4ee4\u63d0\u793a\u7b26(CMD)\u6216\u7ec8\u7aef\uff0c\u6267\u884c\u4ee5\u4e0b\u547d\u4ee4\uff1a<\/p>\n\n\n\n

bash<\/p>\n\n\n\n

\u8fd0\u884c<\/p>\n\n\n\n

# \u514b\u9686 GPT-SoVITS \u4ed3\u5e93<\/p>\n\n\n\n

git clone https:\/\/github.com\/RVC-Boss\/GPT-SoVITS.git<\/p>\n\n\n\n

# \u8fdb\u5165\u9879\u76ee\u76ee\u5f55<\/p>\n\n\n\n

cd GPT-SoVITS<\/p>\n\n\n\n

(2)\u521b\u5efa\u5e76\u6fc0\u6d3b\u865a\u62df\u73af\u5883(\u63a8\u8350)<\/p>\n\n\n\n

bash<\/p>\n\n\n\n

\u8fd0\u884c<\/p>\n\n\n\n

# \u521b\u5efa\u865a\u62df\u73af\u5883<\/p>\n\n\n\n

python -m venv venv<\/p>\n\n\n\n

# Windows \u6fc0\u6d3b\u865a\u62df\u73af\u5883<\/p>\n\n\n\n

venv\\Scripts\\activate<\/p>\n\n\n\n

# Linux\/macOS \u6fc0\u6d3b\u865a\u62df\u73af\u5883<\/p>\n\n\n\n

source venv\/bin\/activate<\/p>\n\n\n\n

(3)\u5b89\u88c5\u4f9d\u8d56\u5305<\/p>\n\n\n\n

bash<\/p>\n\n\n\n

\u8fd0\u884c<\/p>\n\n\n\n

# \u5347\u7ea7 pip<\/p>\n\n\n\n

pip install –upgrade pip<\/p>\n\n\n\n

# \u5b89\u88c5\u57fa\u7840\u4f9d\u8d56(\u56fd\u5185\u7528\u6237\u5efa\u8bae\u52a0\u6e05\u534e\u6e90)<\/p>\n\n\n\n

pip install -r requirements.txt -i https:\/\/pypi.tuna.tsinghua.edu.cn\/simple<\/p>\n\n\n\n

# \u5b89\u88c5 PyTorch(\u5173\u952e!\u9700\u5339\u914d CUDA \u7248\u672c)<\/p>\n\n\n\n

# CUDA 12.1 \u7248\u672c<\/p>\n\n\n\n

pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 –index-url https:\/\/download.pytorch.org\/whl\/cu121<\/p>\n\n\n\n

# \u65e0\u663e\u5361\/CPU \u7248\u672c<\/p>\n\n\n\n

pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 –index-url https:\/\/download.pytorch.org\/whl\/cpu<\/p>\n\n\n\n

(4)\u4e0b\u8f7d\u9884\u8bad\u7ec3\u6a21\u578b<\/p>\n\n\n\n

GPT-SoVITS \u8fd0\u884c\u9700\u8981\u9884\u8bad\u7ec3\u6a21\u578b\uff0c\u6b65\u9aa4\u5982\u4e0b\uff1a<\/p>\n\n\n\n

\u8fdb\u5165\u9879\u76ee\u76ee\u5f55\u4e0b\u7684 pretrained_models \u6587\u4ef6\u5939(\u65e0\u5219\u65b0\u5efa);<\/p>\n\n\n\n

\u4ece\u9879\u76ee\u5b98\u65b9\u5f00\u6e90\u5730\u5740(\u5982 Hugging Face)\u4e0b\u8f7d GPT \u6a21\u578b\u3001SoVITS \u6a21\u578b\u3001bert \u6a21\u578b\uff0c\u653e\u5165\u8be5\u6587\u4ef6\u5939;<\/p>\n\n\n\n

\u786e\u8ba4\u6a21\u578b\u6587\u4ef6\u547d\u540d\u4e0e\u914d\u7f6e\u6587\u4ef6\u4e00\u81f4(\u9ed8\u8ba4\u914d\u7f6e\u65e0\u9700\u4fee\u6539\uff0c\u53ea\u9700\u6587\u4ef6\u5b58\u5728)\u3002<\/p>\n\n\n\n

(5)\u542f\u52a8 WebUI<\/p>\n\n\n\n

bash<\/p>\n\n\n\n

\u8fd0\u884c<\/p>\n\n\n\n

# \u8fd0\u884c\u542f\u52a8\u811a\u672c<\/p>\n\n\n\n

python webui.py<\/p>\n\n\n\n

\u542f\u52a8\u6210\u529f\u540e\uff0c\u7ec8\u7aef\u4f1a\u663e\u793a\u672c\u5730\u8bbf\u95ee\u5730\u5740(\u5982 http:\/\/127.0.0.1:9880)\uff0c\u6253\u5f00\u6d4f\u89c8\u5668\u8bbf\u95ee\u5373\u53ef\u4f7f\u7528\u3002<\/p>\n\n\n\n

\u4e8c\u3001\u5e38\u89c1\u95ee\u9898\u89e3\u7b54(FAQ)<\/p>\n\n\n\n

1. \u8fd0\u884c webui.py \u63d0\u793a\u300cNo module named xxx\u300d?<\/p>\n\n\n\n

\u539f\u56e0\uff1a\u4f9d\u8d56\u5305\u672a\u5b89\u88c5\u5b8c\u6574\uff0c\u6216\u865a\u62df\u73af\u5883\u672a\u6fc0\u6d3b\u3002<\/p>\n\n\n\n

\u89e3\u51b3\uff1a<\/p>\n\n\n\n

\u786e\u8ba4\u5df2\u6fc0\u6d3b\u865a\u62df\u73af\u5883(\u7ec8\u7aef\u524d\u7f00\u6709 (venv));<\/p>\n\n\n\n

\u91cd\u65b0\u6267\u884c pip install -r requirements.txt\uff0c\u82e5\u4ecd\u62a5\u9519\uff0c\u624b\u52a8\u5b89\u88c5\u7f3a\u5931\u7684\u5305(\u5982 pip install xxx);<\/p>\n\n\n\n

\u56fd\u5185\u7528\u6237\u82e5\u4e0b\u8f7d\u5931\u8d25\uff0c\u66f4\u6362\u963f\u91cc\u4e91 \/ \u8c46\u74e3\u6e90\uff1a-i https:\/\/mirrors.aliyun.com\/pypi\/simple\/\u3002<\/p>\n\n\n\n

2. \u663e\u5361\u53ef\u7528\u4f46\u63d0\u793a\u300cCUDA out of memory\u300d?<\/p>\n\n\n\n

\u539f\u56e0\uff1a\u663e\u5b58\u4e0d\u8db3\uff0c\u6216\u6a21\u578b\u52a0\u8f7d\u65f6\u5360\u7528\u8fc7\u9ad8\u3002<\/p>\n\n\n\n

\u89e3\u51b3\uff1a<\/p>\n\n\n\n

\u5173\u95ed\u5176\u4ed6\u5360\u7528\u663e\u5361\u7684\u7a0b\u5e8f(\u5982\u6d4f\u89c8\u5668\u3001\u6e38\u620f);<\/p>\n\n\n\n

\u4fee\u6539 webui.py \u4e2d\u7684\u53c2\u6570\uff0c\u542f\u7528\u663e\u5b58\u4f18\u5316\uff1a\u6dfb\u52a0 –low-vram \u542f\u52a8\u53c2\u6570(python webui.py –low-vram);<\/p>\n\n\n\n

\u82e5\u663e\u5b58 \u2264 8GB\uff0c\u5efa\u8bae\u4f7f\u7528 CPU \u6a21\u5f0f\u6d4b\u8bd5\uff0c\u6216\u964d\u4f4e\u5408\u6210\u65f6\u7684\u97f3\u9891\u957f\u5ea6\u3002<\/p>\n\n\n\n

3. PyTorch \u5b89\u88c5\u540e\u63d0\u793a\u300cCUDA is not available\u300d?<\/p>\n\n\n\n

\u539f\u56e0\uff1aPyTorch \u7248\u672c\u4e0e CUDA \u4e0d\u5339\u914d\uff0c\u6216\u663e\u5361\u4e0d\u652f\u6301 CUDA\u3002<\/p>\n\n\n\n

\u89e3\u51b3\uff1a<\/p>\n\n\n\n

\u6267\u884c python -c “import torch; print(torch.cuda.is_available())”\uff0c\u82e5\u8f93\u51fa False\uff0c\u91cd\u65b0\u5b89\u88c5\u5bf9\u5e94 CUDA \u7248\u672c\u7684 PyTorch;<\/p>\n\n\n\n

\u786e\u8ba4\u663e\u5361\u9a71\u52a8\u5df2\u66f4\u65b0\u5230\u6700\u65b0\u7248\u672c(NVIDIA \u5b98\u7f51\u4e0b\u8f7d GeForce Experience \u66f4\u65b0);<\/p>\n\n\n\n

\u82e5\u4e3a\u7b14\u8bb0\u672c\u53cc\u663e\u5361\uff0c\u9700\u8bbe\u7f6e Python \u8fd0\u884c\u65f6\u4f7f\u7528\u72ec\u7acb\u663e\u5361\u3002<\/p>\n\n\n\n

4. \u5408\u6210\u97f3\u9891\u65f6\u65e0\u58f0\u97f3 \/ \u97f3\u8d28\u5dee?<\/p>\n\n\n\n

\u539f\u56e0\uff1a\u9884\u8bad\u7ec3\u6a21\u578b\u7f3a\u5931\uff0c\u6216\u8f93\u5165\u6587\u672c \/ \u53c2\u8003\u97f3\u9891\u4e0d\u7b26\u5408\u8981\u6c42\u3002<\/p>\n\n\n\n

\u89e3\u51b3\uff1a<\/p>\n\n\n\n

\u786e\u8ba4 pretrained_models \u6587\u4ef6\u5939\u5185\u6a21\u578b\u5b8c\u6574\uff0c\u65e0\u7f3a\u5931\u6587\u4ef6;<\/p>\n\n\n\n

\u53c2\u8003\u97f3\u9891\u9700\u6ee1\u8db3\uff1a\u65f6\u957f 5-30 \u79d2\u3001\u65e0\u6742\u97f3\u3001\u6e05\u6670\u4eba\u58f0;<\/p>\n\n\n\n

\u8f93\u5165\u6587\u672c\u8bed\u8a00\u4e0e\u6a21\u578b\u5339\u914d(\u5982\u4e2d\u6587\u6a21\u578b\u4ec5\u652f\u6301\u4e2d\u6587\u5408\u6210)\u3002<\/p>\n\n\n\n

5. Linux \u7cfb\u7edf\u542f\u52a8\u540e\u65e0\u6cd5\u8bbf\u95ee WebUI?<\/p>\n\n\n\n

\u539f\u56e0\uff1a\u7aef\u53e3\u672a\u5f00\u653e\uff0c\u6216\u7ed1\u5b9a\u5730\u5740\u4ec5\u9650\u672c\u5730\u3002<\/p>\n\n\n\n

\u89e3\u51b3\uff1a<\/p>\n\n\n\n

\u542f\u52a8\u65f6\u6307\u5b9a\u7ed1\u5b9a\u6240\u6709\u5730\u5740\uff1apython webui.py –listen 0.0.0.0 –port 9880;<\/p>\n\n\n\n

\u68c0\u67e5\u9632\u706b\u5899\u662f\u5426\u653e\u884c 9880 \u7aef\u53e3(Ubuntu\uff1asudo ufw allow 9880)\u3002<\/p>\n\n\n\n

6. \u6a21\u578b\u4e0b\u8f7d\u901f\u5ea6\u6162 \/ \u5931\u8d25?<\/p>\n\n\n\n

\u539f\u56e0\uff1a\u7f51\u7edc\u9650\u5236\uff0c\u6216 Hugging Face \u8bbf\u95ee\u4e0d\u7a33\u5b9a\u3002<\/p>\n\n\n\n

\u89e3\u51b3\uff1a<\/p>\n\n\n\n

\u4f7f\u7528\u56fd\u5185\u955c\u50cf\u7ad9\u4e0b\u8f7d\u6a21\u578b(\u5982 ModelScope);<\/p>\n\n\n\n

\u914d\u7f6e Git \u4ee3\u7406\uff1agit config –global http.proxy http:\/\/127.0.0.1:7890(\u9700\u66ff\u6362\u4e3a\u81ea\u5df1\u7684\u4ee3\u7406\u7aef\u53e3)\u3002<\/p>\n\n\n\n

\u603b\u7ed3:<\/p>\n\n\n\n

\u73af\u5883\u6838\u5fc3\uff1aPython 3.10 + \u5339\u914d\u7248\u672c\u7684 PyTorch(CUDA\/CPU)\u662f\u5b89\u88c5\u6210\u529f\u7684\u57fa\u7840\uff0c\u52a1\u5fc5\u786e\u8ba4\u7248\u672c\u517c\u5bb9;<\/p>\n\n\n\n

\u90e8\u7f72\u6b65\u9aa4\uff1a\u514b\u9686\u4ed3\u5e93 \u2192 \u865a\u62df\u73af\u5883 \u2192 \u5b89\u88c5\u4f9d\u8d56 \u2192 \u4e0b\u8f7d\u6a21\u578b \u2192 \u542f\u52a8 WebUI\uff0c\u6309\u987a\u5e8f\u6267\u884c\u53ef\u907f\u514d 80% \u7684\u95ee\u9898;<\/p>\n\n\n\n

\u5e38\u89c1\u95ee\u9898\uff1a\u663e\u5b58\u4e0d\u8db3\u7528 –low-vram \u53c2\u6570\uff0cCUDA \u4e0d\u53ef\u7528\u68c0\u67e5 PyTorch \u4e0e\u663e\u5361\u9a71\u52a8\uff0c\u4f9d\u8d56\u7f3a\u5931\u91cd\u65b0\u5b89\u88c5 requirements.txt\u3002<\/p>\n\n\n\n

\u6574\u4e2a\u5b89\u88c5\u8fc7\u7a0b\u7684\u6838\u5fc3\u662f\u300c\u7248\u672c\u5339\u914d\u300d\u548c\u300c\u6a21\u578b\u5b8c\u6574\u300d\uff0c\u65b0\u624b\u5efa\u8bae\u5148\u5728 Windows \u7cfb\u7edf\u6d4b\u8bd5\uff0c\u719f\u6089\u6d41\u7a0b\u540e\u518d\u90e8\u7f72\u5230 Linux \u670d\u52a1\u5668\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

\u60f3\u8981\u5b8c\u6574\u7684 GPT-SoVITS \u5b89\u88c5\u90e8\u7f72\u6d41\u7a0b,\u9700\u8981\u4ece\u73af\u5883\u51c6…<\/p>\n","protected":false},"author":2,"featured_media":115,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-112","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-zixun"],"views":68,"_links":{"self":[{"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/posts\/112","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/comments?post=112"}],"version-history":[{"count":2,"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/posts\/112\/revisions"}],"predecessor-version":[{"id":123,"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/posts\/112\/revisions\/123"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/media\/115"}],"wp:attachment":[{"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/media?parent=112"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/categories?post=112"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.starverse-ai.com\/guide\/wp-json\/wp\/v2\/tags?post=112"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}