Merge 9e0c1941dc4e59cde8d67a6e10c932c1d535a54e into ad7df5298bea51273c86c05b5b13f28ed7d9fe16

* Update Colab Infer, Add NLTK Download

* Rename GPT_SoVITS_Inference.ipynb to Colab-Inference.ipynb

* Rename

* Update URL

* .

Colab-Inference.ipynb

@@ -0,0 +1,184 @@
+{
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "# GPT-SoVITS Infer"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## Env Setup (Run Once Only)\n",
+        "## 环境配置, 只需运行一次"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### 1."
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "e9b7iFV3dm1f"
+      },
+      "outputs": [],
+      "source": [
+        "%%writefile /content/setup.sh\n",
+        "set -e\n",
+        "\n",
+        "cd /content\n",
+        "\n",
+        "git clone https://github.com/RVC-Boss/GPT-SoVITS.git\n",
+        "\n",
+        "cd GPT-SoVITS\n",
+        "\n",
+        "mkdir GPT_weights\n",
+        "\n",
+        "mkdir SoVITS_weights\n",
+        "\n",
+        "if conda env list | awk '{print $1}' | grep -Fxq \"GPTSoVITS\"; then\n",
+        "    :\n",
+        "else\n",
+        "    conda create -n GPTSoVITS python=3.10 -y\n",
+        "fi\n",
+        "\n",
+        "source activate GPTSoVITS\n",
+        "\n",
+        "pip install ipykernel\n",
+        "\n",
+        "bash install.sh --source HF"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### 2."
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "cellView": "form",
+        "id": "0NgxXg5sjv7z"
+      },
+      "outputs": [],
+      "source": [
+        "%pip install -q condacolab\n",
+        "import condacolab\n",
+        "condacolab.install_from_url(\"https://repo.anaconda.com/archive/Anaconda3-2024.10-1-Linux-x86_64.sh\")\n",
+        "!cd /content && bash setup.sh"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "# Download Model"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### Download From HuggingFace"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "cellView": "form",
+        "id": "vbZY-LnM0tzq"
+      },
+      "outputs": [],
+      "source": [
+        "# Modify These\n",
+        "USER_ID = \"AkitoP\"\n",
+        "REPO_NAME = \"GPT-SoVITS-v2-aegi\"\n",
+        "BRANCH = \"main\"\n",
+        "GPT_PATH = \"new_aegigoe-e100.ckpt\"\n",
+        "SOVITS_PATH = \"new_aegigoe_e60_s32220.pth\"\n",
+        "\n",
+        "# Do Not Modify\n",
+        "HF_BASE = \"https://huggingface.co\"\n",
+        "REPO_ID = f\"{USER_ID}/{REPO_NAME}\"\n",
+        "GPT_URL = f\"{HF_BASE}/{REPO_ID}/blob/{BRANCH}/{GPT_PATH}\"\n",
+        "SOVITS_URL = f\"{HF_BASE}/{REPO_ID}/blob/{BRANCH}/{SOVITS_PATH}\"\n",
+        "\n",
+        "!cd \"/content/GPT-SoVITS/GPT_weights\" && wget \"{GPT_URL}\"\n",
+        "!cd \"/content/GPT-SoVITS/SoVITS_weights\" && wget \"{SOVITS_URL}\"\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### Download From ModelScope"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "# Modify These\n",
+        "USER_ID = \"aihobbyist\"\n",
+        "REPO_NAME = \"GPT-SoVits-V2-models\"\n",
+        "BRANCH = \"master\"\n",
+        "GPT_PATH = \"Genshin_Impact/EN/GPT_GenshinImpact_EN_5.1.ckpt\"\n",
+        "SOVITS_PATH = \"Wuthering_Waves/CN/SV_WutheringWaves_CN_1.3.pth\"\n",
+        "\n",
+        "# Do Not Modify\n",
+        "HF_BASE = \"https://www.modelscope.cn/models\"\n",
+        "REPO_ID = f\"{USER_ID}/{REPO_NAME}\"\n",
+        "GPT_URL = f\"{HF_BASE}/{REPO_ID}/resolve/{BRANCH}/{GPT_PATH}\"\n",
+        "SOVITS_URL = f\"{HF_BASE}/{REPO_ID}/resolve/{BRANCH}/{SOVITS_PATH}\"\n",
+        "\n",
+        "!cd \"/content/GPT-SoVITS/GPT_weights\" && wget \"{GPT_URL}\"\n",
+        "!cd \"/content/GPT-SoVITS/SoVITS_weights\" && wget \"{SOVITS_URL}\""
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "# Launch WebUI\n",
+        "# 启动 WebUI"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "cellView": "form",
+        "id": "4oRGUzkrk8C7"
+      },
+      "outputs": [],
+      "source": [
+        "!cd /content/GPT-SoVITS && source activate GPTSoVITS && export is_share=True && python webui.py"
+      ]
+    }
+  ],
+  "metadata": {
+    "accelerator": "GPU",
+    "colab": {
+      "provenance": []
+    },
+    "kernelspec": {
+      "display_name": "Python 3",
+      "name": "python3"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}

Colab-WebUI.ipynb

@@ -10,21 +10,28 @@
         "<a href=\"https://colab.research.google.com/github/RVC-Boss/GPT-SoVITS/blob/main/colab_webui.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
       ]
     },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "# GPT-SoVITS WebUI"
+      ]
+    },
     {
       "cell_type": "markdown",
       "metadata": {
         "id": "_o6a8GS2lWQM"
       },
       "source": [
-        "# Env Setup (Run Once Only)\n",
-        "# 环境配置, 只需运行一次"
+        "## Env Setup (Run Once Only)\n",
+        "## 环境配置, 只需运行一次"
       ]
     },
     {
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "## 1."
+        "### 1."
       ]
     },
     {
@@ -35,9 +42,11 @@
       "source": [
         "%%writefile /content/setup.sh\n",
         "set -e\n",
+        "\n",
         "cd /content\n",
-        "rm -rf GPT-SoVITS\n",
+        "\n",
         "git clone https://github.com/RVC-Boss/GPT-SoVITS.git\n",
+        "\n",
         "cd GPT-SoVITS\n",
         "\n",
         "if conda env list | awk '{print $1}' | grep -Fxq \"GPTSoVITS\"; then\n",
@@ -48,6 +57,8 @@
         "\n",
         "source activate GPTSoVITS\n",
         "\n",
+        "pip install ipykernel\n",
+        "\n",
         "bash install.sh --source HF --download-uvr5"
       ]
     },
@@ -55,7 +66,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "## 2."
+        "### 2."
       ]
     },
     {
@@ -74,8 +85,8 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "# Launch WebUI\n",
-        "# 启动 WebUI"
+        "## Launch WebUI\n",
+        "## 启动 WebUI"
       ]
     },
     {

GPT_SoVITS_Inference.ipynb

@@ -1,153 +0,0 @@
-{
-  "cells": [
-    {
-      "cell_type": "markdown",
-      "metadata": {
-        "id": "himHYZmra7ix"
-      },
-      "source": [
-        "# Credits for bubarino giving me the huggingface import code (感谢 bubarino 给了我 huggingface 导入代码)"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "id": "e9b7iFV3dm1f"
-      },
-      "outputs": [],
-      "source": [
-        "!git clone https://github.com/RVC-Boss/GPT-SoVITS.git\n",
-        "%cd GPT-SoVITS\n",
-        "!apt-get update && apt-get install -y --no-install-recommends tzdata ffmpeg libsox-dev parallel aria2 git git-lfs && git lfs install\n",
-        "!pip install -r extra-req.txt --no-deps\n",
-        "!pip install -r requirements.txt"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "cellView": "form",
-        "id": "0NgxXg5sjv7z"
-      },
-      "outputs": [],
-      "source": [
-        "# @title Download pretrained models 下载预训练模型\n",
-        "!mkdir -p /content/GPT-SoVITS/GPT_SoVITS/pretrained_models\n",
-        "!mkdir -p /content/GPT-SoVITS/tools/damo_asr/models\n",
-        "!mkdir -p /content/GPT-SoVITS/tools/uvr5\n",
-        "%cd /content/GPT-SoVITS/GPT_SoVITS/pretrained_models\n",
-        "!git clone https://huggingface.co/lj1995/GPT-SoVITS\n",
-        "%cd /content/GPT-SoVITS/tools/damo_asr/models\n",
-        "!git clone https://www.modelscope.cn/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch.git\n",
-        "!git clone https://www.modelscope.cn/damo/speech_fsmn_vad_zh-cn-16k-common-pytorch.git\n",
-        "!git clone https://www.modelscope.cn/damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch.git\n",
-        "# @title UVR5 pretrains 安装uvr5模型\n",
-        "%cd /content/GPT-SoVITS/tools/uvr5\n",
-        "!git clone https://huggingface.co/Delik/uvr5_weights\n",
-        "!git config core.sparseCheckout true\n",
-        "!mv /content/GPT-SoVITS/GPT_SoVITS/pretrained_models/GPT-SoVITS/* /content/GPT-SoVITS/GPT_SoVITS/pretrained_models/"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "cellView": "form",
-        "id": "cPDEH-9czOJF"
-      },
-      "outputs": [],
-      "source": [
-        "#@title Create folder models 创建文件夹模型\n",
-        "import os\n",
-        "base_directory = \"/content/GPT-SoVITS\"\n",
-        "folder_names = [\"SoVITS_weights\", \"GPT_weights\"]\n",
-        "\n",
-        "for folder_name in folder_names:\n",
-        "  if os.path.exists(os.path.join(base_directory, folder_name)):\n",
-        "    print(f\"The folder '{folder_name}' already exists. (文件夹'{folder_name}'已经存在。)\")\n",
-        "  else:\n",
-        "    os.makedirs(os.path.join(base_directory, folder_name))\n",
-        "    print(f\"The folder '{folder_name}' was created successfully! (文件夹'{folder_name}'已成功创建！)\")\n",
-        "\n",
-        "print(\"All folders have been created. (所有文件夹均已创建。)\")"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "cellView": "form",
-        "id": "vbZY-LnM0tzq"
-      },
-      "outputs": [],
-      "source": [
-        "import requests\n",
-        "import zipfile\n",
-        "import shutil\n",
-        "import os\n",
-        "\n",
-        "#@title Import model 导入模型 (HuggingFace)\n",
-        "hf_link = 'https://huggingface.co/modelloosrvcc/Nagisa_Shingetsu_GPT-SoVITS/resolve/main/Nagisa.zip' #@param {type: \"string\"}\n",
-        "\n",
-        "output_path = '/content/'\n",
-        "\n",
-        "response = requests.get(hf_link)\n",
-        "with open(output_path + 'file.zip', 'wb') as file:\n",
-        "    file.write(response.content)\n",
-        "\n",
-        "with zipfile.ZipFile(output_path + 'file.zip', 'r') as zip_ref:\n",
-        "    zip_ref.extractall(output_path)\n",
-        "\n",
-        "os.remove(output_path + \"file.zip\")\n",
-        "\n",
-        "source_directory = output_path\n",
-        "SoVITS_destination_directory = '/content/GPT-SoVITS/SoVITS_weights'\n",
-        "GPT_destination_directory = '/content/GPT-SoVITS/GPT_weights'\n",
-        "\n",
-        "for filename in os.listdir(source_directory):\n",
-        "    if filename.endswith(\".pth\"):\n",
-        "        source_path = os.path.join(source_directory, filename)\n",
-        "        destination_path = os.path.join(SoVITS_destination_directory, filename)\n",
-        "        shutil.move(source_path, destination_path)\n",
-        "\n",
-        "for filename in os.listdir(source_directory):\n",
-        "    if filename.endswith(\".ckpt\"):\n",
-        "        source_path = os.path.join(source_directory, filename)\n",
-        "        destination_path = os.path.join(GPT_destination_directory, filename)\n",
-        "        shutil.move(source_path, destination_path)\n",
-        "\n",
-        "print(f'Model downloaded. (模型已下载。)')"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "cellView": "form",
-        "id": "4oRGUzkrk8C7"
-      },
-      "outputs": [],
-      "source": [
-        "# @title launch WebUI 启动WebUI\n",
-        "!/usr/local/bin/pip install ipykernel\n",
-        "!sed -i '10s/False/True/' /content/GPT-SoVITS/config.py\n",
-        "%cd /content/GPT-SoVITS/\n",
-        "!/usr/local/bin/python  webui.py"
-      ]
-    }
-  ],
-  "metadata": {
-    "accelerator": "GPU",
-    "colab": {
-      "provenance": []
-    },
-    "kernelspec": {
-      "display_name": "Python 3",
-      "name": "python3"
-    }
-  },
-  "nbformat": 4,
-  "nbformat_minor": 0
-}

api.py

@@ -150,7 +150,7 @@ sys.path.append(now_dir)
 sys.path.append("%s/GPT_SoVITS" % (now_dir))
 
 import signal
-from text.LangSegmenter import LangSegmenter
+from GPT_SoVITS.text.LangSegmenter import LangSegmenter
 from time import time as ttime
 import torch
 import torchaudio
@@ -161,14 +161,14 @@ from fastapi.responses import StreamingResponse, JSONResponse
 import uvicorn
 from transformers import AutoModelForMaskedLM, AutoTokenizer
 import numpy as np
-from feature_extractor import cnhubert
+from GPT_SoVITS.feature_extractor import cnhubert
 from io import BytesIO
-from module.models import SynthesizerTrn, SynthesizerTrnV3
+from GPT_SoVITS.module.models import SynthesizerTrn, SynthesizerTrnV3
 from peft import LoraConfig, get_peft_model
-from AR.models.t2s_lightning_module import Text2SemanticLightningModule
-from text import cleaned_text_to_sequence
-from text.cleaner import clean_text
-from module.mel_processing import spectrogram_torch
+from GPT_SoVITS.AR.models.t2s_lightning_module import Text2SemanticLightningModule
+from GPT_SoVITS.text import cleaned_text_to_sequence
+from GPT_SoVITS.text.cleaner import clean_text
+from GPT_SoVITS.module.mel_processing import spectrogram_torch
 import config as global_config
 import logging
 import subprocess
@@ -176,9 +176,9 @@ import subprocess
 
 class DefaultRefer:
     def __init__(self, path, text, language):
-        self.path = args.default_refer_path
-        self.text = args.default_refer_text
-        self.language = args.default_refer_language
+        self.path = path
+        self.text = text
+        self.language = language
 
     def is_ready(self) -> bool:
         return is_full(self.path, self.text, self.language)
@@ -200,7 +200,7 @@ def is_full(*items):  # 任意一项为空返回False
 
 def init_bigvgan():
     global bigvgan_model
-    from BigVGAN import bigvgan
+    from GPT_SoVITS.BigVGAN import bigvgan
 
     bigvgan_model = bigvgan.BigVGAN.from_pretrained(
         "%s/GPT_SoVITS/pretrained_models/models--nvidia--bigvgan_v2_24khz_100band_256x" % (now_dir,),
@@ -225,7 +225,7 @@ def resample(audio_tensor, sr0):
     return resample_transform_dict[sr0](audio_tensor)
 
 
-from module.mel_processing import mel_spectrogram_torch
+from GPT_SoVITS.module.mel_processing import mel_spectrogram_torch
 
 spec_min = -12
 spec_max = 2
@@ -254,6 +254,34 @@ mel_fn = lambda x: mel_spectrogram_torch(
 )
 
 
+class DictToAttrRecursive(dict):
+    def __init__(self, input_dict):
+        super().__init__(input_dict)
+        for key, value in input_dict.items():
+            if isinstance(value, dict):
+                value = DictToAttrRecursive(value)
+            self[key] = value
+            setattr(self, key, value)
+
+    def __getattr__(self, item):
+        try:
+            return self[item]
+        except KeyError:
+            raise AttributeError(f"Attribute {item} not found")
+
+    def __setattr__(self, key, value):
+        if isinstance(value, dict):
+            value = DictToAttrRecursive(value)
+        super(DictToAttrRecursive, self).__setitem__(key, value)
+        super().__setattr__(key, value)
+
+    def __delattr__(self, item):
+        try:
+            del self[item]
+        except KeyError:
+            raise AttributeError(f"Attribute {item} not found")
+
+
 sr_model = None
 
 
@@ -289,7 +317,7 @@ class Sovits:
         self.hps = hps
 
 
-from process_ckpt import get_sovits_version_from_path_fast, load_sovits_new
+from GPT_SoVITS.process_ckpt import get_sovits_version_from_path_fast, load_sovits_new
 
 
 def get_sovits_weights(sovits_path):
@@ -438,7 +466,7 @@ def get_bert_inf(phones, word2ph, norm_text, language):
     return bert
 
 
-from text import chinese
+from GPT_SoVITS.text import chinese
 
 
 def get_phones_and_bert(text, language, version, final=False):
@@ -505,36 +533,8 @@ def get_phones_and_bert(text, language, version, final=False):
     return phones, bert.to(torch.float16 if is_half == True else torch.float32), norm_text
 
 
-class DictToAttrRecursive(dict):
-    def __init__(self, input_dict):
-        super().__init__(input_dict)
-        for key, value in input_dict.items():
-            if isinstance(value, dict):
-                value = DictToAttrRecursive(value)
-            self[key] = value
-            setattr(self, key, value)
-
-    def __getattr__(self, item):
-        try:
-            return self[item]
-        except KeyError:
-            raise AttributeError(f"Attribute {item} not found")
-
-    def __setattr__(self, key, value):
-        if isinstance(value, dict):
-            value = DictToAttrRecursive(value)
-        super(DictToAttrRecursive, self).__setitem__(key, value)
-        super().__setattr__(key, value)
-
-    def __delattr__(self, item):
-        try:
-            del self[item]
-        except KeyError:
-            raise AttributeError(f"Attribute {item} not found")
-
-
 def get_spepc(hps, filename):
-    audio, _ = librosa.load(filename, int(hps.data.sampling_rate))
+    audio, _ = librosa.load(filename, sr=int(hps.data.sampling_rate))
     audio = torch.FloatTensor(audio)
     maxx = audio.abs().max()
     if maxx > 1:
@@ -1058,15 +1058,23 @@ parser.add_argument("-b", "--bert_path", type=str, default=g_config.bert_path, h
 args = parser.parse_args()
 sovits_path = args.sovits_path
 gpt_path = args.gpt_path
+default_refer_path = args.default_refer_path
+default_refer_text = args.default_refer_text
+default_refer_language = args.default_refer_language
 device = args.device
 port = args.port
 host = args.bind_addr
+full_precision = args.full_precision
+half_precision = args.half_precision
+stream_mode = args.stream_mode
+media_type = args.media_type
+sub_type = args.sub_type
+default_cut_punc = args.cut_punc
 cnhubert_base_path = args.hubert_path
 bert_path = args.bert_path
-default_cut_punc = args.cut_punc
 
 # 应用参数配置
-default_refer = DefaultRefer(args.default_refer_path, args.default_refer_text, args.default_refer_language)
+default_refer = DefaultRefer(default_refer_path, default_refer_text, default_refer_language)
 
 # 模型路径检查
 if sovits_path == "":
@@ -1087,24 +1095,24 @@ else:
 
 # 获取半精度
 is_half = g_config.is_half
-if args.full_precision:
+if full_precision:
     is_half = False
-if args.half_precision:
+if half_precision:
     is_half = True
-if args.full_precision and args.half_precision:
+if full_precision and half_precision:
     is_half = g_config.is_half  # 炒饭fallback
 logger.info(f"半精: {is_half}")
 
 # 流式返回模式
-if args.stream_mode.lower() in ["normal", "n"]:
+if stream_mode.lower() in ["normal", "n"]:
     stream_mode = "normal"
     logger.info("流式返回已开启")
 else:
     stream_mode = "close"
 
 # 音频编码格式
-if args.media_type.lower() in ["aac", "ogg"]:
-    media_type = args.media_type.lower()
+if media_type.lower() in ["aac", "ogg"]:
+    media_type = media_type.lower()
 elif stream_mode == "close":
     media_type = "wav"
 else:
@@ -1112,14 +1120,15 @@ else:
 logger.info(f"编码格式: {media_type}")
 
 # 音频数据类型
-if args.sub_type.lower() == "int32":
+if sub_type.lower() == "int32":
     is_int32 = True
-    logger.info("数据类型: int32")
+    logger.info(f"数据类型: int32")
 else:
     is_int32 = False
-    logger.info("数据类型: int16")
+    logger.info(f"数据类型: int16")
 
 # 初始化模型
+os.environ["bert_path"] = bert_path
 cnhubert.cnhubert_base_path = cnhubert_base_path
 tokenizer = AutoTokenizer.from_pretrained(bert_path)
 bert_model = AutoModelForMaskedLM.from_pretrained(bert_path)

gpt-sovits_kaggle.ipynb

@@ -1,5 +1,13 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "9fd922fb",
+   "metadata": {},
+   "source": [
+    "# Deprecated"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,

install.sh

@@ -202,6 +202,8 @@ pip install -r extra-req.txt --no-deps
 
 pip install -r requirements.txt
 
+python -c "import nltk; nltk.download(['averaged_perceptron_tagger','averaged_perceptron_tagger_eng','cmudict'])"
+
 if [ "$USE_ROCM" = true ] && [ "$IS_WSL" = true ]; then
     echo "Update to WSL compatible runtime lib..."
     location=$(pip show torch | grep Location | awk -F ": " '{print $2}')