file add

.gitignore

@@ -0,0 +1,16 @@
+DUMMY1
+DUMMY2
+DUMMY3
+logs
+__pycache__
+.ipynb_checkpoints
+.*.swp
+
+build
+*.c
+monotonic_align/monotonic_align
+/.vs/vits/FileContentIndex
+configs/dracu_japanese_base2.json
+configs/tolove_japanese_base2.json
+
+.idea

File_System.ipynb

@@ -0,0 +1,487 @@
+{
+ "nbformat": 4,
+ "nbformat_minor": 0,
+ "metadata": {
+  "colab": {
+   "provenance": [],
+   "collapsed_sections": [
+    "3jNjgQ0JlCCL"
+   ]
+  },
+  "kernelspec": {
+   "name": "python3",
+   "display_name": "Python 3"
+  },
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "cells": [
+  {
+   "cell_type": "code",
+   "source": [
+    "import gradio as gr\n",
+    "import torch\n",
+    "print(gr.__version__)\n",
+    "print(torch.__version__)"
+   ],
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "W70gW3rnp1QP",
+    "outputId": "60be638b-06cd-4b33-e63d-6e5dc5237ca8"
+   },
+   "execution_count": 2,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "3.23.0\n",
+      "1.13.1\n"
+     ]
+    }
+   ]
+  },
+  {
+   "cell_type": "code",
+   "source": [
+    "from pathlib import Path\n",
+    "\n",
+    "MODEL_PATH = Path(\"models\") \n",
+    "MODEL_PATH.mkdir(parents=True, exist_ok=True)"
+   ],
+   "metadata": {
+    "id": "X-8HUWxxcVHr"
+   },
+   "execution_count": 3,
+   "outputs": []
+  },
+  {
+   "cell_type": "code",
+   "source": [
+    "import json\n",
+    "from PIL import Image\n",
+    "import numpy as np\n",
+    "import os\n",
+    "\n",
+    "with open(\"models/model_info.json\", \"r\", encoding=\"utf-8\") as f:\n",
+    "    models_info = json.load(f)\n",
+    "\n",
+    "\n",
+    "\n",
+    "LANGUAGES = ['EN','CN','JP']\n",
+    "speaker_id = 0\n",
+    "\n",
+    "\n"
+   ],
+   "metadata": {
+    "id": "VGacf2W4AjYm"
+   },
+   "execution_count": 37,
+   "outputs": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "outputs": [],
+   "source": [
+    "class CustomEncoder(json.JSONEncoder):\n",
+    "    def default(self, obj):\n",
+    "        if isinstance(obj, Path):\n",
+    "            return str(obj)\n",
+    "        return super().default(obj)\n"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "source": [
+    "def add_model_fn(example_text, cover, SpeakerID, name_en, name_cn, language):\n",
+    "\n",
+    "\n",
+    "\n",
+    "    # 检查必填字段是否为空\n",
+    "    if not speaker_id or not name_en or not language:\n",
+    "        raise gr.Error(\"Please fill in all required fields!\")\n",
+    "        return \"Failed to add model\"\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "    ### 保存上传的文件\n",
+    "\n",
+    "    # 生成文件路径\n",
+    "    model_save_dir = Path(\"models\")\n",
+    "    model_save_dir = model_save_dir / name_en\n",
+    "    img_save_dir = model_save_dir\n",
+    "    model_save_dir.mkdir(parents=True, exist_ok=True)\n",
+    "\n",
+    "    #shutil.copyfile(file.value,Model_save_path)\n",
+    "\n",
+    "    # file_data = file.data[0]\n",
+    "    # filename = secure_filename(file_data.name)\n",
+    "    # filepath = os.path.join(\"models\", name_en, filename)\n",
+    "    # os.makedirs(os.path.dirname(filepath), exist_ok=True)\n",
+    "\n",
+    "\n",
+    "    # 保存checkpoints 和 cover\n",
+    "    #tensor = torch.FloatTensor(file)\n",
+    "    Model_name = name_en + \".pth\"\n",
+    "    model_save_dir = model_save_dir / Model_name\n",
+    "    #torch.save(tensor, Model_save_path)\n",
+    "\n",
+    "    #\n",
+    "    #  # convert to RGB format if necessary\n",
+    "    # if len(img.shape) == 2 or img.shape[2] == 1:\n",
+    "    #     img = cv2.cvtColor(img, cv2.COLOR_GRAY2RGBA)\n",
+    "    # else:\n",
+    "    #     img = cv2.cvtColor(img, cv2.COLOR_BGR2RGBA)\n",
+    "    #\n",
+    "    # cv2.imwrite(str(img_save_dir / \"cover.png\"), img)\n",
+    "\n",
+    "    if cover is not None:\n",
+    "        img = np.array(cover)\n",
+    "        img = Image.fromarray(img)\n",
+    "        img.save(os.path.join(img_save_dir, 'cover.png'))\n",
+    "\n",
+    "\n",
+    "    \n",
+    "    #获取用户输入\n",
+    "    new_model = {\n",
+    "        \"name_en\": name_en,\n",
+    "        \"name_zh\": name_cn,\n",
+    "        \"cover\": img_save_dir / \"cover.png\",\n",
+    "        \"sid\": SpeakerID,\n",
+    "        \"example\": \"それに新しいお菓子屋さんも出来てみんな買いものを楽しんでいます！\",\n",
+    "        \"language\": language,\n",
+    "        \"type\": \"single\",\n",
+    "        \"model_path\": model_save_dir\n",
+    "    }\n",
+    "\n",
+    "\n",
+    "\n",
+    "    with open(\"models/model_info.json\", \"r\", encoding=\"utf-8\") as f:\n",
+    "        models_info = json.load(f)\n",
+    "\n",
+    "    models_info[name_en] = new_model\n",
+    "    with open(\"models/model_info.json\", \"w\") as f:\n",
+    "        json.dump(models_info, f, cls=CustomEncoder)\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "    #return file.\n",
+    "    return \"Success\"\n"
+   ],
+   "metadata": {
+    "id": "3dynM_kkBytx"
+   },
+   "execution_count": 105,
+   "outputs": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 102,
+   "outputs": [],
+   "source": [
+    "def clear_add_model_info():\n",
+    "    return \"\",None,\"\",\"\",\"\",\"\""
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 104,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running on local URL:  http://127.0.0.1:7880\n",
+      "\n",
+      "To create a public link, set `share=True` in `launch()`.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": "<IPython.core.display.HTML object>",
+      "text/html": "<div><iframe src=\"http://127.0.0.1:7880/\" width=\"100%\" height=\"500\" allow=\"autoplay; camera; microphone; clipboard-read; clipboard-write;\" frameborder=\"0\" allowfullscreen></iframe></div>"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Keyboard interruption in main thread... closing server.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": ""
+     },
+     "execution_count": 104,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "theme = gr.themes.Base()\n",
+    "\n",
+    "with gr.Blocks(theme=theme) as interface:\n",
+    "  with gr.Tab(\"Settings\"):\n",
+    "        with gr.Box():\n",
+    "            gr.Markdown(\"\"\"# Select Model\"\"\")\n",
+    "            with gr.Row():\n",
+    "\n",
+    "                with gr.Column(scale = 5):\n",
+    "                    model_choice = gr.Dropdown(label = \"Model\",\n",
+    "                                           choices=[(model[\"name_en\"]) for name, model in models_info.items()],\n",
+    "                                           interactive=True,\n",
+    "                                           value=models_info['yuuka']['name_en']\n",
+    "                                         )\n",
+    "                with gr.Column(scale = 5):\n",
+    "                    speaker_id = gr.Dropdown(label = \"Speaker ID\",\n",
+    "                                         choices=[(str(model[\"sid\"])) for name, model in models_info.items()],\n",
+    "                                         interactive=True,\n",
+    "                                         value=str(models_info['yuuka']['sid'])\n",
+    "                                         )\n",
+    "\n",
+    "                with gr.Column(scale = 1):\n",
+    "                    refresh_button = gr.Button(\"Refresh\", variant=\"primary\")\n",
+    "                    reset_button = gr.Button(\"Reset\")\n",
+    "\n",
+    "        with gr.Box():\n",
+    "            gr.Markdown(\"# Add Model\\n\"\n",
+    "                        \"> *为必填选项\\n\"\n",
+    "                        \"> 添加完成后将**checkpoints**文件放到对应生成的文件夹中\"\n",
+    "                        )\n",
+    "\n",
+    "\n",
+    "            with gr.Row():\n",
+    "                # file = gr.Files(label = \"VITS Model*\", file_types=[\".pth\"])\n",
+    "                example_text = gr.Textbox(label = \"Example Text\",\n",
+    "                                          lines=16,\n",
+    "                                          placeholder=\"Enter the example text here\",)\n",
+    "                model_cover = gr.Image(label = \"Cover\")\n",
+    "\n",
+    "                with gr.Column():\n",
+    "                    model_speaker_id = gr.Textbox(label = \"Speaker List*\",\n",
+    "                                                  placeholder=\"Single speaker model default=0\")\n",
+    "                    model_name_en = gr.Textbox(label = \"name_en*\")\n",
+    "                    model_name_cn = gr.Textbox(label = \"name_cn\")\n",
+    "                    model_language = gr.Dropdown(label = \"Language*\",\n",
+    "                                               choices=LANGUAGES,\n",
+    "                                               interactive=True)\n",
+    "                    with gr.Row():\n",
+    "                        add_model_button = gr.Button(\"Add Model\", variant=\"primary\")\n",
+    "                        clear_add_model_button = gr.Button(\"Clear\")\n",
+    "            with gr.Box():\n",
+    "              with gr.Row():\n",
+    "                message_box = gr.Textbox(label = \"Message\")\n",
+    "\n",
+    "\n",
+    "\n",
+    "        add_model_button.click(add_model_fn,\n",
+    "                               inputs = [example_text, model_cover, model_speaker_id, model_name_en, model_name_cn, model_language],\n",
+    "                               outputs = message_box\n",
+    "                               )\n",
+    "        clear_add_model_button.click(clear_add_model_info,\n",
+    "                                     outputs = [example_text, model_cover, model_speaker_id, model_name_en, model_name_cn, model_language]\n",
+    "        )\n",
+    "\n",
+    "interface.queue(concurrency_count=1).launch(debug=True)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 62,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running on local URL:  http://127.0.0.1:7866\n",
+      "\n",
+      "To create a public link, set `share=True` in `launch()`.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": "<IPython.core.display.HTML object>",
+      "text/html": "<div><iframe src=\"http://127.0.0.1:7866/\" width=\"100%\" height=\"500\" allow=\"autoplay; camera; microphone; clipboard-read; clipboard-write;\" frameborder=\"0\" allowfullscreen></iframe></div>"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Traceback (most recent call last):\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\routes.py\", line 394, in run_predict\n",
+      "    output = await app.get_blocks().process_api(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 1075, in process_api\n",
+      "    result = await self.call_function(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 884, in call_function\n",
+      "    prediction = await anyio.to_thread.run_sync(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\to_thread.py\", line 31, in run_sync\n",
+      "    return await get_asynclib().run_sync_in_worker_thread(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 937, in run_sync_in_worker_thread\n",
+      "    return await future\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 867, in run\n",
+      "    result = context.run(func, *args)\n",
+      "  File \"C:\\Users\\l4227\\AppData\\Local\\Temp\\ipykernel_11412\\3513495185.py\", line 4, in file_upload\n",
+      "    return file.name\n",
+      "AttributeError: 'list' object has no attribute 'name'\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Keyboard interruption in main thread... closing server.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": ""
+     },
+     "execution_count": 62,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "import gradio as gr\n",
+    "\n",
+    "def file_upload(file):\n",
+    "    return file.name\n",
+    "\n",
+    "\n",
+    "with gr.Blocks() as interface:\n",
+    "\n",
+    "    a = gr.Files(label = \"VITS Model*\", file_types=[\".pth\"])\n",
+    "    b = gr.Files(label = \"Cover\", file_types=[\".png\"])\n",
+    "    c = gr.Button()\n",
+    "    d = gr.Textbox()\n",
+    "\n",
+    "    c.click(fn=file_upload,inputs=a,outputs=d)\n",
+    "\n",
+    "\n",
+    "interface.queue(concurrency_count=1).launch(debug=True)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 66,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running on local URL:  http://127.0.0.1:7867\n",
+      "\n",
+      "To create a public link, set `share=True` in `launch()`.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": "<IPython.core.display.HTML object>",
+      "text/html": "<div><iframe src=\"http://127.0.0.1:7867/\" width=\"100%\" height=\"500\" allow=\"autoplay; camera; microphone; clipboard-read; clipboard-write;\" frameborder=\"0\" allowfullscreen></iframe></div>"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": ""
+     },
+     "execution_count": 66,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "import gradio as gr\n",
+    "\n",
+    "def file_upload(file):\n",
+    "    return file.name\n",
+    "\n",
+    "iface = gr.Interface(fn=file_upload, inputs=\"file\", outputs=\"text\")\n",
+    "iface.launch()"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 65,
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\deprecation.py:40: UserWarning: `optional` parameter is deprecated, and it has no effect\n",
+      "  warnings.warn(value)\n",
+      "D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\deprecation.py:40: UserWarning: `keep_filename` parameter is deprecated, and it has no effect\n",
+      "  warnings.warn(value)\n"
+     ]
+    },
+    {
+     "ename": "AttributeError",
+     "evalue": "module 'gradio.outputs' has no attribute 'Text'",
+     "output_type": "error",
+     "traceback": [
+      "\u001B[1;31m---------------------------------------------------------------------------\u001B[0m",
+      "\u001B[1;31mAttributeError\u001B[0m                            Traceback (most recent call last)",
+      "Cell \u001B[1;32mIn[65], line 8\u001B[0m\n\u001B[0;32m      5\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124m/path/to/output/file.txt\u001B[39m\u001B[38;5;124m\"\u001B[39m\n\u001B[0;32m      7\u001B[0m input_file \u001B[38;5;241m=\u001B[39m gr\u001B[38;5;241m.\u001B[39minputs\u001B[38;5;241m.\u001B[39mFile()\n\u001B[1;32m----> 8\u001B[0m output_file_path \u001B[38;5;241m=\u001B[39m \u001B[43mgr\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43moutputs\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mText\u001B[49m(\u001B[38;5;28mtype\u001B[39m\u001B[38;5;241m=\u001B[39m\u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mfile\u001B[39m\u001B[38;5;124m\"\u001B[39m)\n\u001B[0;32m     10\u001B[0m gr\u001B[38;5;241m.\u001B[39mInterface(get_file_path, inputs\u001B[38;5;241m=\u001B[39minput_file, outputs\u001B[38;5;241m=\u001B[39moutput_file_path)\u001B[38;5;241m.\u001B[39mlaunch()\n",
+      "\u001B[1;31mAttributeError\u001B[0m: module 'gradio.outputs' has no attribute 'Text'"
+     ]
+    }
+   ],
+   "source": [
+    "import gradio as gr\n",
+    "\n",
+    "def get_file_path(input_file):\n",
+    "    # do something with input file\n",
+    "    return \"/path/to/output/file.txt\"\n",
+    "\n",
+    "input_file = gr.inputs.File()\n",
+    "output_file_path = gr.outputs.Text(type=\"file\")\n",
+    "\n",
+    "gr.Interface(get_file_path, inputs=input_file, outputs=output_file_path).launch()\n"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [],
+   "metadata": {
+    "collapsed": false
+   }
+  }
+ ]
+}

GUI_Usage.ipynb

@@ -0,0 +1,1138 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "1cRKuRl7Z8Nj"
+   },
+   "source": [
+    "# Requirment"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {
+    "colab": {
+     "background_save": true
+    },
+    "id": "ouQGqsHfsDv6",
+    "outputId": "8a464347-c2ba-489e-8f45-3707e9ba2e1d"
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "G:\\AI\\VITS_WebUI\\monotonic_align\n",
+      "running build_ext\n",
+      "copying build\\lib.win-amd64-3.9\\monotonic_align\\core.cp39-win_amd64.pyd -> monotonic_align\n",
+      "G:\\AI\\VITS_WebUI\n"
+     ]
+    }
+   ],
+   "source": [
+    "%cd G:\\AI\\VITS_WebUI\\monotonic_align\n",
+    "!python setup.py build_ext --inplace\n",
+    "%cd .."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Fri Apr 21 22:47:53 2023       \n",
+      "+---------------------------------------------------------------------------------------+\n",
+      "| NVIDIA-SMI 531.14                 Driver Version: 531.14       CUDA Version: 12.1     |\n",
+      "|-----------------------------------------+----------------------+----------------------+\n",
+      "| GPU  Name                      TCC/WDDM | Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
+      "| Fan  Temp  Perf            Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
+      "|                                         |                      |               MIG M. |\n",
+      "|=========================================+======================+======================|\n",
+      "|   0  NVIDIA GeForce RTX 3060       WDDM | 00000000:01:00.0  On |                  N/A |\n",
+      "|  0%   33C    P8               20W / 170W|   8114MiB / 12288MiB |     32%      Default |\n",
+      "|                                         |                      |                  N/A |\n",
+      "+-----------------------------------------+----------------------+----------------------+\n",
+      "                                                                                         \n",
+      "+---------------------------------------------------------------------------------------+\n",
+      "| Processes:                                                                            |\n",
+      "|  GPU   GI   CI        PID   Type   Process name                            GPU Memory |\n",
+      "|        ID   ID                                                             Usage      |\n",
+      "|=======================================================================================|\n",
+      "|    0   N/A  N/A      5040    C+G   ...\\cef\\cef.win7x64\\steamwebhelper.exe    N/A      |\n",
+      "|    0   N/A  N/A      5872    C+G   ...on\\wallpaper_engine\\wallpaper32.exe    N/A      |\n",
+      "|    0   N/A  N/A      7144    C+G   ....0_x64__kzh8wxbdkxb8p\\DCv2\\DCv2.exe    N/A      |\n",
+      "|    0   N/A  N/A      8724    C+G   C:\\Windows\\explorer.exe                   N/A      |\n",
+      "|    0   N/A  N/A      9632    C+G   ....Search_cw5n1h2txyewy\\SearchApp.exe    N/A      |\n",
+      "|    0   N/A  N/A      9704    C+G   D:\\CloudMusic\\cloudmusic.exe              N/A      |\n",
+      "|    0   N/A  N/A      9812    C+G   ...2txyewy\\StartMenuExperienceHost.exe    N/A      |\n",
+      "|    0   N/A  N/A     10872    C+G   ...0.0_x64__p7pnf6hceqser\\snipaste.exe    N/A      |\n",
+      "|    0   N/A  N/A     11900    C+G   D:\\Typora\\Typora.exe                      N/A      |\n",
+      "|    0   N/A  N/A     12268    C+G   ...t.LockApp_cw5n1h2txyewy\\LockApp.exe    N/A      |\n",
+      "|    0   N/A  N/A     13320    C+G   ...rPicker\\PowerToys.ColorPickerUI.exe    N/A      |\n",
+      "|    0   N/A  N/A     13600    C+G   ...FancyZones\\PowerToys.FancyZones.exe    N/A      |\n",
+      "|    0   N/A  N/A     13660    C+G   ...5n1h2txyewy\\ShellExperienceHost.exe    N/A      |\n",
+      "|    0   N/A  N/A     13904    C+G   D:\\Eagle\\Eagle.exe                        N/A      |\n",
+      "|    0   N/A  N/A     16220    C+G   ...GeForce Experience\\NVIDIA Share.exe    N/A      |\n",
+      "|    0   N/A  N/A     16240    C+G   ...GeForce Experience\\NVIDIA Share.exe    N/A      |\n",
+      "|    0   N/A  N/A     16332    C+G   ...CBS_cw5n1h2txyewy\\TextInputHost.exe    N/A      |\n",
+      "|    0   N/A  N/A     17608    C+G   ...B\\system_tray\\lghub_system_tray.exe    N/A      |\n",
+      "|    0   N/A  N/A     17696    C+G   C:\\Program Files\\LGHUB\\lghub.exe          N/A      |\n",
+      "|    0   N/A  N/A     20848    C+G   ...oogle\\Chrome\\Application\\chrome.exe    N/A      |\n",
+      "|    0   N/A  N/A     23484    C+G   ...auncher\\PowerToys.PowerLauncher.exe    N/A      |\n",
+      "|    0   N/A  N/A     26616    C+G   D:\\motrix\\Motrix.exe                      N/A      |\n",
+      "|    0   N/A  N/A     27388    C+G   D:\\BaiduNetdisk\\baidunetdiskrender.exe    N/A      |\n",
+      "|    0   N/A  N/A     28064    C+G   ...on\\112.0.1722.39\\msedgewebview2.exe    N/A      |\n",
+      "|    0   N/A  N/A     28988    C+G   ...3\\extracted\\runtime\\WeChatAppEx.exe    N/A      |\n",
+      "|    0   N/A  N/A     32628    C+G   D:\\RaiDrive\\RaiDrive.exe                  N/A      |\n",
+      "|    0   N/A  N/A     34352    C+G   ...1.0_x64__8wekyb3d8bbwe\\Video.UI.exe    N/A      |\n",
+      "|    0   N/A  N/A     64972    C+G   ...ft Office\\root\\Office16\\WINWORD.EXE    N/A      |\n",
+      "|    0   N/A  N/A     86756    C+G   ..._8wekyb3d8bbwe\\Microsoft.Photos.exe    N/A      |\n",
+      "|    0   N/A  N/A    159156    C+G   ...siveControlPanel\\SystemSettings.exe    N/A      |\n",
+      "|    0   N/A  N/A    186184    C+G   ...9.0.0_x64__gqbn7fs4pywxm\\Db.App.exe    N/A      |\n",
+      "|    0   N/A  N/A    326192    C+G   ...les\\Microsoft OneDrive\\OneDrive.exe    N/A      |\n",
+      "|    0   N/A  N/A    366360    C+G   ...ekyb3d8bbwe\\PhoneExperienceHost.exe    N/A      |\n",
+      "|    0   N/A  N/A    455764    C+G   ...-ins\\Spaces\\Adobe Spaces Helper.exe    N/A      |\n",
+      "|    0   N/A  N/A    456280    C+G   ...obe Photoshop CC 2019\\Photoshop.exe    N/A      |\n",
+      "|    0   N/A  N/A    456496    C+G   ...CEP\\CEPHtmlEngine\\CEPHtmlEngine.exe    N/A      |\n",
+      "|    0   N/A  N/A    469488    C+G   ...t Office\\root\\Office16\\POWERPNT.EXE    N/A      |\n",
+      "|    0   N/A  N/A    493720    C+G   ...rm 2022.3.2\\jbr\\bin\\jcef_helper.exe    N/A      |\n",
+      "|    0   N/A  N/A    498508    C+G   ...crosoft\\Edge\\Application\\msedge.exe    N/A      |\n",
+      "+---------------------------------------------------------------------------------------+\n"
+     ]
+    }
+   ],
+   "source": [
+    "!nvidia-smi"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "SxpEIauJZ0s6"
+   },
+   "source": [
+    "# Settings"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {
+    "cellView": "form",
+    "id": "v10x1lO7Z5AK"
+   },
+   "outputs": [],
+   "source": [
+    "#@title  Edit config\n",
+    "import json\n",
+    "batchsize = 16  #@param {type:\"number\"}\n",
+    "training_files = \"filelists/yuuka_train.txt.cleaned\" #@param {type:\"string\"}\n",
+    "validation_files = \"filelists/yuuka_val.txt.cleaned\" #@param {type:\"string\"}\n",
+    "config = json.load(open(\"configs/config.json\"))\n",
+    "config['train']['batch_size'] = batchsize\n",
+    "config['data']['training_files'] = training_files\n",
+    "config['data']['validation_files'] = validation_files\n",
+    "with open(\"configs/config.json\", 'w+') as f:\n",
+    "    json.dump(config, f, indent=4)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "XBNba8Qpa7XF"
+   },
+   "source": [
+    "# GUI"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {
+    "id": "zF5IUSAQa_EB"
+   },
+   "outputs": [],
+   "source": [
+    "import gradio as gr\n",
+    "import numpy as np"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {
+    "id": "gcO8hd1Jr2t6"
+   },
+   "outputs": [],
+   "source": [
+    "%matplotlib inline\n",
+    "import matplotlib.pyplot as plt\n",
+    "import IPython.display as ipd\n",
+    "import os\n",
+    "import json\n",
+    "import math\n",
+    "import torch\n",
+    "import commons\n",
+    "import utils\n",
+    "from models import SynthesizerTrn\n",
+    "from text.symbols import symbols\n",
+    "from text import text_to_sequence\n",
+    "from scipy.io.wavfile import write\n",
+    "from gradio.processing_utils import download_tmp_copy_of_file\n",
+    "from PIL import Image\n",
+    "import numpy as np\n",
+    "import os\n",
+    "from pathlib import Path\n",
+    "import openai\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {
+    "id": "tp-8n_YBg5FN"
+   },
+   "outputs": [],
+   "source": [
+    "LANGUAGES = ['EN','CN','JP']\n",
+    "SPEAKER_ID = 0\n",
+    "COVER = \"models/Yuuka/cover.png\"\n",
+    "speaker_choice = \"Yuuka\"\n",
+    "MODEL_ZH_NAME = \"早濑优香\"\n",
+    "EXAMPLE_TEXT = \"先生。今日も全力であなたをアシストしますね。\"\n",
+    "USER_INPUT_TEXT = \"\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "INFO:root:Loaded checkpoint 'models/Yuuka/Yuuka.pth' (iteration 445)\n"
+     ]
+    }
+   ],
+   "source": [
+    "CONFIG_PATH = \"configs/config.json\"\n",
+    "MODEL_PATH = \"models/Yuuka/Yuuka.pth\"\n",
+    "\n",
+    "hps = utils.get_hparams_from_file(CONFIG_PATH)\n",
+    "net_g = SynthesizerTrn(\n",
+    "    len(hps.symbols),\n",
+    "    hps.data.filter_length // 2 + 1,\n",
+    "    hps.train.segment_size // hps.data.hop_length,\n",
+    "    n_speakers=hps.data.n_speakers,\n",
+    "    **hps.model).cuda()\n",
+    "\n",
+    "model = net_g.eval()\n",
+    "model = utils.load_checkpoint(MODEL_PATH, net_g, None)\n",
+    "\n",
+    "def tts_fn(text, noise_scale, noise_scale_w, length_scale):\n",
+    "  stn_tst = get_text(text, hps)\n",
+    "  with torch.no_grad():\n",
+    "    x_tst = stn_tst.cuda().unsqueeze(0)\n",
+    "    x_tst_lengths = torch.LongTensor([stn_tst.size(0)]).cuda()\n",
+    "    sid = torch.LongTensor([SPEAKER_ID]).cuda()\n",
+    "    audio = net_g.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale, noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0,0].data.cpu().float().numpy()\n",
+    "  return  (22050, audio)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "outputs": [],
+   "source": [
+    "with open(\"models/model_info.json\", \"r\", encoding=\"utf-8\") as f:\n",
+    "    models_info = json.load(f)\n",
+    "\n",
+    "for i,model_info in models_info.items():\n",
+    "    name_en = model_info['name_en']"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "outputs": [],
+   "source": [
+    "def load_model():\n",
+    "    global hps,net_g,model\n",
+    "\n",
+    "    hps = utils.get_hparams_from_file(CONFIG_PATH)\n",
+    "    net_g = SynthesizerTrn(\n",
+    "    len(hps.symbols),\n",
+    "    hps.data.filter_length // 2 + 1,\n",
+    "    hps.train.segment_size // hps.data.hop_length,\n",
+    "    n_speakers=hps.data.n_speakers,\n",
+    "    **hps.model).cuda()\n",
+    "\n",
+    "    model = net_g.eval()\n",
+    "    model = utils.load_checkpoint(MODEL_PATH, net_g, None)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "outputs": [],
+   "source": [
+    "def get_text(text, hps):\n",
+    "    text_norm = text_to_sequence(text, hps.data.text_cleaners)\n",
+    "    if hps.data.add_blank:\n",
+    "        text_norm = commons.intersperse(text_norm, 0)\n",
+    "    text_norm = torch.LongTensor(text_norm)\n",
+    "    return text_norm"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "outputs": [],
+   "source": [
+    "def tts_fn(text, noise_scale, noise_scale_w, length_scale):\n",
+    "  stn_tst = get_text(text, hps)\n",
+    "  with torch.no_grad():\n",
+    "    x_tst = stn_tst.cuda().unsqueeze(0)\n",
+    "    x_tst_lengths = torch.LongTensor([stn_tst.size(0)]).cuda()\n",
+    "    sid = torch.LongTensor([SPEAKER_ID]).cuda()\n",
+    "    audio = net_g.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale, noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0,0].data.cpu().float().numpy()\n",
+    "  return  (22050, audio)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "outputs": [],
+   "source": [
+    "def add_model_fn(example_text, cover, speakerID, name_en, name_cn, language):\n",
+    "\n",
+    "    # 检查必填字段是否为空\n",
+    "    if not SPEAKER_ID or not name_en or not language:\n",
+    "        raise gr.Error(\"Please fill in all required fields!\")\n",
+    "        return \"Failed to add model\"\n",
+    "\n",
+    "    ### 保存上传的文件\n",
+    "\n",
+    "    # 生成文件路径\n",
+    "    model_save_dir = Path(\"models\")\n",
+    "    model_save_dir = model_save_dir / name_en\n",
+    "    img_save_dir = model_save_dir\n",
+    "    model_save_dir.mkdir(parents=True, exist_ok=True)\n",
+    "\n",
+    "    Model_name = name_en + \".pth\"\n",
+    "    model_save_dir = model_save_dir / Model_name\n",
+    "\n",
+    "    # 保存上传的图片\n",
+    "    if cover is not None:\n",
+    "        img = np.array(cover)\n",
+    "        img = Image.fromarray(img)\n",
+    "        img.save(os.path.join(img_save_dir, 'cover_white_background.png'))\n",
+    "\n",
+    "    #获取用户输入\n",
+    "    new_model = {\n",
+    "        \"name_en\": name_en,\n",
+    "        \"name_zh\": name_cn,\n",
+    "        \"cover\": img_save_dir / \"cover.png\",\n",
+    "        \"sid\": speakerID,\n",
+    "        \"example\": example_text,\n",
+    "        \"language\": language,\n",
+    "        \"type\": \"single\",\n",
+    "        \"model_path\": model_save_dir\n",
+    "    }\n",
+    "\n",
+    "    #写入json\n",
+    "    with open(\"models/model_info.json\", \"r\", encoding=\"utf-8\") as f:\n",
+    "        models_info = json.load(f)\n",
+    "\n",
+    "    models_info[name_en] = new_model\n",
+    "    with open(\"models/model_info.json\", \"w\") as f:\n",
+    "        json.dump(models_info, f, cls=CustomEncoder)\n",
+    "\n",
+    "\n",
+    "    return \"Success\""
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "outputs": [],
+   "source": [
+    "def clear_input_text():\n",
+    "    return \"\""
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "outputs": [],
+   "source": [
+    "def clear_add_model_info():\n",
+    "    return \"\",None,\"\",\"\",\"\",\"\""
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "outputs": [],
+   "source": [
+    "def get_options():\n",
+    "  with open(\"models/model_info.json\", \"r\", encoding=\"utf-8\") as f:\n",
+    "    global models_info\n",
+    "    models_info = json.load(f)\n",
+    "\n",
+    "  for i,model_info in models_info.items():\n",
+    "    global name_en\n",
+    "    name_en = model_info['name_en']"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "outputs": [],
+   "source": [
+    "def reset_options():\n",
+    "  value_model_choice = models_info['Yuuka']['name_en']\n",
+    "  value_speaker_id = models_info['Yuuka']['sid']\n",
+    "  return value_model_choice,value_speaker_id"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "outputs": [],
+   "source": [
+    "def refresh_options():\n",
+    "  get_options()\n",
+    "  value_model_choice = models_info[speaker_choice]['name_en']\n",
+    "  value_speaker_id = models_info[speaker_choice]['sid']\n",
+    "  return value_model_choice,value_speaker_id"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "outputs": [],
+   "source": [
+    "def change_dropdown(choice):\n",
+    "  global speaker_choice\n",
+    "  speaker_choice = choice\n",
+    "  global COVER\n",
+    "  COVER = str(models_info[speaker_choice]['cover'])\n",
+    "  global MODEL_PATH\n",
+    "  MODEL_PATH = str(models_info[speaker_choice]['model_path'])\n",
+    "  global MODEL_ZH_NAME\n",
+    "  MODEL_ZH_NAME = str(models_info[speaker_choice]['name_zh'])\n",
+    "  global EXAMPLE_TEXT\n",
+    "  EXAMPLE_TEXT = str(models_info[speaker_choice]['example'])\n",
+    "\n",
+    "  speaker_id_change = gr.update(value=str(models_info[speaker_choice]['sid']))\n",
+    "  cover_change = gr.update(value='<div align=\"center\">'\n",
+    "                f'<img style=\"width:auto;height:512px;\" src=\"file/{COVER}\">' if COVER else \"\"\n",
+    "                f'<a><strong>{speaker_choice}</strong></a>'\n",
+    "                                                                                           '</div>')\n",
+    "  title_change = gr.update(value=\n",
+    "                '<div align=\"center\">'\n",
+    "                f'<h3><a><strong>{\"语音名称: \"}{MODEL_ZH_NAME}</strong></a>'\n",
+    "                f'<h3><strong>{\"checkpoint: \"}{speaker_choice}</strong>'\n",
+    "                                                                                           '</div>')\n",
+    "\n",
+    "\n",
+    "  lan_change = gr.update(value=str(models_info[speaker_choice]['language']))\n",
+    "\n",
+    "  example_change = gr.update(value=EXAMPLE_TEXT)\n",
+    "\n",
+    "  ChatGPT_cover_change = gr.update(value='<div align=\"center\">'\n",
+    "                f'<img style=\"width:auto;height:512px;\" src=\"file/{COVER}\">' if COVER else \"\"\n",
+    "                f'<a><strong>{speaker_choice}</strong></a>'\n",
+    "                                                                                           '</div>')\n",
+    "  ChatGPT_title_change = gr.update(value=\n",
+    "                '<div align=\"center\">'\n",
+    "                f'<h3><a><strong>{\"语音名称: \"}{MODEL_ZH_NAME}</strong></a>'\n",
+    "                f'<h3><strong>{\"checkpoint: \"}{speaker_choice}</strong>'\n",
+    "                                                                                           '</div>')\n",
+    "\n",
+    "  load_model()\n",
+    "\n",
+    "  return [speaker_id_change,cover_change,title_change,lan_change,example_change,cover_change,title_change,lan_change]"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "outputs": [],
+   "source": [
+    "def load_api_key(value):\n",
+    "  openai.api_key = value"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "outputs": [],
+   "source": [
+    "def usr_input_update(value):\n",
+    "  global USER_INPUT_TEXT\n",
+    "  USER_INPUT_TEXT = value"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "outputs": [],
+   "source": [
+    "# def ChatGPT_Bot(history):\n",
+    "#     response = openai.ChatCompletion.create(\n",
+    "#       model=\"gpt-3.5-turbo\",\n",
+    "#       messages=[\n",
+    "#           {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n",
+    "#           {\"role\": \"user\", \"content\": USER_INPUT_TEXT}\n",
+    "#         ]\n",
+    "#     )\n",
+    "#\n",
+    "#     history[-1][1] = response['choices'][0]['message']['content']\n",
+    "#\n",
+    "#     return history\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "def ChatGPT_Bot(history):\n",
+    "    response = \"先生。今日も全力であなたをアシストしますね。\"\n",
+    "\n",
+    "    history[-1][1] = response\n",
+    "\n",
+    "    return history\n",
+    "\n",
+    "\n",
+    "\n",
+    "def bot(history):\n",
+    "    last_input = history[-1][0]\n",
+    "\n",
+    "    audio = tts_fn(last_input,0.6,0.668,1.0)\n",
+    "    history[-1][1] = audio\n",
+    "\n",
+    "    return history\n"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "outputs": [],
+   "source": [
+    "def add_text(history, text):\n",
+    "    history = history + [(text, None)]\n",
+    "    return history, \"\""
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "outputs": [],
+   "source": [
+    "class CustomEncoder(json.JSONEncoder):\n",
+    "    def default(self, obj):\n",
+    "        if isinstance(obj, Path):\n",
+    "            return str(obj)\n",
+    "        return super().default(obj)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "outputs": [],
+   "source": [
+    "download_audio_js = \"\"\"\n",
+    "() =>{{\n",
+    "    let root = document.querySelector(\"body > gradio-app\");\n",
+    "    if (root.shadowRoot != null)\n",
+    "        root = root.shadowRoot;\n",
+    "    let audio = root.querySelector(\"#tts-audio-{audio_id}\").querySelector(\"audio\");\n",
+    "    let text = root.querySelector(\"#input-text-{audio_id}\").querySelector(\"textarea\");\n",
+    "    if (audio == undefined)\n",
+    "        return;\n",
+    "    text = text.value;\n",
+    "    if (text == undefined)\n",
+    "        text = Math.floor(Math.random()*100000000);\n",
+    "    audio = audio.src;\n",
+    "    let oA = document.createElement(\"a\");\n",
+    "    oA.download = text.substr(0, 20)+'.wav';\n",
+    "    oA.href = audio;\n",
+    "    document.body.appendChild(oA);\n",
+    "    oA.click();\n",
+    "    oA.remove();\n",
+    "}}\n",
+    "\"\"\""
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): api.gradio.app:443\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): huggingface.co:443\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.reference:entering reference: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.reference:entering reference: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.reference:entering reference: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.reference:entering reference: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.reference:entering reference: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.heading:entering heading: StateBlock(line=0,level=0,tokens=0), 0, 1, False\n",
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.reference:entering reference: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.heading:entering heading: StateBlock(line=0,level=0,tokens=0), 0, 3, False\n",
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=1,level=0,tokens=3), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=1,level=0,tokens=3), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=1,level=0,tokens=3), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.code:entering code: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.reference:entering reference: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.heading:entering heading: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.lheading:entering lheading: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.heading:entering heading: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.paragraph:entering paragraph: StateBlock(line=1,level=1,tokens=4), 1, 3, False\n",
+      "DEBUG:markdown_it.rules_block.fence:entering fence: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.blockquote:entering blockquote: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.hr:entering hr: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.list:entering list: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.html_block:entering html_block: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n",
+      "DEBUG:markdown_it.rules_block.heading:entering heading: StateBlock(line=1,level=1,tokens=4), 2, 3, True\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\deprecation.py:43: UserWarning: You have unused kwarg parameters in Textbox, please remove them: {'scale': 2}\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:asyncio:Using selector: SelectSelector\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): 127.0.0.1:7860\n",
+      "DEBUG:urllib3.connectionpool:http://127.0.0.1:7860 \"GET /startup-events HTTP/1.1\" 200 5\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): 127.0.0.1:7860\n",
+      "DEBUG:urllib3.connectionpool:http://127.0.0.1:7860 \"HEAD / HTTP/1.1\" 200 0\n",
+      "Running on local URL:  http://127.0.0.1:7860\n",
+      "\n",
+      "To create a public link, set `share=True` in `launch()`.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": "<IPython.core.display.HTML object>",
+      "text/html": "<div><iframe src=\"http://127.0.0.1:7860/\" width=\"100%\" height=\"500\" allow=\"autoplay; camera; microphone; clipboard-read; clipboard-write;\" frameborder=\"0\" allowfullscreen></iframe></div>"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): api.gradio.app:443\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): huggingface.co:443\n",
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/api/predict \"HTTP/1.1 200 OK\"\n",
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/reset \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\processing_utils.py:234: UserWarning: Trying to convert audio automatically from float32 to 16-bit int format.\n",
+      "  warnings.warn(warning.format(data.dtype))\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/api/predict \"HTTP/1.1 500 Internal Server Error\"\n",
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/reset \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Traceback (most recent call last):\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\routes.py\", line 394, in run_predict\n",
+      "    output = await app.get_blocks().process_api(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 1075, in process_api\n",
+      "    result = await self.call_function(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 884, in call_function\n",
+      "    prediction = await anyio.to_thread.run_sync(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\to_thread.py\", line 31, in run_sync\n",
+      "    return await get_asynclib().run_sync_in_worker_thread(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 937, in run_sync_in_worker_thread\n",
+      "    return await future\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 867, in run\n",
+      "    result = context.run(func, *args)\n",
+      "  File \"C:\\Users\\l4227\\AppData\\Local\\Temp\\ipykernel_501044\\4197914779.py\", line 7, in tts_fn\n",
+      "    audio = net_g.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale, noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0,0].data.cpu().float().numpy()\n",
+      "  File \"G:\\AI\\VITS_WebUI\\models.py\", line 500, in infer\n",
+      "    x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1194, in _call_impl\n",
+      "    return forward_call(*input, **kwargs)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\models.py\", line 172, in forward\n",
+      "    x = self.encoder(x * x_mask, x_mask)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1194, in _call_impl\n",
+      "    return forward_call(*input, **kwargs)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 39, in forward\n",
+      "    y = self.attn_layers[i](x, x, attn_mask)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1194, in _call_impl\n",
+      "    return forward_call(*input, **kwargs)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 143, in forward\n",
+      "    x, self.attn = self.attention(q, k, v, mask=attn_mask)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 160, in attention\n",
+      "    scores_local = self._relative_position_to_absolute_position(rel_logits)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 225, in _relative_position_to_absolute_position\n",
+      "    x_flat = F.pad(x_flat, commons.convert_pad_shape([[0,0],[0,0],[0,length-1]]))\n",
+      "torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 3.46 GiB (GPU 0; 12.00 GiB total capacity; 9.99 GiB already allocated; 0 bytes free; 10.09 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/api/predict \"HTTP/1.1 500 Internal Server Error\"\n",
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/reset \"HTTP/1.1 200 OK\"\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Traceback (most recent call last):\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\routes.py\", line 394, in run_predict\n",
+      "    output = await app.get_blocks().process_api(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 1075, in process_api\n",
+      "    result = await self.call_function(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 884, in call_function\n",
+      "    prediction = await anyio.to_thread.run_sync(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\to_thread.py\", line 31, in run_sync\n",
+      "    return await get_asynclib().run_sync_in_worker_thread(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 937, in run_sync_in_worker_thread\n",
+      "    return await future\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 867, in run\n",
+      "    result = context.run(func, *args)\n",
+      "  File \"C:\\Users\\l4227\\AppData\\Local\\Temp\\ipykernel_501044\\4197914779.py\", line 7, in tts_fn\n",
+      "    audio = net_g.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale, noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0,0].data.cpu().float().numpy()\n",
+      "  File \"G:\\AI\\VITS_WebUI\\models.py\", line 500, in infer\n",
+      "    x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1194, in _call_impl\n",
+      "    return forward_call(*input, **kwargs)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\models.py\", line 172, in forward\n",
+      "    x = self.encoder(x * x_mask, x_mask)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1194, in _call_impl\n",
+      "    return forward_call(*input, **kwargs)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 39, in forward\n",
+      "    y = self.attn_layers[i](x, x, attn_mask)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1194, in _call_impl\n",
+      "    return forward_call(*input, **kwargs)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 143, in forward\n",
+      "    x, self.attn = self.attention(q, k, v, mask=attn_mask)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 160, in attention\n",
+      "    scores_local = self._relative_position_to_absolute_position(rel_logits)\n",
+      "  File \"G:\\AI\\VITS_WebUI\\attentions.py\", line 221, in _relative_position_to_absolute_position\n",
+      "    x = F.pad(x, commons.convert_pad_shape([[0,0],[0,0],[0,0],[0,1]]))\n",
+      "torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 572.00 MiB (GPU 0; 12.00 GiB total capacity; 11.00 GiB already allocated; 0 bytes free; 11.06 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/api/predict \"HTTP/1.1 200 OK\"\n",
+      "DEBUG:httpx._client:HTTP Request: POST http://127.0.0.1:7860/reset \"HTTP/1.1 200 OK\"\n",
+      "Keyboard interruption in main thread... closing server.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": ""
+     },
+     "execution_count": 27,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "get_options()\n",
+    "\n",
+    "theme = gr.themes.Base()\n",
+    "\n",
+    "with gr.Blocks(theme=theme) as interface:\n",
+    "    with gr.Tab(\"Text to Speech\"):\n",
+    "        with gr.Column():\n",
+    "            cover_markdown = gr.Markdown(\n",
+    "                '<div align=\"center\">'\n",
+    "                f'<img style=\"width:auto;height:512px;\" src=\"file/{COVER}\">' if COVER else \"\"\n",
+    "                                                                                           '</div>')\n",
+    "            title_markdown = gr.Markdown(\n",
+    "                '<div align=\"center\">'\n",
+    "                f'<h3><a><strong>{\"语音名称: \"}{MODEL_ZH_NAME}</strong></a>'\n",
+    "                f'<h3><strong>{\"checkpoint: \"}{speaker_choice}</strong>'\n",
+    "                                                                                           '</div>')\n",
+    "\n",
+    "            with gr.Row():\n",
+    "                with gr.Column(scale = 4):\n",
+    "                    input_text = gr.Textbox(\n",
+    "                        label=\"Input\",\n",
+    "                        lines=2,\n",
+    "                        placeholder=\"Enter the text you want to process here\",\n",
+    "                        elem_id=f\"input-text-en-{name_en.replace(' ','')}\",\n",
+    "                        scale = 2\n",
+    "                    )\n",
+    "                with gr.Column(scale = 1):\n",
+    "                    gen_button = gr.Button(\"Generate\", variant=\"primary\")\n",
+    "                    clear_input_button = gr.Button(\"Clear\")\n",
+    "\n",
+    "            with gr.Row():\n",
+    "                with gr.Column(scale = 2):\n",
+    "                    lan = gr.Radio(label=\"Language\", choices=LANGUAGES, value=\"JP\")\n",
+    "                    noise_scale = gr.Slider(minimum=0.1, maximum=1.0, step=0.1, label = \"Noise Scale (情感变化程度)\", value = 0.6)\n",
+    "                    noise_scale_w = gr.Slider(minimum=0.1, maximum=1.0, step=0.1, label = \"Noise Scale w (发音长度)\", value = 0.668)\n",
+    "                    length_scale = gr.Slider(minimum=0.1, maximum=2.0, step=0.1, label = \"Length Scale (语速)\", value=1.0)\n",
+    "\n",
+    "                with gr.Column(scale = 1):\n",
+    "                    example_text_box = gr.Textbox(label=\"Example:\",\n",
+    "                                                  value=EXAMPLE_TEXT)\n",
+    "\n",
+    "                    output_audio = gr.Audio(label=\"Output\", elem_id=f\"tts-audio-en-{name_en.replace(' ','')}\")\n",
+    "                    download_button = gr.Button(\"Download\")\n",
+    "\n",
+    "                    # example = gr.Examples(\n",
+    "                    #     examples = [EXAMPLE_TEXT],\n",
+    "                    #     inputs=input_text,\n",
+    "                    #     outputs = output_audio,\n",
+    "                    #     fn=example_tts_fn,\n",
+    "                    #     cache_examples=True\n",
+    "                    # )\n",
+    "\n",
+    "\n",
+    "        gen_button.click(\n",
+    "                    tts_fn,\n",
+    "                    inputs = [input_text, noise_scale, noise_scale_w, length_scale],\n",
+    "                    outputs = output_audio)\n",
+    "        clear_input_button.click(\n",
+    "                    clear_input_text,\n",
+    "                    outputs = input_text\n",
+    "        )\n",
+    "        download_button.click(None, [], [], _js=download_audio_js.format(audio_id=f\"en-{name_en.replace(' ', '')}\"))\n",
+    "\n",
+    "\n",
+    "\n",
+    "#------------------------------------------------------------------------------------------------------------------------\n",
+    "    with gr.Tab(\"AI Singer\"):\n",
+    "        input_text_gpt = gr.Textbox()\n",
+    "\n",
+    "\n",
+    "#------------------------------------------------------------------------------------------------------------------------\n",
+    "    with gr.Tab(\"TTS with ChatGPT\"):\n",
+    "        with gr.Row():\n",
+    "            with gr.Column(scale=7):\n",
+    "                api_key = gr.Textbox(\n",
+    "                    label = \"API Key\",\n",
+    "                    type=\"password\")\n",
+    "                api_key.change(fn=load_api_key,inputs=api_key)\n",
+    "            with gr.Column(scale=1):\n",
+    "                lan_ChatGPT = gr.Radio(label=\"Language\", choices=LANGUAGES, value=\"JP\")\n",
+    "\n",
+    "        with gr.Row():\n",
+    "            with gr.Column(scale=1):\n",
+    "                user_input = gr.Textbox(\n",
+    "                    show_label=False,\n",
+    "                    placeholder=\"Enter text and press enter\")\n",
+    "\n",
+    "                with gr.Row():\n",
+    "                    submit_button = gr.Button(\"Submit\", variant=\"primary\")\n",
+    "                    submit_clear_button = gr.Button(\"Clear\")\n",
+    "\n",
+    "                cover_markdown_ChatGPT = gr.Markdown(\n",
+    "                '<div align=\"center\">'\n",
+    "                f'<img style=\"width:auto;height:512px;\" src=\"file/{COVER}\">' if COVER else \"\"\n",
+    "                                                                                           '</div>')\n",
+    "                title_markdown_ChatGPT = gr.Markdown(\n",
+    "                '<div align=\"center\">'\n",
+    "                f'<h3><a><strong>{\"语音名称: \"}{MODEL_ZH_NAME}</strong></a>'\n",
+    "                f'<h3><strong>{\"checkpoint: \"}{speaker_choice}</strong>'\n",
+    "                                                                                           '</div>')\n",
+    "            with gr.Column(scale=2):\n",
+    "                chatbot = gr.Chatbot([], elem_id=\"chatbot\").style(height=750)\n",
+    "\n",
+    "\n",
+    "\n",
+    "        user_input.change(fn=usr_input_update, inputs=user_input)\n",
+    "\n",
+    "        user_input.submit(add_text, [chatbot ,user_input], [chatbot ,user_input], queue=False).then(bot, chatbot, chatbot)\n",
+    "\n",
+    "        submit_button.click(\n",
+    "            fn=add_text,\n",
+    "            inputs=[chatbot ,user_input],\n",
+    "            outputs=[chatbot ,user_input],\n",
+    "            queue=False).then(ChatGPT_Bot, chatbot, chatbot)\n",
+    "\n",
+    "\n",
+    "\n",
+    "#------------------------------------------------------------------------------------------------------------------------\n",
+    "    with gr.Tab(\"Settings\"):\n",
+    "        with gr.Box():\n",
+    "            gr.Markdown(\"\"\"# Select Model\"\"\")\n",
+    "            with gr.Row():\n",
+    "\n",
+    "                with gr.Column(scale = 5):\n",
+    "                    model_choice = gr.Dropdown(label = \"Model\",\n",
+    "                                           choices=[(model[\"name_en\"]) for name, model in models_info.items()],\n",
+    "                                           interactive=True,\n",
+    "                                           value=models_info['Yuuka']['name_en']\n",
+    "                                         )\n",
+    "                with gr.Column(scale = 5):\n",
+    "                    speaker_id_choice = gr.Dropdown(label = \"Speaker ID\",\n",
+    "                                         choices=[(str(model[\"sid\"])) for name, model in models_info.items()],\n",
+    "                                         interactive=True,\n",
+    "                                         value=str(models_info['Yuuka']['sid'])\n",
+    "                                         )\n",
+    "\n",
+    "                with gr.Column(scale = 1):\n",
+    "                    refresh_button = gr.Button(\"Refresh\", variant=\"primary\")\n",
+    "                    reset_button = gr.Button(\"Reset\")\n",
+    "\n",
+    "        ### 切换模型功能实现\n",
+    "        model_choice.change(fn=change_dropdown, inputs=model_choice, outputs=[speaker_id_choice,cover_markdown,title_markdown,lan,example_text_box,cover_markdown_ChatGPT,title_markdown_ChatGPT,lan_ChatGPT])\n",
+    "\n",
+    "        refresh_button.click(fn=refresh_options, outputs = [model_choice,speaker_id_choice])\n",
+    "        reset_button.click(reset_options, outputs = [model_choice,speaker_id_choice])\n",
+    "\n",
+    "\n",
+    "        with gr.Box():\n",
+    "            gr.Markdown(\"# Add Model\\n\"\n",
+    "                        \"> *为必填选项\\n\"\n",
+    "                        \"> 添加完成后将**checkpoints**文件放到对应生成的文件夹中\"\n",
+    "                        )\n",
+    "\n",
+    "\n",
+    "            with gr.Row():\n",
+    "                # file = gr.Files(label = \"VITS Model*\", file_types=[\".pth\"])\n",
+    "                example_text = gr.Textbox(label = \"Example Text\",\n",
+    "                                          lines=16,\n",
+    "                                          placeholder=\"Enter the example text here\",)\n",
+    "                model_cover = gr.Image(label = \"Cover\")\n",
+    "\n",
+    "                with gr.Column():\n",
+    "                    model_speaker_id = gr.Textbox(label = \"Speaker List*\",\n",
+    "                                                  placeholder=\"Single speaker model default=0\")\n",
+    "                    model_name_en = gr.Textbox(label = \"name_en*\")\n",
+    "                    model_name_cn = gr.Textbox(label = \"name_cn\")\n",
+    "                    model_language = gr.Dropdown(label = \"Language*\",\n",
+    "                                               choices=LANGUAGES,\n",
+    "                                               interactive=True)\n",
+    "                    with gr.Row():\n",
+    "                        add_model_button = gr.Button(\"Add Model\", variant=\"primary\")\n",
+    "                        clear_add_model_button = gr.Button(\"Clear\")\n",
+    "            with gr.Box():\n",
+    "              with gr.Row():\n",
+    "                message_box = gr.Textbox(label = \"Message\")\n",
+    "\n",
+    "\n",
+    "\n",
+    "        add_model_button.click(add_model_fn,\n",
+    "                               inputs = [example_text, model_cover, model_speaker_id, model_name_en, model_name_cn, model_language],\n",
+    "                               outputs = message_box\n",
+    "                               )\n",
+    "        clear_add_model_button.click(clear_add_model_info,\n",
+    "                                     outputs = [example_text, model_cover, model_speaker_id, model_name_en, model_name_cn, model_language]\n",
+    "        )\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "interface.queue(concurrency_count=1).launch(debug=True)\n",
+    "\n"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [],
+   "metadata": {
+    "collapsed": false,
+    "pycharm": {
+     "is_executing": true
+    }
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [],
+   "metadata": {
+    "collapsed": false
+   }
+  }
+ ],
+ "metadata": {
+  "colab": {
+   "authorship_tag": "ABX9TyOanTxjuTkrY9G5Z/F1JKYD",
+   "collapsed_sections": [
+    "1cRKuRl7Z8Nj",
+    "uYQ2esCNI4IT",
+    "YvWwpaTKI5Ut",
+    "1rerX8gxPmLf",
+    "vs-wM321Zk0u",
+    "SxpEIauJZ0s6"
+   ],
+   "provenance": []
+  },
+  "gpuClass": "standard",
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.9.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 1
+}

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2021 Jaehyeon Kim
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

SoftVC.ipynb

@@ -0,0 +1,345 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "import gradio as gr\n",
+    "import os\n",
+    "\n",
+    "import logging\n",
+    "\n",
+    "import librosa\n",
+    "import torch\n",
+    "\n",
+    "import commons\n",
+    "import utils\n",
+    "from models import SynthesizerTrn\n",
+    "from text.symbols import symbols\n",
+    "from text import text_to_sequence\n",
+    "import numpy as np"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "G:\\AI\\so-vits-svc_v2\\VITS_WebUI\\monotonic_align\n",
+      "G:\\AI\\so-vits-svc_v2\\VITS_WebUI\n"
+     ]
+    }
+   ],
+   "source": [
+    "%cd G:\\AI\\so-vits-svc_v2\\VITS_WebUI\\monotonic_align\n",
+    "!python setup.py build_ext --inplace\n",
+    "%cd .."
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "outputs": [],
+   "source": [
+    "def resize2d(source, target_len):\n",
+    "    source[source<0.001] = np.nan\n",
+    "    target = np.interp(np.arange(0, len(source)*target_len, len(source))/ target_len, np.arange(0, len(source)), source)\n",
+    "    return np.nan_to_num(target)\n",
+    "def convert_wav_22050_to_f0(audio):\n",
+    "    tmp = librosa.pyin(audio,\n",
+    "                fmin=librosa.note_to_hz('C0'),\n",
+    "                fmax=librosa.note_to_hz('C7'),\n",
+    "                frame_length=1780)[0]\n",
+    "    f0 = np.zeros_like(tmp)\n",
+    "    f0[tmp>0] = tmp[tmp>0]\n",
+    "    return f0"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "outputs": [],
+   "source": [
+    "def get_text(text, hps):\n",
+    "    text_norm = text_to_sequence(text, hps.data.text_cleaners)\n",
+    "    if hps.data.add_blank:\n",
+    "        text_norm = commons.intersperse(text_norm, 0)\n",
+    "    text_norm = torch.LongTensor(text_norm)\n",
+    "    print(text_norm.shape)\n",
+    "    return text_norm"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "outputs": [],
+   "source": [
+    "CONFIG_PATH = \"configs/config.json\"\n",
+    "MODEL_PATH = \"models/Yuuka/Yuuka.pth\"\n",
+    "\n",
+    "hps = utils.get_hparams_from_file(CONFIG_PATH)\n",
+    "net_g_ms = SynthesizerTrn(\n",
+    "    len(hps.symbols),\n",
+    "    hps.data.filter_length // 2 + 1,\n",
+    "    hps.train.segment_size // hps.data.hop_length,\n",
+    "    n_speakers=hps.data.n_speakers,\n",
+    "    **hps.model).cuda()\n",
+    "\n"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "INFO:root:Loaded checkpoint 'models/Yuuka/Yuuka.pth' (iteration 445)\n"
+     ]
+    }
+   ],
+   "source": [
+    "_ = utils.load_checkpoint(MODEL_PATH, net_g_ms, None)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Using cache found in C:\\Users\\l4227/.cache\\torch\\hub\\bshall_hubert_main\n"
+     ]
+    }
+   ],
+   "source": [
+    "hubert = torch.hub.load(\"bshall/hubert:main\", \"hubert_soft\")"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "outputs": [],
+   "source": [
+    "def vc_fn(input_audio,vc_transform):\n",
+    "    if input_audio is None:\n",
+    "        return \"You need to upload an audio\", None\n",
+    "    sampling_rate, audio = input_audio\n",
+    "    # print(audio.shape,sampling_rate)\n",
+    "    duration = audio.shape[0] / sampling_rate\n",
+    "    if duration > 30:\n",
+    "        return \"Error: Audio is too long\", None\n",
+    "    audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32)\n",
+    "    if len(audio.shape) > 1:\n",
+    "        audio = librosa.to_mono(audio.transpose(1, 0))\n",
+    "    if sampling_rate != 16000:\n",
+    "        audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=16000)\n",
+    "\n",
+    "    audio22050 = librosa.resample(audio, orig_sr=16000, target_sr=22050)\n",
+    "    f0 = convert_wav_22050_to_f0(audio22050)\n",
+    "\n",
+    "    source = torch.FloatTensor(audio).unsqueeze(0).unsqueeze(0)\n",
+    "    print(source.shape)\n",
+    "    with torch.inference_mode():\n",
+    "        units = hubert.units(source)\n",
+    "        soft = units.squeeze(0).numpy()\n",
+    "        print(sampling_rate)\n",
+    "        f0 = resize2d(f0, len(soft[:, 0])) * vc_transform\n",
+    "        soft[:, 0] = f0 / 10\n",
+    "    sid = torch.LongTensor([0])\n",
+    "    stn_tst = torch.FloatTensor(soft)\n",
+    "    with torch.no_grad():\n",
+    "        x_tst = stn_tst.unsqueeze(0)\n",
+    "        x_tst_lengths = torch.LongTensor([stn_tst.size(0)])\n",
+    "        audio = net_g_ms.infer(x_tst, x_tst_lengths,sid=sid, noise_scale=0.1, noise_scale_w=0.1, length_scale=1)[0][\n",
+    "            0, 0].data.float().numpy()\n",
+    "\n",
+    "    return \"Success\", (hps.data.sampling_rate, audio)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): api.gradio.app:443\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): huggingface.co:443\n",
+      "DEBUG:asyncio:Using selector: SelectSelector\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): 127.0.0.1:7890\n",
+      "DEBUG:urllib3.connectionpool:http://127.0.0.1:7890 \"GET http://127.0.0.1:7861/startup-events HTTP/1.1\" 200 5\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): 127.0.0.1:7890\n",
+      "DEBUG:urllib3.connectionpool:http://127.0.0.1:7890 \"HEAD http://127.0.0.1:7861/ HTTP/1.1\" 200 0\n",
+      "Running on local URL:  http://127.0.0.1:7861\n",
+      "\n",
+      "To create a public link, set `share=True` in `launch()`.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": "<IPython.core.display.HTML object>",
+      "text/html": "<div><iframe src=\"http://127.0.0.1:7861/\" width=\"100%\" height=\"500\" allow=\"autoplay; camera; microphone; clipboard-read; clipboard-write;\" frameborder=\"0\" allowfullscreen></iframe></div>"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): api.gradio.app:443\n",
+      "DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): huggingface.co:443\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Traceback (most recent call last):\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\soundfile.py\", line 161, in <module>\n",
+      "    import _soundfile_data  # ImportError if this doesn't exist\n",
+      "ModuleNotFoundError: No module named '_soundfile_data'\n",
+      "\n",
+      "During handling of the above exception, another exception occurred:\n",
+      "\n",
+      "Traceback (most recent call last):\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\soundfile.py\", line 171, in <module>\n",
+      "    _snd = _ffi.dlopen(_libname)\n",
+      "OSError: cannot load library 'D:\\Anaconda\\envs\\ML\\Library\\bin\\sndfile.dll': error 0x7e\n",
+      "\n",
+      "During handling of the above exception, another exception occurred:\n",
+      "\n",
+      "Traceback (most recent call last):\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\routes.py\", line 394, in run_predict\n",
+      "    output = await app.get_blocks().process_api(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 1075, in process_api\n",
+      "    result = await self.call_function(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\gradio\\blocks.py\", line 884, in call_function\n",
+      "    prediction = await anyio.to_thread.run_sync(\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\to_thread.py\", line 28, in run_sync\n",
+      "    return await get_asynclib().run_sync_in_worker_thread(func, *args, cancellable=cancellable,\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 818, in run_sync_in_worker_thread\n",
+      "    return await future\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\anyio\\_backends\\_asyncio.py\", line 754, in run\n",
+      "    result = context.run(func, *args)\n",
+      "  File \"C:\\Users\\l4227\\AppData\\Local\\Temp\\ipykernel_23416\\731703501.py\", line 13, in vc_fn\n",
+      "    audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=16000)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\lazy_loader\\__init__.py\", line 77, in __getattr__\n",
+      "    attr = getattr(submod, name)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\lazy_loader\\__init__.py\", line 76, in __getattr__\n",
+      "    submod = importlib.import_module(submod_path)\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\importlib\\__init__.py\", line 127, in import_module\n",
+      "    return _bootstrap._gcd_import(name[level:], package, level)\n",
+      "  File \"<frozen importlib._bootstrap>\", line 1030, in _gcd_import\n",
+      "  File \"<frozen importlib._bootstrap>\", line 1007, in _find_and_load\n",
+      "  File \"<frozen importlib._bootstrap>\", line 986, in _find_and_load_unlocked\n",
+      "  File \"<frozen importlib._bootstrap>\", line 680, in _load_unlocked\n",
+      "  File \"<frozen importlib._bootstrap_external>\", line 850, in exec_module\n",
+      "  File \"<frozen importlib._bootstrap>\", line 228, in _call_with_frames_removed\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\librosa\\core\\audio.py\", line 10, in <module>\n",
+      "    import soundfile as sf\n",
+      "  File \"D:\\Anaconda\\envs\\ML\\lib\\site-packages\\soundfile.py\", line 192, in <module>\n",
+      "    _snd = _ffi.dlopen(_explicit_libname)\n",
+      "OSError: cannot load library 'libsndfile.dll': error 0x7e\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Keyboard interruption in main thread... closing server.\n"
+     ]
+    }
+   ],
+   "source": [
+    "app = gr.Blocks()\n",
+    "with app:\n",
+    "    with gr.Tabs():\n",
+    "        with gr.TabItem(\"Basic\"):\n",
+    "            vc_input3 = gr.Audio(label=\"Input Audio (30s limitation)\")\n",
+    "            vc_transform = gr.Number(label=\"transform\",value=1.0)\n",
+    "            vc_submit = gr.Button(\"Convert\", variant=\"primary\")\n",
+    "            vc_output1 = gr.Textbox(label=\"Output Message\")\n",
+    "            vc_output2 = gr.Audio(label=\"Output Audio\")\n",
+    "        vc_submit.click(vc_fn, [ vc_input3,vc_transform], [vc_output1, vc_output2])\n",
+    "\n",
+    "    app.launch(debug=True)"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [],
+   "metadata": {
+    "collapsed": false
+   }
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 0
+}

attentions.py

@@ -0,0 +1,303 @@
+import copy
+import math
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+import commons
+import modules
+from modules import LayerNorm
+   
+
+class Encoder(nn.Module):
+  def __init__(self, hidden_channels, filter_channels, n_heads, n_layers, kernel_size=1, p_dropout=0., window_size=4, **kwargs):
+    super().__init__()
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.window_size = window_size
+
+    self.drop = nn.Dropout(p_dropout)
+    self.attn_layers = nn.ModuleList()
+    self.norm_layers_1 = nn.ModuleList()
+    self.ffn_layers = nn.ModuleList()
+    self.norm_layers_2 = nn.ModuleList()
+    for i in range(self.n_layers):
+      self.attn_layers.append(MultiHeadAttention(hidden_channels, hidden_channels, n_heads, p_dropout=p_dropout, window_size=window_size))
+      self.norm_layers_1.append(LayerNorm(hidden_channels))
+      self.ffn_layers.append(FFN(hidden_channels, hidden_channels, filter_channels, kernel_size, p_dropout=p_dropout))
+      self.norm_layers_2.append(LayerNorm(hidden_channels))
+
+  def forward(self, x, x_mask):
+    attn_mask = x_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+    x = x * x_mask
+    for i in range(self.n_layers):
+      y = self.attn_layers[i](x, x, attn_mask)
+      y = self.drop(y)
+      x = self.norm_layers_1[i](x + y)
+
+      y = self.ffn_layers[i](x, x_mask)
+      y = self.drop(y)
+      x = self.norm_layers_2[i](x + y)
+    x = x * x_mask
+    return x
+
+
+class Decoder(nn.Module):
+  def __init__(self, hidden_channels, filter_channels, n_heads, n_layers, kernel_size=1, p_dropout=0., proximal_bias=False, proximal_init=True, **kwargs):
+    super().__init__()
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.proximal_bias = proximal_bias
+    self.proximal_init = proximal_init
+
+    self.drop = nn.Dropout(p_dropout)
+    self.self_attn_layers = nn.ModuleList()
+    self.norm_layers_0 = nn.ModuleList()
+    self.encdec_attn_layers = nn.ModuleList()
+    self.norm_layers_1 = nn.ModuleList()
+    self.ffn_layers = nn.ModuleList()
+    self.norm_layers_2 = nn.ModuleList()
+    for i in range(self.n_layers):
+      self.self_attn_layers.append(MultiHeadAttention(hidden_channels, hidden_channels, n_heads, p_dropout=p_dropout, proximal_bias=proximal_bias, proximal_init=proximal_init))
+      self.norm_layers_0.append(LayerNorm(hidden_channels))
+      self.encdec_attn_layers.append(MultiHeadAttention(hidden_channels, hidden_channels, n_heads, p_dropout=p_dropout))
+      self.norm_layers_1.append(LayerNorm(hidden_channels))
+      self.ffn_layers.append(FFN(hidden_channels, hidden_channels, filter_channels, kernel_size, p_dropout=p_dropout, causal=True))
+      self.norm_layers_2.append(LayerNorm(hidden_channels))
+
+  def forward(self, x, x_mask, h, h_mask):
+    """
+    x: decoder input
+    h: encoder output
+    """
+    self_attn_mask = commons.subsequent_mask(x_mask.size(2)).to(device=x.device, dtype=x.dtype)
+    encdec_attn_mask = h_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+    x = x * x_mask
+    for i in range(self.n_layers):
+      y = self.self_attn_layers[i](x, x, self_attn_mask)
+      y = self.drop(y)
+      x = self.norm_layers_0[i](x + y)
+
+      y = self.encdec_attn_layers[i](x, h, encdec_attn_mask)
+      y = self.drop(y)
+      x = self.norm_layers_1[i](x + y)
+      
+      y = self.ffn_layers[i](x, x_mask)
+      y = self.drop(y)
+      x = self.norm_layers_2[i](x + y)
+    x = x * x_mask
+    return x
+
+
+class MultiHeadAttention(nn.Module):
+  def __init__(self, channels, out_channels, n_heads, p_dropout=0., window_size=None, heads_share=True, block_length=None, proximal_bias=False, proximal_init=False):
+    super().__init__()
+    assert channels % n_heads == 0
+
+    self.channels = channels
+    self.out_channels = out_channels
+    self.n_heads = n_heads
+    self.p_dropout = p_dropout
+    self.window_size = window_size
+    self.heads_share = heads_share
+    self.block_length = block_length
+    self.proximal_bias = proximal_bias
+    self.proximal_init = proximal_init
+    self.attn = None
+
+    self.k_channels = channels // n_heads
+    self.conv_q = nn.Conv1d(channels, channels, 1)
+    self.conv_k = nn.Conv1d(channels, channels, 1)
+    self.conv_v = nn.Conv1d(channels, channels, 1)
+    self.conv_o = nn.Conv1d(channels, out_channels, 1)
+    self.drop = nn.Dropout(p_dropout)
+
+    if window_size is not None:
+      n_heads_rel = 1 if heads_share else n_heads
+      rel_stddev = self.k_channels**-0.5
+      self.emb_rel_k = nn.Parameter(torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels) * rel_stddev)
+      self.emb_rel_v = nn.Parameter(torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels) * rel_stddev)
+
+    nn.init.xavier_uniform_(self.conv_q.weight)
+    nn.init.xavier_uniform_(self.conv_k.weight)
+    nn.init.xavier_uniform_(self.conv_v.weight)
+    if proximal_init:
+      with torch.no_grad():
+        self.conv_k.weight.copy_(self.conv_q.weight)
+        self.conv_k.bias.copy_(self.conv_q.bias)
+      
+  def forward(self, x, c, attn_mask=None):
+    q = self.conv_q(x)
+    k = self.conv_k(c)
+    v = self.conv_v(c)
+    
+    x, self.attn = self.attention(q, k, v, mask=attn_mask)
+
+    x = self.conv_o(x)
+    return x
+
+  def attention(self, query, key, value, mask=None):
+    # reshape [b, d, t] -> [b, n_h, t, d_k]
+    b, d, t_s, t_t = (*key.size(), query.size(2))
+    query = query.view(b, self.n_heads, self.k_channels, t_t).transpose(2, 3)
+    key = key.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
+    value = value.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
+
+    scores = torch.matmul(query / math.sqrt(self.k_channels), key.transpose(-2, -1))
+    if self.window_size is not None:
+      assert t_s == t_t, "Relative attention is only available for self-attention."
+      key_relative_embeddings = self._get_relative_embeddings(self.emb_rel_k, t_s)
+      rel_logits = self._matmul_with_relative_keys(query /math.sqrt(self.k_channels), key_relative_embeddings)
+      scores_local = self._relative_position_to_absolute_position(rel_logits)
+      scores = scores + scores_local
+    if self.proximal_bias:
+      assert t_s == t_t, "Proximal bias is only available for self-attention."
+      scores = scores + self._attention_bias_proximal(t_s).to(device=scores.device, dtype=scores.dtype)
+    if mask is not None:
+      scores = scores.masked_fill(mask == 0, -1e4)
+      if self.block_length is not None:
+        assert t_s == t_t, "Local attention is only available for self-attention."
+        block_mask = torch.ones_like(scores).triu(-self.block_length).tril(self.block_length)
+        scores = scores.masked_fill(block_mask == 0, -1e4)
+    p_attn = F.softmax(scores, dim=-1) # [b, n_h, t_t, t_s]
+    p_attn = self.drop(p_attn)
+    output = torch.matmul(p_attn, value)
+    if self.window_size is not None:
+      relative_weights = self._absolute_position_to_relative_position(p_attn)
+      value_relative_embeddings = self._get_relative_embeddings(self.emb_rel_v, t_s)
+      output = output + self._matmul_with_relative_values(relative_weights, value_relative_embeddings)
+    output = output.transpose(2, 3).contiguous().view(b, d, t_t) # [b, n_h, t_t, d_k] -> [b, d, t_t]
+    return output, p_attn
+
+  def _matmul_with_relative_values(self, x, y):
+    """
+    x: [b, h, l, m]
+    y: [h or 1, m, d]
+    ret: [b, h, l, d]
+    """
+    ret = torch.matmul(x, y.unsqueeze(0))
+    return ret
+
+  def _matmul_with_relative_keys(self, x, y):
+    """
+    x: [b, h, l, d]
+    y: [h or 1, m, d]
+    ret: [b, h, l, m]
+    """
+    ret = torch.matmul(x, y.unsqueeze(0).transpose(-2, -1))
+    return ret
+
+  def _get_relative_embeddings(self, relative_embeddings, length):
+    max_relative_position = 2 * self.window_size + 1
+    # Pad first before slice to avoid using cond ops.
+    pad_length = max(length - (self.window_size + 1), 0)
+    slice_start_position = max((self.window_size + 1) - length, 0)
+    slice_end_position = slice_start_position + 2 * length - 1
+    if pad_length > 0:
+      padded_relative_embeddings = F.pad(
+          relative_embeddings,
+          commons.convert_pad_shape([[0, 0], [pad_length, pad_length], [0, 0]]))
+    else:
+      padded_relative_embeddings = relative_embeddings
+    used_relative_embeddings = padded_relative_embeddings[:,slice_start_position:slice_end_position]
+    return used_relative_embeddings
+
+  def _relative_position_to_absolute_position(self, x):
+    """
+    x: [b, h, l, 2*l-1]
+    ret: [b, h, l, l]
+    """
+    batch, heads, length, _ = x.size()
+    # Concat columns of pad to shift from relative to absolute indexing.
+    x = F.pad(x, commons.convert_pad_shape([[0,0],[0,0],[0,0],[0,1]]))
+
+    # Concat extra elements so to add up to shape (len+1, 2*len-1).
+    x_flat = x.view([batch, heads, length * 2 * length])
+    x_flat = F.pad(x_flat, commons.convert_pad_shape([[0,0],[0,0],[0,length-1]]))
+
+    # Reshape and slice out the padded elements.
+    x_final = x_flat.view([batch, heads, length+1, 2*length-1])[:, :, :length, length-1:]
+    return x_final
+
+  def _absolute_position_to_relative_position(self, x):
+    """
+    x: [b, h, l, l]
+    ret: [b, h, l, 2*l-1]
+    """
+    batch, heads, length, _ = x.size()
+    # padd along column
+    x = F.pad(x, commons.convert_pad_shape([[0, 0], [0, 0], [0, 0], [0, length-1]]))
+    x_flat = x.view([batch, heads, length**2 + length*(length -1)])
+    # add 0's in the beginning that will skew the elements after reshape
+    x_flat = F.pad(x_flat, commons.convert_pad_shape([[0, 0], [0, 0], [length, 0]]))
+    x_final = x_flat.view([batch, heads, length, 2*length])[:,:,:,1:]
+    return x_final
+
+  def _attention_bias_proximal(self, length):
+    """Bias for self-attention to encourage attention to close positions.
+    Args:
+      length: an integer scalar.
+    Returns:
+      a Tensor with shape [1, 1, length, length]
+    """
+    r = torch.arange(length, dtype=torch.float32)
+    diff = torch.unsqueeze(r, 0) - torch.unsqueeze(r, 1)
+    return torch.unsqueeze(torch.unsqueeze(-torch.log1p(torch.abs(diff)), 0), 0)
+
+
+class FFN(nn.Module):
+  def __init__(self, in_channels, out_channels, filter_channels, kernel_size, p_dropout=0., activation=None, causal=False):
+    super().__init__()
+    self.in_channels = in_channels
+    self.out_channels = out_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.activation = activation
+    self.causal = causal
+
+    if causal:
+      self.padding = self._causal_padding
+    else:
+      self.padding = self._same_padding
+
+    self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size)
+    self.conv_2 = nn.Conv1d(filter_channels, out_channels, kernel_size)
+    self.drop = nn.Dropout(p_dropout)
+
+  def forward(self, x, x_mask):
+    x = self.conv_1(self.padding(x * x_mask))
+    if self.activation == "gelu":
+      x = x * torch.sigmoid(1.702 * x)
+    else:
+      x = torch.relu(x)
+    x = self.drop(x)
+    x = self.conv_2(self.padding(x * x_mask))
+    return x * x_mask
+  
+  def _causal_padding(self, x):
+    if self.kernel_size == 1:
+      return x
+    pad_l = self.kernel_size - 1
+    pad_r = 0
+    padding = [[0, 0], [0, 0], [pad_l, pad_r]]
+    x = F.pad(x, commons.convert_pad_shape(padding))
+    return x
+
+  def _same_padding(self, x):
+    if self.kernel_size == 1:
+      return x
+    pad_l = (self.kernel_size - 1) // 2
+    pad_r = self.kernel_size // 2
+    padding = [[0, 0], [0, 0], [pad_l, pad_r]]
+    x = F.pad(x, commons.convert_pad_shape(padding))
+    return x

commons.py

@@ -0,0 +1,161 @@
+import math
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+
+def init_weights(m, mean=0.0, std=0.01):
+  classname = m.__class__.__name__
+  if classname.find("Conv") != -1:
+    m.weight.data.normal_(mean, std)
+
+
+def get_padding(kernel_size, dilation=1):
+  return int((kernel_size*dilation - dilation)/2)
+
+
+def convert_pad_shape(pad_shape):
+  l = pad_shape[::-1]
+  pad_shape = [item for sublist in l for item in sublist]
+  return pad_shape
+
+
+def intersperse(lst, item):
+  result = [item] * (len(lst) * 2 + 1)
+  result[1::2] = lst
+  return result
+
+
+def kl_divergence(m_p, logs_p, m_q, logs_q):
+  """KL(P||Q)"""
+  kl = (logs_q - logs_p) - 0.5
+  kl += 0.5 * (torch.exp(2. * logs_p) + ((m_p - m_q)**2)) * torch.exp(-2. * logs_q)
+  return kl
+
+
+def rand_gumbel(shape):
+  """Sample from the Gumbel distribution, protect from overflows."""
+  uniform_samples = torch.rand(shape) * 0.99998 + 0.00001
+  return -torch.log(-torch.log(uniform_samples))
+
+
+def rand_gumbel_like(x):
+  g = rand_gumbel(x.size()).to(dtype=x.dtype, device=x.device)
+  return g
+
+
+def slice_segments(x, ids_str, segment_size=4):
+  ret = torch.zeros_like(x[:, :, :segment_size])
+  for i in range(x.size(0)):
+    idx_str = ids_str[i]
+    idx_end = idx_str + segment_size
+    ret[i] = x[i, :, idx_str:idx_end]
+  return ret
+
+
+def rand_slice_segments(x, x_lengths=None, segment_size=4):
+  b, d, t = x.size()
+  if x_lengths is None:
+    x_lengths = t
+  ids_str_max = x_lengths - segment_size + 1
+  ids_str = (torch.rand([b]).to(device=x.device) * ids_str_max).to(dtype=torch.long)
+  ret = slice_segments(x, ids_str, segment_size)
+  return ret, ids_str
+
+
+def get_timing_signal_1d(
+    length, channels, min_timescale=1.0, max_timescale=1.0e4):
+  position = torch.arange(length, dtype=torch.float)
+  num_timescales = channels // 2
+  log_timescale_increment = (
+      math.log(float(max_timescale) / float(min_timescale)) /
+      (num_timescales - 1))
+  inv_timescales = min_timescale * torch.exp(
+      torch.arange(num_timescales, dtype=torch.float) * -log_timescale_increment)
+  scaled_time = position.unsqueeze(0) * inv_timescales.unsqueeze(1)
+  signal = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], 0)
+  signal = F.pad(signal, [0, 0, 0, channels % 2])
+  signal = signal.view(1, channels, length)
+  return signal
+
+
+def add_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4):
+  b, channels, length = x.size()
+  signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+  return x + signal.to(dtype=x.dtype, device=x.device)
+
+
+def cat_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4, axis=1):
+  b, channels, length = x.size()
+  signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+  return torch.cat([x, signal.to(dtype=x.dtype, device=x.device)], axis)
+
+
+def subsequent_mask(length):
+  mask = torch.tril(torch.ones(length, length)).unsqueeze(0).unsqueeze(0)
+  return mask
+
+
+@torch.jit.script
+def fused_add_tanh_sigmoid_multiply(input_a, input_b, n_channels):
+  n_channels_int = n_channels[0]
+  in_act = input_a + input_b
+  t_act = torch.tanh(in_act[:, :n_channels_int, :])
+  s_act = torch.sigmoid(in_act[:, n_channels_int:, :])
+  acts = t_act * s_act
+  return acts
+
+
+def convert_pad_shape(pad_shape):
+  l = pad_shape[::-1]
+  pad_shape = [item for sublist in l for item in sublist]
+  return pad_shape
+
+
+def shift_1d(x):
+  x = F.pad(x, convert_pad_shape([[0, 0], [0, 0], [1, 0]]))[:, :, :-1]
+  return x
+
+
+def sequence_mask(length, max_length=None):
+  if max_length is None:
+    max_length = length.max()
+  x = torch.arange(max_length, dtype=length.dtype, device=length.device)
+  return x.unsqueeze(0) < length.unsqueeze(1)
+
+
+def generate_path(duration, mask):
+  """
+  duration: [b, 1, t_x]
+  mask: [b, 1, t_y, t_x]
+  """
+  device = duration.device
+  
+  b, _, t_y, t_x = mask.shape
+  cum_duration = torch.cumsum(duration, -1)
+  
+  cum_duration_flat = cum_duration.view(b * t_x)
+  path = sequence_mask(cum_duration_flat, t_y).to(mask.dtype)
+  path = path.view(b, t_x, t_y)
+  path = path - F.pad(path, convert_pad_shape([[0, 0], [1, 0], [0, 0]]))[:, :-1]
+  path = path.unsqueeze(1).transpose(2,3) * mask
+  return path
+
+
+def clip_grad_value_(parameters, clip_value, norm_type=2):
+  if isinstance(parameters, torch.Tensor):
+    parameters = [parameters]
+  parameters = list(filter(lambda p: p.grad is not None, parameters))
+  norm_type = float(norm_type)
+  if clip_value is not None:
+    clip_value = float(clip_value)
+
+  total_norm = 0
+  for p in parameters:
+    param_norm = p.grad.data.norm(norm_type)
+    total_norm += param_norm.item() ** norm_type
+    if clip_value is not None:
+      p.grad.data.clamp_(min=-clip_value, max=clip_value)
+  total_norm = total_norm ** (1. / norm_type)
+  return total_norm

configs/config-single-speaker.json

@@ -0,0 +1,54 @@
+{
+  "train": {
+    "log_interval": 200,
+    "eval_interval": 1000,
+    "seed": 1234,
+    "epochs": 10000,
+    "learning_rate": 2e-5,
+    "betas": [0.8, 0.99],
+    "eps": 1e-9,
+    "batch_size": 16,
+    "fp16_run": true,
+    "lr_decay": 0.999875,
+    "segment_size": 8192,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0
+  },
+  "data": {
+    "training_files":"filelists/takina_train.txt.cleaned",
+    "validation_files":"filelists/takina_val.txt.cleaned",
+	"text_cleaners":["japanese_cleaners"],
+    "max_wav_value": 32768.0,
+    "sampling_rate": 22050,
+    "filter_length": 1024,
+    "hop_length": 256,
+    "win_length": 1024,
+    "n_mel_channels": 80,
+    "mel_fmin": 0.0,
+    "mel_fmax": null,
+    "add_blank": true,
+    "n_speakers": 0,
+    "cleaned_text": true
+  },
+  "model": {
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.1,
+    "resblock": "1",
+    "resblock_kernel_sizes": [3,7,11],
+    "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
+    "upsample_rates": [8,8,2,2],
+    "upsample_initial_channel": 512,
+    "upsample_kernel_sizes": [16,16,4,4],
+    "n_layers_q": 3,
+    "use_spectral_norm": false
+  },
+  "speakers": ["takina"],
+  "symbols": ["_", ",", ".", "!", "?", "-", "~", "\u2026", "A", "E", "I", "N", "O", "Q", "U", "a", "b", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o", "p", "r", "s", "t", "u", "v", "w", "y", "z", "\u0283", "\u02a7", "\u02a6", "\u026f", "\u0279", "\u0259", "\u0265", "\u207c", "\u02b0", "`", "\u2192", "\u2193", "\u2191", " "]
+}

configs/config.json

@@ -0,0 +1,948 @@
+{
+    "train": {
+        "log_interval": 200,
+        "eval_interval": 1000,
+        "seed": 1234,
+        "epochs": 10000,
+        "learning_rate": 0.0002,
+        "betas": [
+            0.8,
+            0.99
+        ],
+        "eps": 1e-09,
+        "batch_size": 16,
+        "fp16_run": true,
+        "lr_decay": 0.999875,
+        "segment_size": 8192,
+        "init_lr_ratio": 1,
+        "warmup_epochs": 0,
+        "c_mel": 45,
+        "c_kl": 1.0
+    },
+    "data": {
+        "training_files": "filelists/yuuka_train.txt.cleaned",
+        "validation_files": "filelists/yuuka_val.txt.cleaned",
+        "text_cleaners": [
+            "japanese_cleaners"
+        ],
+        "max_wav_value": 32768.0,
+        "sampling_rate": 22050,
+        "filter_length": 1024,
+        "hop_length": 256,
+        "win_length": 1024,
+        "n_mel_channels": 80,
+        "mel_fmin": 0.0,
+        "mel_fmax": null,
+        "add_blank": true,
+        "n_speakers": 804,
+        "cleaned_text": true
+    },
+    "model": {
+        "inter_channels": 192,
+        "hidden_channels": 192,
+        "filter_channels": 768,
+        "n_heads": 2,
+        "n_layers": 6,
+        "kernel_size": 3,
+        "p_dropout": 0.1,
+        "resblock": "1",
+        "resblock_kernel_sizes": [
+            3,
+            7,
+            11
+        ],
+        "resblock_dilation_sizes": [
+            [
+                1,
+                3,
+                5
+            ],
+            [
+                1,
+                3,
+                5
+            ],
+            [
+                1,
+                3,
+                5
+            ]
+        ],
+        "upsample_rates": [
+            8,
+            8,
+            2,
+            2
+        ],
+        "upsample_initial_channel": 512,
+        "upsample_kernel_sizes": [
+            16,
+            16,
+            4,
+            4
+        ],
+        "n_layers_q": 3,
+        "use_spectral_norm": false,
+        "gin_channels": 256
+    },
+    "speakers": [
+        "\u7279\u522b\u5468",
+        "\u65e0\u58f0\u94c3\u9e7f",
+        "\u4e1c\u6d77\u5e1d\u7687\uff08\u5e1d\u5b9d\uff0c\u5e1d\u738b\uff09",
+        "\u4e38\u5584\u65af\u57fa",
+        "\u5bcc\u58eb\u5947\u8ff9",
+        "\u5c0f\u6817\u5e3d",
+        "\u9ec4\u91d1\u8239",
+        "\u4f0f\u7279\u52a0",
+        "\u5927\u548c\u8d64\u9aa5",
+        "\u5927\u6811\u5feb\u8f66",
+        "\u8349\u4e0a\u98de",
+        "\u83f1\u4e9a\u9a6c\u900a",
+        "\u76ee\u767d\u9ea6\u6606",
+        "\u795e\u9e70",
+        "\u597d\u6b4c\u5267",
+        "\u6210\u7530\u767d\u4ec1",
+        "\u9c81\u9053\u592b\u8c61\u5f81\uff08\u7687\u5e1d\uff09",
+        "\u6c14\u69fd",
+        "\u7231\u4e3d\u6570\u7801",
+        "\u661f\u4e91\u5929\u7a7a",
+        "\u7389\u85fb\u5341\u5b57",
+        "\u7f8e\u5999\u59ff\u52bf",
+        "\u7435\u7436\u6668\u5149",
+        "\u6469\u8036\u91cd\u70ae",
+        "\u66fc\u57ce\u8336\u5ea7",
+        "\u7f8e\u6d66\u6ce2\u65c1",
+        "\u76ee\u767d\u8d56\u6069",
+        "\u83f1\u66d9",
+        "\u96ea\u4e2d\u7f8e\u4eba",
+        "\u7c73\u6d74",
+        "\u827e\u5c3c\u65af\u98ce\u795e",
+        "\u7231\u4e3d\u901f\u5b50\uff08\u7231\u4e3d\u5feb\u5b50\uff09",
+        "\u7231\u6155\u7ec7\u59ec",
+        "\u7a3b\u8377\u4e00",
+        "\u80dc\u5229\u5956\u5238",
+        "\u7a7a\u4e2d\u795e\u5bab",
+        "\u8363\u8fdb\u95ea\u8000",
+        "\u771f\u673a\u4f36",
+        "\u5ddd\u4e0a\u516c\u4e3b",
+        "\u9ec4\u91d1\u57ce\uff08\u9ec4\u91d1\u57ce\u5e02\uff09",
+        "\u6a31\u82b1\u8fdb\u738b",
+        "\u91c7\u73e0",
+        "\u65b0\u5149\u98ce",
+        "\u4e1c\u5546\u53d8\u9769",
+        "\u8d85\u7ea7\u5c0f\u6d77\u6e7e",
+        "\u9192\u76ee\u98de\u9e70\uff08\u5bc4\u5bc4\u5b50\uff09",
+        "\u8352\u6f20\u82f1\u96c4",
+        "\u4e1c\u701b\u4f50\u6566",
+        "\u4e2d\u5c71\u5e86\u5178",
+        "\u6210\u7530\u5927\u8fdb",
+        "\u897f\u91ce\u82b1",
+        "\u6625\u4e3d\uff08\u4e4c\u62c9\u62c9\uff09",
+        "\u9752\u7af9\u56de\u5fc6",
+        "\u5fae\u5149\u98de\u9a79",
+        "\u7f8e\u4e3d\u5468\u65e5",
+        "\u5f85\u517c\u798f\u6765",
+        "mr cb\uff08cb\u5148\u751f\uff09",
+        "\u540d\u5c06\u6012\u6d9b\uff08\u540d\u5c06\u6237\u4ec1\uff09",
+        "\u76ee\u767d\u591a\u4f2f",
+        "\u4f18\u79c0\u7d20\u8d28",
+        "\u5e1d\u738b\u5149\u8f89",
+        "\u5f85\u517c\u8bd7\u6b4c\u5267",
+        "\u751f\u91ce\u72c4\u675c\u65af",
+        "\u76ee\u767d\u5584\u4fe1",
+        "\u5927\u62d3\u592a\u9633\u795e",
+        "\u53cc\u6da1\u8f6e\uff08\u4e24\u7acb\u76f4\uff0c\u4e24\u55b7\u5c04\uff0c\u4e8c\u9505\u5934\uff0c\u9006\u55b7\u5c04\uff09",
+        "\u91cc\u89c1\u5149\u94bb\uff08\u8428\u6258\u8bfa\u91d1\u521a\u77f3\uff09",
+        "\u5317\u90e8\u7384\u9a79",
+        "\u6a31\u82b1\u5343\u4ee3\u738b",
+        "\u5929\u72fc\u661f\u8c61\u5f81",
+        "\u76ee\u767d\u963f\u5c14\u4e39",
+        "\u516b\u91cd\u65e0\u654c",
+        "\u9e64\u4e38\u521a\u5fd7",
+        "\u76ee\u767d\u5149\u660e",
+        "\u6210\u7530\u62dc\u4ec1\uff08\u6210\u7530\u8def\uff09",
+        "\u4e5f\u6587\u6444\u8f89",
+        "\u5c0f\u6797\u5386\u5947",
+        "\u5317\u6e2f\u706b\u5c71",
+        "\u5947\u9510\u9a8f",
+        "\u82e6\u6da9\u7cd6\u971c",
+        "\u5c0f\u5c0f\u8695\u8327",
+        "\u9a8f\u5ddd\u624b\u7eb2\uff08\u7eff\u5e3d\u6076\u9b54\uff09",
+        "\u79cb\u5ddd\u5f25\u751f\uff08\u5c0f\u5c0f\u7406\u4e8b\u957f\uff09",
+        "\u4e59\u540d\u53f2\u60a6\u5b50\uff08\u4e59\u540d\u8bb0\u8005\uff09",
+        "\u6850\u751f\u9662\u8475",
+        "\u5b89\u5fc3\u6cfd\u523a\u523a\u7f8e",
+        "\u6a2b\u672c\u7406\u5b50",
+        "\u795e\u91cc\u7eeb\u534e\uff08\u9f9f\u9f9f\uff09",
+        "\u7434",
+        "\u7a7a\uff08\u7a7a\u54e5\uff09",
+        "\u4e3d\u838e",
+        "\u8367\uff08\u8367\u59b9\uff09",
+        "\u82ad\u82ad\u62c9",
+        "\u51ef\u4e9a",
+        "\u8fea\u5362\u514b",
+        "\u96f7\u6cfd",
+        "\u5b89\u67cf",
+        "\u6e29\u8fea",
+        "\u9999\u83f1",
+        "\u5317\u6597",
+        "\u884c\u79cb",
+        "\u9b48",
+        "\u51dd\u5149",
+        "\u53ef\u8389",
+        "\u949f\u79bb",
+        "\u83f2\u8c22\u5c14\uff08\u7687\u5973\uff09",
+        "\u73ed\u5c3c\u7279",
+        "\u8fbe\u8fbe\u5229\u4e9a\uff08\u516c\u5b50\uff09",
+        "\u8bfa\u827e\u5c14\uff08\u5973\u4ec6\uff09",
+        "\u4e03\u4e03",
+        "\u91cd\u4e91",
+        "\u7518\u96e8\uff08\u6930\u7f8a\uff09",
+        "\u963f\u8d1d\u591a",
+        "\u8fea\u5965\u5a1c\uff08\u732b\u732b\uff09",
+        "\u83ab\u5a1c",
+        "\u523b\u6674",
+        "\u7802\u7cd6",
+        "\u8f9b\u7131",
+        "\u7f57\u838e\u8389\u4e9a",
+        "\u80e1\u6843",
+        "\u67ab\u539f\u4e07\u53f6\uff08\u4e07\u53f6\uff09",
+        "\u70df\u7eef",
+        "\u5bb5\u5bab",
+        "\u6258\u9a6c",
+        "\u4f18\u83c8",
+        "\u96f7\u7535\u5c06\u519b\uff08\u96f7\u795e\uff09",
+        "\u65e9\u67da",
+        "\u73ca\u745a\u5bab\u5fc3\u6d77\uff08\u5fc3\u6d77\uff0c\u6263\u6263\u7c73\uff09",
+        "\u4e94\u90ce",
+        "\u4e5d\u6761\u88df\u7f57",
+        "\u8352\u6cf7\u4e00\u6597\uff08\u4e00\u6597\uff09",
+        "\u57c3\u6d1b\u4f0a",
+        "\u7533\u9e64",
+        "\u516b\u91cd\u795e\u5b50\uff08\u795e\u5b50\uff09",
+        "\u795e\u91cc\u7eeb\u4eba\uff08\u7eeb\u4eba\uff09",
+        "\u591c\u5170",
+        "\u4e45\u5c90\u5fcd",
+        "\u9e7f\u91ce\u82d1\u5e73\u85cf",
+        "\u63d0\u7eb3\u91cc",
+        "\u67ef\u83b1",
+        "\u591a\u8389",
+        "\u4e91\u5807",
+        "\u7eb3\u897f\u59b2\uff08\u8349\u795e\uff09",
+        "\u6df1\u6e0a\u4f7f\u5f92",
+        "\u59ae\u9732",
+        "\u8d5b\u8bfa",
+        "\u503a\u52a1\u5904\u7406\u4eba",
+        "\u574e\u8482\u4e1d",
+        "\u771f\u5f13\u5feb\u8f66",
+        "\u79cb\u4eba",
+        "\u671b\u65cf",
+        "\u827e\u5c14\u83f2",
+        "\u827e\u8389\u4e1d",
+        "\u827e\u4f26",
+        "\u963f\u6d1b\u74e6",
+        "\u5929\u91ce",
+        "\u5929\u76ee\u5341\u4e94",
+        "\u611a\u4eba\u4f17-\u5b89\u5fb7\u70c8",
+        "\u5b89\u987a",
+        "\u5b89\u897f",
+        "\u8475",
+        "\u9752\u6728",
+        "\u8352\u5ddd\u5e78\u6b21",
+        "\u8352\u8c37",
+        "\u6709\u6cfd",
+        "\u6d45\u5ddd",
+        "\u9ebb\u7f8e",
+        "\u51dd\u5149\u52a9\u624b",
+        "\u963f\u6258",
+        "\u7afa\u5b50",
+        "\u767e\u8bc6",
+        "\u767e\u95fb",
+        "\u767e\u6653",
+        "\u767d\u672f",
+        "\u8d1d\u96c5\u7279\u4e3d\u5947",
+        "\u4e3d\u5854",
+        "\u5931\u843d\u8ff7\u8fed",
+        "\u7f2d\u4e71\u661f\u68d8",
+        "\u4f0a\u7538",
+        "\u4f0f\u7279\u52a0\u5973\u5b69",
+        "\u72c2\u70ed\u84dd\u8c03",
+        "\u8389\u8389\u5a05",
+        "\u841d\u838e\u8389\u5a05",
+        "\u516b\u91cd\u6a31",
+        "\u516b\u91cd\u971e",
+        "\u5361\u83b2",
+        "\u7b2c\u516d\u591c\u60f3\u66f2",
+        "\u5361\u841d\u5c14",
+        "\u59ec\u5b50",
+        "\u6781\u5730\u6218\u5203",
+        "\u5e03\u6d1b\u59ae\u5a05",
+        "\u6b21\u751f\u94f6\u7ffc",
+        "\u7406\u4e4b\u5f8b\u8005%26\u5e0c\u513f",
+        "\u7406\u4e4b\u5f8b\u8005",
+        "\u8ff7\u57ce\u9a87\u5154",
+        "\u5e0c\u513f",
+        "\u9b47\u591c\u661f\u6e0a",
+        "\u9ed1\u5e0c\u513f",
+        "\u5e15\u6735\u83f2\u8389\u4e1d",
+        "\u4e0d\u706d\u661f\u951a",
+        "\u5929\u5143\u9a91\u82f1",
+        "\u5e7d\u5170\u9edb\u5c14",
+        "\u6d3e\u8499bh3",
+        "\u7231\u9171",
+        "\u7eef\u7389\u4e38",
+        "\u5fb7\u4e3d\u838e",
+        "\u6708\u4e0b\u521d\u62e5",
+        "\u6714\u591c\u89c2\u661f",
+        "\u66ae\u5149\u9a91\u58eb",
+        "\u683c\u857e\u4fee",
+        "\u7559\u4e91\u501f\u98ce\u771f\u541b",
+        "\u6885\u6bd4\u4e4c\u65af",
+        "\u4eff\u72b9\u5927",
+        "\u514b\u83b1\u56e0",
+        "\u5723\u5251\u5e7d\u5170\u9edb\u5c14",
+        "\u5996\u7cbe\u7231\u8389",
+        "\u7279\u65af\u62c9zero",
+        "\u82cd\u7384",
+        "\u82e5\u6c34",
+        "\u897f\u7433",
+        "\u6234\u56e0\u65af\u96f7\u5e03",
+        "\u8d1d\u62c9",
+        "\u8d64\u9e22",
+        "\u9547\u9b42\u6b4c",
+        "\u6e21\u9e26",
+        "\u4eba\u4e4b\u5f8b\u8005",
+        "\u7231\u8389\u5e0c\u96c5",
+        "\u5929\u7a79\u6e38\u4fa0",
+        "\u742a\u4e9a\u5a1c",
+        "\u7a7a\u4e4b\u5f8b\u8005",
+        "\u85aa\u708e\u4e4b\u5f8b\u8005",
+        "\u4e91\u58a8\u4e39\u5fc3",
+        "\u7b26\u534e",
+        "\u8bc6\u4e4b\u5f8b\u8005",
+        "\u7279\u74e6\u6797",
+        "\u7ef4\u5c14\u8587",
+        "\u82bd\u8863",
+        "\u96f7\u4e4b\u5f8b\u8005",
+        "\u65ad\u7f6a\u5f71\u821e",
+        "\u963f\u6ce2\u5c3c\u4e9a",
+        "\u698e\u672c",
+        "\u5384\u5c3c\u65af\u7279",
+        "\u6076\u9f99",
+        "\u8303\u4e8c\u7237",
+        "\u6cd5\u62c9",
+        "\u611a\u4eba\u4f17\u58eb\u5175",
+        "\u611a\u4eba\u4f17\u58eb\u5175a",
+        "\u611a\u4eba\u4f17\u58eb\u5175b",
+        "\u611a\u4eba\u4f17\u58eb\u5175c",
+        "\u611a\u4eba\u4f17a",
+        "\u611a\u4eba\u4f17b",
+        "\u98de\u98de",
+        "\u83f2\u5229\u514b\u65af",
+        "\u5973\u6027\u8ddf\u968f\u8005",
+        "\u9022\u5ca9",
+        "\u6446\u6e21\u4eba",
+        "\u72c2\u8e81\u7684\u7537\u4eba",
+        "\u5965\u5179",
+        "\u8299\u841d\u62c9",
+        "\u8ddf\u968f\u8005",
+        "\u871c\u6c41\u751f\u7269",
+        "\u9ec4\u9ebb\u5b50",
+        "\u6e0a\u4e0a",
+        "\u85e4\u6728",
+        "\u6df1\u89c1",
+        "\u798f\u672c",
+        "\u8299\u84c9",
+        "\u53e4\u6cfd",
+        "\u53e4\u7530",
+        "\u53e4\u5c71",
+        "\u53e4\u8c37\u6607",
+        "\u5085\u4e09\u513f",
+        "\u9ad8\u8001\u516d",
+        "\u77ff\u5de5\u5192",
+        "\u5143\u592a",
+        "\u5fb7\u5b89\u516c",
+        "\u8302\u624d\u516c",
+        "\u6770\u62c9\u5fb7",
+        "\u845b\u7f57\u4e3d",
+        "\u91d1\u5ffd\u5f8b",
+        "\u516c\u4fca",
+        "\u9505\u5df4",
+        "\u6b4c\u5fb7",
+        "\u963f\u8c6a",
+        "\u72d7\u4e09\u513f",
+        "\u845b\u745e\u4e1d",
+        "\u82e5\u5fc3",
+        "\u963f\u5c71\u5a46",
+        "\u602a\u9e1f",
+        "\u5e7f\u7af9",
+        "\u89c2\u6d77",
+        "\u5173\u5b8f",
+        "\u871c\u6c41\u536b\u5175",
+        "\u5b88\u536b1",
+        "\u50b2\u6162\u7684\u5b88\u536b",
+        "\u5bb3\u6015\u7684\u5b88\u536b",
+        "\u8d35\u5b89",
+        "\u76d6\u4f0a",
+        "\u963f\u521b",
+        "\u54c8\u592b\u4e39",
+        "\u65e5\u8bed\u963f\u8d1d\u591a\uff08\u91ce\u5c9b\u5065\u513f\uff09",
+        "\u65e5\u8bed\u57c3\u6d1b\u4f0a\uff08\u9ad8\u57a3\u5f69\u9633\uff09",
+        "\u65e5\u8bed\u5b89\u67cf\uff08\u77f3\u89c1\u821e\u83dc\u9999\uff09",
+        "\u65e5\u8bed\u795e\u91cc\u7eeb\u534e\uff08\u65e9\u89c1\u6c99\u7ec7\uff09",
+        "\u65e5\u8bed\u795e\u91cc\u7eeb\u4eba\uff08\u77f3\u7530\u5f70\uff09",
+        "\u65e5\u8bed\u767d\u672f\uff08\u6e38\u4f50\u6d69\u4e8c\uff09",
+        "\u65e5\u8bed\u82ad\u82ad\u62c9\uff08\u9b3c\u5934\u660e\u91cc\uff09",
+        "\u65e5\u8bed\u5317\u6597\uff08\u5c0f\u6e05\u6c34\u4e9a\u7f8e\uff09",
+        "\u65e5\u8bed\u73ed\u5c3c\u7279\uff08\u9022\u5742\u826f\u592a\uff09",
+        "\u65e5\u8bed\u574e\u8482\u4e1d\uff08\u67da\u6728\u51c9\u9999\uff09",
+        "\u65e5\u8bed\u91cd\u4e91\uff08\u9f50\u85e4\u58ee\u9a6c\uff09",
+        "\u65e5\u8bed\u67ef\u83b1\uff08\u524d\u5ddd\u51c9\u5b50\uff09",
+        "\u65e5\u8bed\u8d5b\u8bfa\uff08\u5165\u91ce\u81ea\u7531\uff09",
+        "\u65e5\u8bed\u6234\u56e0\u65af\u96f7\u5e03\uff08\u6d25\u7530\u5065\u6b21\u90ce\uff09",
+        "\u65e5\u8bed\u8fea\u5362\u514b\uff08\u5c0f\u91ce\u8d24\u7ae0\uff09",
+        "\u65e5\u8bed\u8fea\u5965\u5a1c\uff08\u4e95\u6cfd\u8bd7\u7ec7\uff09",
+        "\u65e5\u8bed\u591a\u8389\uff08\u91d1\u7530\u670b\u5b50\uff09",
+        "\u65e5\u8bed\u4f18\u83c8\uff08\u4f50\u85e4\u5229\u5948\uff09",
+        "\u65e5\u8bed\u83f2\u8c22\u5c14\uff08\u5185\u7530\u771f\u793c\uff09",
+        "\u65e5\u8bed\u7518\u96e8\uff08\u4e0a\u7530\u4e3d\u5948\uff09",
+        "\u65e5\u8bed\uff08\u7560\u4e2d\u7950\uff09",
+        "\u65e5\u8bed\u9e7f\u91ce\u9662\u5e73\u85cf\uff08\u4e95\u53e3\u7950\u4e00\uff09",
+        "\u65e5\u8bed\u7a7a\uff08\u5800\u6c5f\u77ac\uff09",
+        "\u65e5\u8bed\u8367\uff08\u60a0\u6728\u78a7\uff09",
+        "\u65e5\u8bed\u80e1\u6843\uff08\u9ad8\u6865\u674e\u4f9d\uff09",
+        "\u65e5\u8bed\u4e00\u6597\uff08\u897f\u5ddd\u8d35\u6559\uff09",
+        "\u65e5\u8bed\u51ef\u4e9a\uff08\u9e1f\u6d77\u6d69\u8f85\uff09",
+        "\u65e5\u8bed\u4e07\u53f6\uff08\u5c9b\u5d0e\u4fe1\u957f\uff09",
+        "\u65e5\u8bed\u523b\u6674\uff08\u559c\u591a\u6751\u82f1\u68a8\uff09",
+        "\u65e5\u8bed\u53ef\u8389\uff08\u4e45\u91ce\u7f8e\u54b2\uff09",
+        "\u65e5\u8bed\u5fc3\u6d77\uff08\u4e09\u68ee\u94c3\u5b50\uff09",
+        "\u65e5\u8bed\u4e5d\u6761\u88df\u7f57\uff08\u6fd1\u6237\u9ebb\u6c99\u7f8e\uff09",
+        "\u65e5\u8bed\u4e3d\u838e\uff08\u7530\u4e2d\u7406\u60e0\uff09",
+        "\u65e5\u8bed\u83ab\u5a1c\uff08\u5c0f\u539f\u597d\u7f8e\uff09",
+        "\u65e5\u8bed\u7eb3\u897f\u59b2\uff08\u7530\u6751\u7531\u52a0\u8389\uff09",
+        "\u65e5\u8bed\u59ae\u9732\uff08\u91d1\u5143\u5bff\u5b50\uff09",
+        "\u65e5\u8bed\u51dd\u5149\uff08\u5927\u539f\u6c99\u8036\u9999\uff09",
+        "\u65e5\u8bed\u8bfa\u827e\u5c14\uff08\u9ad8\u5c3e\u594f\u97f3\uff09",
+        "\u65e5\u8bed\u5965\u5179\uff08\u589e\u8c37\u5eb7\u7eaa\uff09",
+        "\u65e5\u8bed\u6d3e\u8499\uff08\u53e4\u8d3a\u8475\uff09",
+        "\u65e5\u8bed\u7434\uff08\u658b\u85e4\u5343\u548c\uff09",
+        "\u65e5\u8bed\u4e03\u4e03\uff08\u7530\u6751\u7531\u52a0\u8389\uff09",
+        "\u65e5\u8bed\u96f7\u7535\u5c06\u519b\uff08\u6cfd\u57ce\u7f8e\u96ea\uff09",
+        "\u65e5\u8bed\u96f7\u6cfd\uff08\u5185\u5c71\u6602\u8f89\uff09",
+        "\u65e5\u8bed\u7f57\u838e\u8389\u4e9a\uff08\u52a0\u9688\u4e9a\u8863\uff09",
+        "\u65e5\u8bed\u65e9\u67da\uff08\u6d32\u5d0e\u7eeb\uff09",
+        "\u65e5\u8bed\u6563\u5175\uff08\u67ff\u539f\u5f7b\u4e5f\uff09",
+        "\u65e5\u8bed\u7533\u9e64\uff08\u5ddd\u6f84\u7eeb\u5b50\uff09",
+        "\u65e5\u8bed\u4e45\u5c90\u5fcd\uff08\u6c34\u6865\u9999\u7ec7\uff09",
+        "\u65e5\u8bed\u5973\u58eb\uff08\u5e84\u5b50\u88d5\u8863\uff09",
+        "\u65e5\u8bed\u7802\u7cd6\uff08\u85e4\u7530\u831c\uff09",
+        "\u65e5\u8bed\u8fbe\u8fbe\u5229\u4e9a\uff08\u6728\u6751\u826f\u5e73\uff09",
+        "\u65e5\u8bed\u6258\u9a6c\uff08\u68ee\u7530\u6210\u4e00\uff09",
+        "\u65e5\u8bed\u63d0\u7eb3\u91cc\uff08\u5c0f\u6797\u6c99\u82d7\uff09",
+        "\u65e5\u8bed\u6e29\u8fea\uff08\u6751\u6fd1\u6b65\uff09",
+        "\u65e5\u8bed\u9999\u83f1\uff08\u5c0f\u6cfd\u4e9a\u674e\uff09",
+        "\u65e5\u8bed\u9b48\uff08\u677e\u5188\u796f\u4e1e\uff09",
+        "\u65e5\u8bed\u884c\u79cb\uff08\u7686\u5ddd\u7eaf\u5b50\uff09",
+        "\u65e5\u8bed\u8f9b\u7131\uff08\u9ad8\u6865\u667a\u79cb\uff09",
+        "\u65e5\u8bed\u516b\u91cd\u795e\u5b50\uff08\u4f50\u4ed3\u7eeb\u97f3\uff09",
+        "\u65e5\u8bed\u70df\u7eef\uff08\u82b1\u5b88\u7531\u7f8e\u91cc\uff09",
+        "\u65e5\u8bed\u591c\u5170\uff08\u8fdc\u85e4\u7eeb\uff09",
+        "\u65e5\u8bed\u5bb5\u5bab\uff08\u690d\u7530\u4f73\u5948\uff09",
+        "\u65e5\u8bed\u4e91\u5807\uff08\u5c0f\u5ca9\u4e95\u5c0f\u9e1f\uff09",
+        "\u65e5\u8bed\u949f\u79bb\uff08\u524d\u91ce\u667a\u662d\uff09",
+        "\u6770\u514b",
+        "\u963f\u5409",
+        "\u6c5f\u821f",
+        "\u9274\u79cb",
+        "\u5609\u4e49",
+        "\u7eaa\u82b3",
+        "\u666f\u6f84",
+        "\u7ecf\u7eb6",
+        "\u666f\u660e",
+        "\u664b\u4f18",
+        "\u963f\u9e20",
+        "\u9152\u5ba2",
+        "\u4e54\u5c14",
+        "\u4e54\u745f\u592b",
+        "\u7ea6\u987f",
+        "\u4e54\u4f0a\u65af",
+        "\u5c45\u5b89",
+        "\u541b\u541b",
+        "\u987a\u5409",
+        "\u7eaf\u4e5f",
+        "\u91cd\u4f50",
+        "\u5927\u5c9b\u7eaf\u5e73",
+        "\u84b2\u6cfd",
+        "\u52d8\u89e3\u7531\u5c0f\u8def\u5065\u4e09\u90ce",
+        "\u67ab",
+        "\u67ab\u539f\u4e49\u5e86",
+        "\u836b\u5c71",
+        "\u7532\u6590\u7530\u9f8d\u99ac",
+        "\u6d77\u6597",
+        "\u60df\u795e\u6674\u4e4b\u4ecb",
+        "\u9e7f\u91ce\u5948\u5948",
+        "\u5361\u7435\u8389\u4e9a",
+        "\u51ef\u745f\u7433",
+        "\u52a0\u85e4\u4fe1\u609f",
+        "\u52a0\u85e4\u6d0b\u5e73",
+        "\u80dc\u5bb6",
+        "\u8305\u847a\u4e00\u5e86",
+        "\u548c\u662d",
+        "\u4e00\u6b63",
+        "\u4e00\u9053",
+        "\u6842\u4e00",
+        "\u5e86\u6b21\u90ce",
+        "\u963f\u8d24",
+        "\u5065\u53f8",
+        "\u5065\u6b21\u90ce",
+        "\u5065\u4e09\u90ce",
+        "\u5929\u7406",
+        "\u6740\u624ba",
+        "\u6740\u624bb",
+        "\u6728\u5357\u674f\u5948",
+        "\u6728\u6751",
+        "\u56fd\u738b",
+        "\u6728\u4e0b",
+        "\u5317\u6751",
+        "\u6e05\u60e0",
+        "\u6e05\u4eba",
+        "\u514b\u5217\u95e8\u7279",
+        "\u9a91\u58eb",
+        "\u5c0f\u6797",
+        "\u5c0f\u6625",
+        "\u5eb7\u62c9\u5fb7",
+        "\u5927\u8089\u4e38",
+        "\u7434\u7f8e",
+        "\u5b8f\u4e00",
+        "\u5eb7\u4ecb",
+        "\u5e78\u5fb7",
+        "\u9ad8\u5584",
+        "\u68a2",
+        "\u514b\u7f57\u7d22",
+        "\u4e45\u4fdd",
+        "\u4e5d\u6761\u9570\u6cbb",
+        "\u4e45\u6728\u7530",
+        "\u6606\u94a7",
+        "\u83ca\u5730\u541b",
+        "\u4e45\u5229\u987b",
+        "\u9ed1\u7530",
+        "\u9ed1\u6cfd\u4eac\u4e4b\u4ecb",
+        "\u54cd\u592a",
+        "\u5c9a\u59d0",
+        "\u5170\u6eaa",
+        "\u6f9c\u9633",
+        "\u52b3\u4f26\u65af",
+        "\u4e50\u660e",
+        "\u83b1\u8bfa",
+        "\u83b2",
+        "\u826f\u5b50",
+        "\u674e\u5f53",
+        "\u674e\u4e01",
+        "\u5c0f\u4e50",
+        "\u7075",
+        "\u5c0f\u73b2",
+        "\u7433\u7405a",
+        "\u7433\u7405b",
+        "\u5c0f\u5f6c",
+        "\u5c0f\u5fb7",
+        "\u5c0f\u697d",
+        "\u5c0f\u9f99",
+        "\u5c0f\u5434",
+        "\u5c0f\u5434\u7684\u8bb0\u5fc6",
+        "\u7406\u6b63",
+        "\u963f\u9f99",
+        "\u5362\u5361",
+        "\u6d1b\u6210",
+        "\u7f57\u5de7",
+        "\u5317\u98ce\u72fc",
+        "\u5362\u6b63",
+        "\u840d\u59e5\u59e5",
+        "\u524d\u7530",
+        "\u771f\u663c",
+        "\u9ebb\u7eaa",
+        "\u771f",
+        "\u611a\u4eba\u4f17-\u9a6c\u514b\u897f\u59c6",
+        "\u5973\u6027a",
+        "\u5973\u6027b",
+        "\u5973\u6027a\u7684\u8ddf\u968f\u8005",
+        "\u963f\u5b88",
+        "\u739b\u683c\u4e3d\u7279",
+        "\u771f\u7406",
+        "\u739b\u4e54\u4e3d",
+        "\u739b\u6587",
+        "\u6b63\u80dc",
+        "\u660c\u4fe1",
+        "\u5c06\u53f8",
+        "\u6b63\u4eba",
+        "\u8def\u7237",
+        "\u8001\u7ae0",
+        "\u677e\u7530",
+        "\u677e\u672c",
+        "\u677e\u6d66",
+        "\u677e\u5742",
+        "\u8001\u5b5f",
+        "\u5b5f\u4e39",
+        "\u5546\u4eba\u968f\u4ece",
+        "\u4f20\u4ee4\u5175",
+        "\u7c73\u6b47\u5c14",
+        "\u5fa1\u8206\u6e90\u4e00\u90ce",
+        "\u5fa1\u8206\u6e90\u6b21\u90ce",
+        "\u5343\u5ca9\u519b\u6559\u5934",
+        "\u5343\u5ca9\u519b\u58eb\u5175",
+        "\u660e\u535a",
+        "\u660e\u4fca",
+        "\u7f8e\u94c3",
+        "\u7f8e\u548c",
+        "\u963f\u5e78",
+        "\u524a\u6708\u7b51\u9633\u771f\u541b",
+        "\u94b1\u773c\u513f",
+        "\u68ee\u5f66",
+        "\u5143\u52a9",
+        "\u7406\u6c34\u53e0\u5c71\u771f\u541b",
+        "\u7406\u6c34\u758a\u5c71\u771f\u541b",
+        "\u6731\u8001\u677f",
+        "\u6728\u6728",
+        "\u6751\u4e0a",
+        "\u6751\u7530",
+        "\u6c38\u91ce",
+        "\u957f\u91ce\u539f\u9f99\u4e4b\u4ecb",
+        "\u957f\u6fd1",
+        "\u4e2d\u91ce\u5fd7\u4e43",
+        "\u83dc\u83dc\u5b50",
+        "\u6960\u6960",
+        "\u6210\u6fd1",
+        "\u963f\u5185",
+        "\u5b81\u7984",
+        "\u725b\u5fd7",
+        "\u4fe1\u535a",
+        "\u4f38\u592b",
+        "\u91ce\u65b9",
+        "\u8bfa\u62c9",
+        "\u7eaa\u9999",
+        "\u8bfa\u66fc",
+        "\u4fee\u5973",
+        "\u7eaf\u6c34\u7cbe\u7075",
+        "\u5c0f\u5ddd",
+        "\u5c0f\u4ed3\u6faa",
+        "\u5188\u6797",
+        "\u5188\u5d0e\u7ed8\u91cc\u9999",
+        "\u5188\u5d0e\u9646\u6597",
+        "\u5965\u62c9\u592b",
+        "\u8001\u79d1",
+        "\u9b3c\u5a46\u5a46",
+        "\u5c0f\u91ce\u5bfa",
+        "\u5927\u6cb3\u539f\u4e94\u53f3\u536b\u95e8",
+        "\u5927\u4e45\u4fdd\u5927\u4ecb",
+        "\u5927\u68ee",
+        "\u5927\u52a9",
+        "\u5965\u7279",
+        "\u6d3e\u8499",
+        "\u6d3e\u84992",
+        "\u75c5\u4ebaa",
+        "\u75c5\u4ebab",
+        "\u5df4\u987f",
+        "\u6d3e\u6069",
+        "\u670b\u4e49",
+        "\u56f4\u89c2\u7fa4\u4f17",
+        "\u56f4\u89c2\u7fa4\u4f17a",
+        "\u56f4\u89c2\u7fa4\u4f17b",
+        "\u56f4\u89c2\u7fa4\u4f17c",
+        "\u56f4\u89c2\u7fa4\u4f17d",
+        "\u56f4\u89c2\u7fa4\u4f17e",
+        "\u94dc\u96c0",
+        "\u963f\u80a5",
+        "\u5174\u53d4",
+        "\u8001\u5468\u53d4",
+        "\u516c\u4e3b",
+        "\u5f7c\u5f97",
+        "\u4e7e\u5b50",
+        "\u828a\u828a",
+        "\u4e7e\u73ae",
+        "\u7eee\u547d",
+        "\u675e\u5e73",
+        "\u79cb\u6708",
+        "\u6606\u6069",
+        "\u96f7\u7535\u5f71",
+        "\u5170\u9053\u5c14",
+        "\u96f7\u8499\u5fb7",
+        "\u5192\u5931\u7684\u5e15\u62c9\u5fb7",
+        "\u4f36\u4e00",
+        "\u73b2\u82b1",
+        "\u963f\u4ec1",
+        "\u5bb6\u81e3\u4eec",
+        "\u68a8\u7ed8",
+        "\u8363\u6c5f",
+        "\u620e\u4e16",
+        "\u6d6a\u4eba",
+        "\u7f57\u4f0a\u65af",
+        "\u5982\u610f",
+        "\u51c9\u5b50",
+        "\u5f69\u9999",
+        "\u9152\u4e95",
+        "\u5742\u672c",
+        "\u6714\u6b21\u90ce",
+        "\u6b66\u58eba",
+        "\u6b66\u58ebb",
+        "\u6b66\u58ebc",
+        "\u6b66\u58ebd",
+        "\u73ca\u745a",
+        "\u4e09\u7530",
+        "\u838e\u62c9",
+        "\u7b39\u91ce",
+        "\u806a\u7f8e",
+        "\u806a",
+        "\u5c0f\u767e\u5408",
+        "\u6563\u5175",
+        "\u5bb3\u6015\u7684\u5c0f\u5218",
+        "\u8212\u4f2f\u7279",
+        "\u8212\u8328",
+        "\u6d77\u9f99",
+        "\u4e16\u5b50",
+        "\u8c22\u5c14\u76d6",
+        "\u5bb6\u4e01",
+        "\u5546\u534e",
+        "\u6c99\u5bc5",
+        "\u963f\u5347",
+        "\u67f4\u7530",
+        "\u963f\u8302",
+        "\u5f0f\u5927\u5c06",
+        "\u6e05\u6c34",
+        "\u5fd7\u6751\u52d8\u5175\u536b",
+        "\u65b0\u4e4b\u4e1e",
+        "\u5fd7\u7ec7",
+        "\u77f3\u5934",
+        "\u8bd7\u7fbd",
+        "\u8bd7\u7b60",
+        "\u77f3\u58ee",
+        "\u7fd4\u592a",
+        "\u6b63\u4e8c",
+        "\u5468\u5e73",
+        "\u8212\u6768",
+        "\u9f50\u683c\u8299\u4e3d\u96c5",
+        "\u5973\u58eb",
+        "\u601d\u52e4",
+        "\u516d\u6307\u4e54\u745f",
+        "\u611a\u4eba\u4f17\u5c0f\u5175d",
+        "\u611a\u4eba\u4f17\u5c0f\u5175a",
+        "\u611a\u4eba\u4f17\u5c0f\u5175b",
+        "\u611a\u4eba\u4f17\u5c0f\u5175c",
+        "\u5434\u8001\u4e94",
+        "\u5434\u8001\u4e8c",
+        "\u6ed1\u5934\u9b3c",
+        "\u8a00\u7b11",
+        "\u5434\u8001\u4e03",
+        "\u58eb\u5175h",
+        "\u58eb\u5175i",
+        "\u58eb\u5175a",
+        "\u58eb\u5175b",
+        "\u58eb\u5175c",
+        "\u58eb\u5175d",
+        "\u58eb\u5175e",
+        "\u58eb\u5175f",
+        "\u58eb\u5175g",
+        "\u594f\u592a",
+        "\u65af\u5766\u5229",
+        "\u6387\u661f\u652b\u8fb0\u5929\u541b",
+        "\u5c0f\u5934",
+        "\u5927\u6b66",
+        "\u9676\u4e49\u9686",
+        "\u6749\u672c",
+        "\u82cf\u897f",
+        "\u5acc\u7591\u4ebaa",
+        "\u5acc\u7591\u4ebab",
+        "\u5acc\u7591\u4ebac",
+        "\u5acc\u7591\u4ebad",
+        "\u65af\u4e07",
+        "\u5251\u5ba2a",
+        "\u5251\u5ba2b",
+        "\u963f\u4e8c",
+        "\u5fe0\u80dc",
+        "\u5fe0\u592b",
+        "\u963f\u656c",
+        "\u5b5d\u5229",
+        "\u9e70\u53f8\u8fdb",
+        "\u9ad8\u5c71",
+        "\u4e5d\u6761\u5b5d\u884c",
+        "\u6bc5",
+        "\u7af9\u5185",
+        "\u62d3\u771f",
+        "\u5353\u4e5f",
+        "\u592a\u90ce\u4e38",
+        "\u6cf0\u52d2",
+        "\u624b\u5c9b",
+        "\u54f2\u5e73",
+        "\u54f2\u592b",
+        "\u6258\u514b",
+        "\u5927boss",
+        "\u963f\u5f3a",
+        "\u6258\u5c14\u5fb7\u62c9",
+        "\u65c1\u89c2\u8005",
+        "\u5929\u6210",
+        "\u963f\u5927",
+        "\u8482\u739b\u4e4c\u65af",
+        "\u63d0\u7c73",
+        "\u6237\u7530",
+        "\u963f\u4e09",
+        "\u4e00\u8d77\u7684\u4eba",
+        "\u5fb7\u7530",
+        "\u5fb7\u957f",
+        "\u667a\u6811",
+        "\u5229\u5f66",
+        "\u80d6\u4e4e\u4e4e\u7684\u65c5\u884c\u8005",
+        "\u85cf\u5b9d\u4ebaa",
+        "\u85cf\u5b9d\u4ebab",
+        "\u85cf\u5b9d\u4ebac",
+        "\u85cf\u5b9d\u4ebad",
+        "\u963f\u7947",
+        "\u6052\u96c4",
+        "\u9732\u5b50",
+        "\u8bdd\u5267\u56e2\u56e2\u957f",
+        "\u5185\u6751",
+        "\u4e0a\u91ce",
+        "\u4e0a\u6749",
+        "\u8001\u6234",
+        "\u8001\u9ad8",
+        "\u8001\u8d3e",
+        "\u8001\u58a8",
+        "\u8001\u5b59",
+        "\u5929\u67a2\u661f",
+        "\u8001\u4e91",
+        "\u6709\u4e50\u658b",
+        "\u4e11\u96c4",
+        "\u4e4c\u7ef4",
+        "\u74e6\u4eac",
+        "\u83f2\u5c14\u6208\u9edb\u7279",
+        "\u7ef4\u591a\u5229\u4e9a",
+        "\u8587\u5c14",
+        "\u74e6\u683c\u7eb3",
+        "\u963f\u5916",
+        "\u4f8d\u5973",
+        "\u74e6\u62c9",
+        "\u671b\u96c5",
+        "\u5b9b\u70df",
+        "\u742c\u7389",
+        "\u6218\u58eba",
+        "\u6218\u58ebb",
+        "\u6e21\u8fba",
+        "\u6e21\u90e8",
+        "\u963f\u4f1f",
+        "\u6587\u749f",
+        "\u6587\u6e0a",
+        "\u97e6\u5c14\u7eb3",
+        "\u738b\u6273\u624b",
+        "\u6b66\u6c9b",
+        "\u6653\u98de",
+        "\u8f9b\u7a0b",
+        "\u661f\u706b",
+        "\u661f\u7a00",
+        "\u8f9b\u79c0",
+        "\u79c0\u534e",
+        "\u963f\u65ed",
+        "\u5f90\u5218\u5e08",
+        "\u77e2\u90e8",
+        "\u516b\u6728",
+        "\u5c71\u4e0a",
+        "\u963f\u9633",
+        "\u989c\u7b11",
+        "\u5eb7\u660e",
+        "\u6cf0\u4e45",
+        "\u5b89\u6b66",
+        "\u77e2\u7530\u5e78\u559c",
+        "\u77e2\u7530\u8f9b\u559c",
+        "\u4e49\u575a",
+        "\u83ba\u513f",
+        "\u76c8\u4e30",
+        "\u5b9c\u5e74",
+        "\u94f6\u674f",
+        "\u9038\u8f69",
+        "\u6a2a\u5c71",
+        "\u6c38\u8d35",
+        "\u6c38\u4e1a",
+        "\u5609\u4e45",
+        "\u5409\u5ddd",
+        "\u4e49\u9ad8",
+        "\u7528\u9ad8",
+        "\u9633\u592a",
+        "\u5143\u84c9",
+        "\u73a5\u8f89",
+        "\u6bd3\u534e",
+        "\u6709\u9999",
+        "\u5e78\u4e5f",
+        "\u7531\u771f",
+        "\u7ed3\u83dc",
+        "\u97f5\u5b81",
+        "\u767e\u5408",
+        "\u767e\u5408\u534e",
+        "\u5c24\u82cf\u6ce2\u592b",
+        "\u88d5\u5b50",
+        "\u60a0\u7b56",
+        "\u60a0\u4e5f",
+        "\u4e8e\u5ae3",
+        "\u67da\u5b50",
+        "\u8001\u90d1",
+        "\u6b63\u8302",
+        "\u5fd7\u6210",
+        "\u82b7\u5de7",
+        "\u77e5\u6613",
+        "\u652f\u652f",
+        "\u5468\u826f",
+        "\u73e0\u51fd",
+        "\u795d\u660e",
+        "\u795d\u6d9b"
+    ],
+    "symbols": [
+        "_",
+        ",",
+        ".",
+        "!",
+        "?",
+        "-",
+        "~",
+        "\u2026",
+        "A",
+        "E",
+        "I",
+        "N",
+        "O",
+        "Q",
+        "U",
+        "a",
+        "b",
+        "d",
+        "e",
+        "f",
+        "g",
+        "h",
+        "i",
+        "j",
+        "k",
+        "l",
+        "m",
+        "n",
+        "o",
+        "p",
+        "r",
+        "s",
+        "t",
+        "u",
+        "v",
+        "w",
+        "y",
+        "z",
+        "\u0283",
+        "\u02a7",
+        "\u02a6",
+        "\u026f",
+        "\u0279",
+        "\u0259",
+        "\u0265",
+        "\u207c",
+        "\u02b0",
+        "`",
+        "\u2192",
+        "\u2193",
+        "\u2191",
+        " "
+    ]
+}

configs/config2.json

@@ -0,0 +1,94 @@
+{
+  "train": {
+    "log_interval": 100,
+    "eval_interval": 500,
+    "seed": 109652,
+    "epochs": 20001,
+    "learning_rate": 0.00015,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 30,
+    "fp16_run": false,
+    "bf16_run": false,
+    "lr_decay": 0.999875,
+    "segment_size": 10240,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "use_sr": true,
+    "max_speclen": 512,
+    "port": "8001",
+    "keep_ckpts": 3
+  },
+  "data": {
+    "training_files": "filelists/44k/train.txt",
+    "validation_files": "filelists/44k/val.txt",
+    "max_wav_value": 32768.0,
+    "sampling_rate": 44100,
+    "filter_length": 2048,
+    "hop_length": 512,
+    "win_length": 2048,
+    "n_mel_channels": 80,
+    "mel_fmin": 0.0,
+    "mel_fmax": 22050
+  },
+  "model": {
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.1,
+    "resblock": "1",
+    "resblock_kernel_sizes": [
+      3,
+      7,
+      11
+    ],
+    "resblock_dilation_sizes": [
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ]
+    ],
+    "upsample_rates": [
+      8,
+      8,
+      2,
+      2,
+      2
+    ],
+    "upsample_initial_channel": 512,
+    "upsample_kernel_sizes": [
+      16,
+      16,
+      4,
+      4,
+      4
+    ],
+    "n_layers_q": 3,
+    "use_spectral_norm": false,
+    "gin_channels": 256,
+    "ssl_dim": 256,
+    "n_speakers": 200
+  },
+  "spk": {
+    "parappa": 0
+  }
+}

data_utils.py

@@ -0,0 +1,392 @@
+import time
+import os
+import random
+import numpy as np
+import torch
+import torch.utils.data
+
+import commons 
+from mel_processing import spectrogram_torch
+from utils import load_wav_to_torch, load_filepaths_and_text
+from text import text_to_sequence, cleaned_text_to_sequence
+
+
+class TextAudioLoader(torch.utils.data.Dataset):
+    """
+        1) loads audio, text pairs
+        2) normalizes text and converts them to sequences of integers
+        3) computes spectrograms from audio files.
+    """
+    def __init__(self, audiopaths_and_text, hparams):
+        self.audiopaths_and_text = load_filepaths_and_text(audiopaths_and_text)
+        self.text_cleaners  = hparams.text_cleaners
+        self.max_wav_value  = hparams.max_wav_value
+        self.sampling_rate  = hparams.sampling_rate
+        self.filter_length  = hparams.filter_length 
+        self.hop_length     = hparams.hop_length 
+        self.win_length     = hparams.win_length
+        self.sampling_rate  = hparams.sampling_rate 
+
+        self.cleaned_text = getattr(hparams, "cleaned_text", False)
+
+        self.add_blank = hparams.add_blank
+        self.min_text_len = getattr(hparams, "min_text_len", 1)
+        self.max_text_len = getattr(hparams, "max_text_len", 190)
+
+        random.seed(1234)
+        random.shuffle(self.audiopaths_and_text)
+        self._filter()
+
+
+    def _filter(self):
+        """
+        Filter text & store spec lengths
+        """
+        # Store spectrogram lengths for Bucketing
+        # wav_length ~= file_size / (wav_channels * Bytes per dim) = file_size / (1 * 2)
+        # spec_length = wav_length // hop_length
+
+        audiopaths_and_text_new = []
+        lengths = []
+        for audiopath, text in self.audiopaths_and_text:
+            if self.min_text_len <= len(text) and len(text) <= self.max_text_len:
+                audiopaths_and_text_new.append([audiopath, text])
+                lengths.append(os.path.getsize(audiopath) // (2 * self.hop_length))
+        self.audiopaths_and_text = audiopaths_and_text_new
+        self.lengths = lengths
+
+    def get_audio_text_pair(self, audiopath_and_text):
+        # separate filename and text
+        audiopath, text = audiopath_and_text[0], audiopath_and_text[1]
+        text = self.get_text(text)
+        spec, wav = self.get_audio(audiopath)
+        return (text, spec, wav)
+
+    def get_audio(self, filename):
+        audio, sampling_rate = load_wav_to_torch(filename)
+        if sampling_rate != self.sampling_rate:
+            raise ValueError("{} {} SR doesn't match target {} SR".format(
+                sampling_rate, self.sampling_rate))
+        audio_norm = audio / self.max_wav_value
+        audio_norm = audio_norm.unsqueeze(0)
+        spec_filename = filename.replace(".wav", ".spec.pt")
+        if os.path.exists(spec_filename):
+            spec = torch.load(spec_filename)
+        else:
+            spec = spectrogram_torch(audio_norm, self.filter_length,
+                self.sampling_rate, self.hop_length, self.win_length,
+                center=False)
+            spec = torch.squeeze(spec, 0)
+            torch.save(spec, spec_filename)
+        return spec, audio_norm
+
+    def get_text(self, text):
+        if self.cleaned_text:
+            text_norm = cleaned_text_to_sequence(text)
+        else:
+            text_norm = text_to_sequence(text, self.text_cleaners)
+        if self.add_blank:
+            text_norm = commons.intersperse(text_norm, 0)
+        text_norm = torch.LongTensor(text_norm)
+        return text_norm
+
+    def __getitem__(self, index):
+        return self.get_audio_text_pair(self.audiopaths_and_text[index])
+
+    def __len__(self):
+        return len(self.audiopaths_and_text)
+
+
+class TextAudioCollate():
+    """ Zero-pads model inputs and targets
+    """
+    def __init__(self, return_ids=False):
+        self.return_ids = return_ids
+
+    def __call__(self, batch):
+        """Collate's training batch from normalized text and aduio
+        PARAMS
+        ------
+        batch: [text_normalized, spec_normalized, wav_normalized]
+        """
+        # Right zero-pad all one-hot text sequences to max input length
+        _, ids_sorted_decreasing = torch.sort(
+            torch.LongTensor([x[1].size(1) for x in batch]),
+            dim=0, descending=True)
+
+        max_text_len = max([len(x[0]) for x in batch])
+        max_spec_len = max([x[1].size(1) for x in batch])
+        max_wav_len = max([x[2].size(1) for x in batch])
+
+        text_lengths = torch.LongTensor(len(batch))
+        spec_lengths = torch.LongTensor(len(batch))
+        wav_lengths = torch.LongTensor(len(batch))
+
+        text_padded = torch.LongTensor(len(batch), max_text_len)
+        spec_padded = torch.FloatTensor(len(batch), batch[0][1].size(0), max_spec_len)
+        wav_padded = torch.FloatTensor(len(batch), 1, max_wav_len)
+        text_padded.zero_()
+        spec_padded.zero_()
+        wav_padded.zero_()
+        for i in range(len(ids_sorted_decreasing)):
+            row = batch[ids_sorted_decreasing[i]]
+
+            text = row[0]
+            text_padded[i, :text.size(0)] = text
+            text_lengths[i] = text.size(0)
+
+            spec = row[1]
+            spec_padded[i, :, :spec.size(1)] = spec
+            spec_lengths[i] = spec.size(1)
+
+            wav = row[2]
+            wav_padded[i, :, :wav.size(1)] = wav
+            wav_lengths[i] = wav.size(1)
+
+        if self.return_ids:
+            return text_padded, text_lengths, spec_padded, spec_lengths, wav_padded, wav_lengths, ids_sorted_decreasing
+        return text_padded, text_lengths, spec_padded, spec_lengths, wav_padded, wav_lengths
+
+
+"""Multi speaker version"""
+class TextAudioSpeakerLoader(torch.utils.data.Dataset):
+    """
+        1) loads audio, speaker_id, text pairs
+        2) normalizes text and converts them to sequences of integers
+        3) computes spectrograms from audio files.
+    """
+    def __init__(self, audiopaths_sid_text, hparams):
+        self.audiopaths_sid_text = load_filepaths_and_text(audiopaths_sid_text)
+        self.text_cleaners = hparams.text_cleaners
+        self.max_wav_value = hparams.max_wav_value
+        self.sampling_rate = hparams.sampling_rate
+        self.filter_length  = hparams.filter_length
+        self.hop_length     = hparams.hop_length
+        self.win_length     = hparams.win_length
+        self.sampling_rate  = hparams.sampling_rate
+
+        self.cleaned_text = getattr(hparams, "cleaned_text", False)
+
+        self.add_blank = hparams.add_blank
+        self.min_text_len = getattr(hparams, "min_text_len", 1)
+        self.max_text_len = getattr(hparams, "max_text_len", 190)
+
+        random.seed(1234)
+        random.shuffle(self.audiopaths_sid_text)
+        self._filter()
+
+    def _filter(self):
+        """
+        Filter text & store spec lengths
+        """
+        # Store spectrogram lengths for Bucketing
+        # wav_length ~= file_size / (wav_channels * Bytes per dim) = file_size / (1 * 2)
+        # spec_length = wav_length // hop_length
+
+        audiopaths_sid_text_new = []
+        lengths = []
+        for audiopath, sid, text in self.audiopaths_sid_text:
+            if self.min_text_len <= len(text) and len(text) <= self.max_text_len:
+                audiopaths_sid_text_new.append([audiopath, sid, text])
+                lengths.append(os.path.getsize(audiopath) // (2 * self.hop_length))
+        self.audiopaths_sid_text = audiopaths_sid_text_new
+        self.lengths = lengths
+
+    def get_audio_text_speaker_pair(self, audiopath_sid_text):
+        # separate filename, speaker_id and text
+        audiopath, sid, text = audiopath_sid_text[0], audiopath_sid_text[1], audiopath_sid_text[2]
+        text = self.get_text(text)
+        spec, wav = self.get_audio(audiopath)
+        sid = self.get_sid(sid)
+        return (text, spec, wav, sid)
+
+    def get_audio(self, filename):
+        audio, sampling_rate = load_wav_to_torch(filename)
+        if sampling_rate != self.sampling_rate:
+            raise ValueError("{} {} SR doesn't match target {} SR".format(
+                sampling_rate, self.sampling_rate))
+        audio_norm = audio / self.max_wav_value
+        audio_norm = audio_norm.unsqueeze(0)
+        spec_filename = filename.replace(".wav", ".spec.pt")
+        if os.path.exists(spec_filename):
+            spec = torch.load(spec_filename)
+        else:
+            spec = spectrogram_torch(audio_norm, self.filter_length,
+                self.sampling_rate, self.hop_length, self.win_length,
+                center=False)
+            spec = torch.squeeze(spec, 0)
+            torch.save(spec, spec_filename)
+        return spec, audio_norm
+
+    def get_text(self, text):
+        if self.cleaned_text:
+            text_norm = cleaned_text_to_sequence(text)
+        else:
+            text_norm = text_to_sequence(text, self.text_cleaners)
+        if self.add_blank:
+            text_norm = commons.intersperse(text_norm, 0)
+        text_norm = torch.LongTensor(text_norm)
+        return text_norm
+
+    def get_sid(self, sid):
+        sid = torch.LongTensor([int(sid)])
+        return sid
+
+    def __getitem__(self, index):
+        return self.get_audio_text_speaker_pair(self.audiopaths_sid_text[index])
+
+    def __len__(self):
+        return len(self.audiopaths_sid_text)
+
+
+class TextAudioSpeakerCollate():
+    """ Zero-pads model inputs and targets
+    """
+    def __init__(self, return_ids=False):
+        self.return_ids = return_ids
+
+    def __call__(self, batch):
+        """Collate's training batch from normalized text, audio and speaker identities
+        PARAMS
+        ------
+        batch: [text_normalized, spec_normalized, wav_normalized, sid]
+        """
+        # Right zero-pad all one-hot text sequences to max input length
+        _, ids_sorted_decreasing = torch.sort(
+            torch.LongTensor([x[1].size(1) for x in batch]),
+            dim=0, descending=True)
+
+        max_text_len = max([len(x[0]) for x in batch])
+        max_spec_len = max([x[1].size(1) for x in batch])
+        max_wav_len = max([x[2].size(1) for x in batch])
+
+        text_lengths = torch.LongTensor(len(batch))
+        spec_lengths = torch.LongTensor(len(batch))
+        wav_lengths = torch.LongTensor(len(batch))
+        sid = torch.LongTensor(len(batch))
+
+        text_padded = torch.LongTensor(len(batch), max_text_len)
+        spec_padded = torch.FloatTensor(len(batch), batch[0][1].size(0), max_spec_len)
+        wav_padded = torch.FloatTensor(len(batch), 1, max_wav_len)
+        text_padded.zero_()
+        spec_padded.zero_()
+        wav_padded.zero_()
+        for i in range(len(ids_sorted_decreasing)):
+            row = batch[ids_sorted_decreasing[i]]
+
+            text = row[0]
+            text_padded[i, :text.size(0)] = text
+            text_lengths[i] = text.size(0)
+
+            spec = row[1]
+            spec_padded[i, :, :spec.size(1)] = spec
+            spec_lengths[i] = spec.size(1)
+
+            wav = row[2]
+            wav_padded[i, :, :wav.size(1)] = wav
+            wav_lengths[i] = wav.size(1)
+
+            sid[i] = row[3]
+
+        if self.return_ids:
+            return text_padded, text_lengths, spec_padded, spec_lengths, wav_padded, wav_lengths, sid, ids_sorted_decreasing
+        return text_padded, text_lengths, spec_padded, spec_lengths, wav_padded, wav_lengths, sid
+
+
+class DistributedBucketSampler(torch.utils.data.distributed.DistributedSampler):
+    """
+    Maintain similar input lengths in a batch.
+    Length groups are specified by boundaries.
+    Ex) boundaries = [b1, b2, b3] -> any batch is included either {x | b1 < length(x) <=b2} or {x | b2 < length(x) <= b3}.
+  
+    It removes samples which are not included in the boundaries.
+    Ex) boundaries = [b1, b2, b3] -> any x s.t. length(x) <= b1 or length(x) > b3 are discarded.
+    """
+    def __init__(self, dataset, batch_size, boundaries, num_replicas=None, rank=None, shuffle=True):
+        super().__init__(dataset, num_replicas=num_replicas, rank=rank, shuffle=shuffle)
+        self.lengths = dataset.lengths
+        self.batch_size = batch_size
+        self.boundaries = boundaries
+  
+        self.buckets, self.num_samples_per_bucket = self._create_buckets()
+        self.total_size = sum(self.num_samples_per_bucket)
+        self.num_samples = self.total_size // self.num_replicas
+  
+    def _create_buckets(self):
+        buckets = [[] for _ in range(len(self.boundaries) - 1)]
+        for i in range(len(self.lengths)):
+            length = self.lengths[i]
+            idx_bucket = self._bisect(length)
+            if idx_bucket != -1:
+                buckets[idx_bucket].append(i)
+  
+        for i in range(len(buckets) - 1, 0, -1):
+            if len(buckets[i]) == 0:
+                buckets.pop(i)
+                self.boundaries.pop(i+1)
+  
+        num_samples_per_bucket = []
+        for i in range(len(buckets)):
+            len_bucket = len(buckets[i])
+            total_batch_size = self.num_replicas * self.batch_size
+            rem = (total_batch_size - (len_bucket % total_batch_size)) % total_batch_size
+            num_samples_per_bucket.append(len_bucket + rem)
+        return buckets, num_samples_per_bucket
+  
+    def __iter__(self):
+      # deterministically shuffle based on epoch
+      g = torch.Generator()
+      g.manual_seed(self.epoch)
+  
+      indices = []
+      if self.shuffle:
+          for bucket in self.buckets:
+              indices.append(torch.randperm(len(bucket), generator=g).tolist())
+      else:
+          for bucket in self.buckets:
+              indices.append(list(range(len(bucket))))
+  
+      batches = []
+      for i in range(len(self.buckets)):
+          bucket = self.buckets[i]
+          len_bucket = len(bucket)
+          ids_bucket = indices[i]
+          num_samples_bucket = self.num_samples_per_bucket[i]
+  
+          # add extra samples to make it evenly divisible
+          rem = num_samples_bucket - len_bucket
+          ids_bucket = ids_bucket + ids_bucket * (rem // len_bucket) + ids_bucket[:(rem % len_bucket)]
+  
+          # subsample
+          ids_bucket = ids_bucket[self.rank::self.num_replicas]
+  
+          # batching
+          for j in range(len(ids_bucket) // self.batch_size):
+              batch = [bucket[idx] for idx in ids_bucket[j*self.batch_size:(j+1)*self.batch_size]]
+              batches.append(batch)
+  
+      if self.shuffle:
+          batch_ids = torch.randperm(len(batches), generator=g).tolist()
+          batches = [batches[i] for i in batch_ids]
+      self.batches = batches
+  
+      assert len(self.batches) * self.batch_size == self.num_samples
+      return iter(self.batches)
+  
+    def _bisect(self, x, lo=0, hi=None):
+      if hi is None:
+          hi = len(self.boundaries) - 1
+  
+      if hi > lo:
+          mid = (hi + lo) // 2
+          if self.boundaries[mid] < x and x <= self.boundaries[mid+1]:
+              return mid
+          elif x <= self.boundaries[mid]:
+              return self._bisect(x, lo, mid)
+          else:
+              return self._bisect(x, mid + 1, hi)
+      else:
+          return -1
+
+    def __len__(self):
+        return self.num_samples // self.batch_size

filelists/yuuka_train.txt

@@ -0,0 +1,108 @@
+wavs/yuuka/1.wav|0|ブルーアーカイブ
+wavs/yuuka/2.wav|0|条件はクリアされました
+wavs/yuuka/3.wav|0|私たちは今日この瞬間は
+wavs/yuuka/4.wav|0|証言することになるでしょう
+wavs/yuuka/5.wav|0|今日も全力であなたをアシストしますね
+wavs/yuuka/6.wav|0|ようこそ
+wavs/yuuka/7.wav|0|先生
+wavs/yuuka/8.wav|0|今から反省会を始めます
+wavs/yuuka/9.wav|0|どこに行くんですか
+wavs/yuuka/10.wav|0|先生
+wavs/yuuka/11.wav|0|もう少し頑張って下さい
+wavs/yuuka/12.wav|0|今
+wavs/yuuka/13.wav|0|先生の行動について言いたいことが
+wavs/yuuka/14.wav|0|そのうちのひとつは
+wavs/yuuka/15.wav|0|特別な言葉だけど
+wavs/yuuka/16.wav|0|でも最近は
+wavs/yuuka/17.wav|0|先生らしくなった気がします
+wavs/yuuka/18.wav|0|私のおかげ
+wavs/yuuka/19.wav|0|これからの目標と
+wavs/yuuka/20.wav|0|方向性を測定してください
+wavs/yuuka/21.wav|0|バカ
+wavs/yuuka/22.wav|0|大人なんですからしっかりと大人らしく
+wavs/yuuka/23.wav|0|計画的な消費をしてください
+wavs/yuuka/24.wav|0|お小遣いをもらって
+wavs/yuuka/25.wav|0|次はもう
+wavs/yuuka/26.wav|0|コンビニで購入した
+wavs/yuuka/27.wav|0|新刊の漫画購入で
+wavs/yuuka/28.wav|0|おとといの夜
+wavs/yuuka/29.wav|0|生徒たちの模範となるべき教育者だ
+wavs/yuuka/30.wav|0|信じられません
+wavs/yuuka/31.wav|0|最低です
+wavs/yuuka/32.wav|0|薬っていう名前の
+wavs/yuuka/48.wav|0|買わなきゃ
+wavs/yuuka/49.wav|0|体力も時間も
+wavs/yuuka/50.wav|0|たくさん使っちゃった気がしますけど
+wavs/yuuka/51.wav|0|先生といると
+wavs/yuuka/52.wav|0|得した気分
+wavs/yuuka/53.wav|0|計算ができません
+wavs/yuuka/54.wav|0|この気持ちがですよ
+wavs/yuuka/55.wav|0|計算だけを考えてたけど
+wavs/yuuka/56.wav|0|今は
+wavs/yuuka/57.wav|0|ただこの瞬間を感じていたい
+wavs/yuuka/58.wav|0|説明できない感覚だけど
+wavs/yuuka/59.wav|0|嫌いじゃない
+wavs/yuuka/60.wav|0|合理的な選択ね
+wavs/yuuka/61.wav|0|報酬は出るんです
+wavs/yuuka/62.wav|0|やっぱり先生は賢いですね
+wavs/yuuka/63.wav|0|敵の位置を確認
+wavs/yuuka/64.wav|0|敵の位置を確認
+wavs/yuuka/65.wav|0|チキン
+wavs/yuuka/66.wav|0|権力
+wavs/yuuka/67.wav|0|まだ終わらないわよ
+wavs/yuuka/68.wav|0|彼は立つね
+wavs/yuuka/69.wav|0|攻撃が私に命中する確率は
+wavs/yuuka/70.wav|0|極めて低い
+wavs/yuuka/71.wav|0|単なる計算結果にすぎないわ
+wavs/yuuka/72.wav|0|勝利は
+wavs/yuuka/73.wav|0|証明する
+wavs/yuuka/74.wav|0|計算の完璧
+wavs/yuuka/75.wav|0|落ち着いていこう
+wavs/yuuka/76.wav|0|私たちの勝率はかなり高い
+wavs/yuuka/77.wav|0|ありがとう
+wavs/yuuka/78.wav|0|私の計算は完璧よ
+wavs/yuuka/79.wav|0|隠れるは
+wavs/yuuka/80.wav|0|支援をお願い
+wavs/yuuka/81.wav|0|無駄よ
+wavs/yuuka/82.wav|0|ここまでは計算どおりね
+wavs/yuuka/83.wav|0|計算が合っていれば
+wavs/yuuka/84.wav|0|すぐに次の目標が見えてくるはず
+wavs/yuuka/85.wav|0|ちょうどいいよね
+wavs/yuuka/86.wav|0|陣痛が
+wavs/yuuka/87.wav|0|大杉
+wavs/yuuka/88.wav|0|行くわよ
+wavs/yuuka/89.wav|0|計算通り
+wavs/yuuka/90.wav|0|岸壁
+wavs/yuuka/91.wav|0|数学が真実を導く
+wavs/yuuka/92.wav|0|たかちよ
+wavs/yuuka/93.wav|0|スマートに調理
+wavs/yuuka/94.wav|0|運が良かった
+wavs/yuuka/95.wav|0|いいね
+wavs/yuuka/96.wav|0|計算通りです
+wavs/yuuka/97.wav|0|設計の方が間違ってましたから
+wavs/yuuka/98.wav|0|計算機が
+wavs/yuuka/99.wav|0|故障していたとは
+wavs/yuuka/100.wav|0|背中を見せるチャンス
+wavs/yuuka/101.wav|0|待ってました
+wavs/yuuka/102.wav|0|周知で感じられる結果というのは
+wavs/yuuka/103.wav|0|人をドキドキさせますね
+wavs/yuuka/104.wav|0|次の任務は
+wavs/yuuka/105.wav|0|切れ痔を鍛えてくださいね
+wavs/yuuka/106.wav|0|こんな表現は
+wavs/yuuka/107.wav|0|あまり好きじゃないけど
+wavs/yuuka/108.wav|0|今の私
+wavs/yuuka/109.wav|0|測定不能です
+wavs/yuuka/110.wav|0|これは
+wavs/yuuka/111.wav|0|ものすごいプラスになるやつだ
+wavs/yuuka/112.wav|0|今日は私の誕生日です
+wavs/yuuka/113.wav|0|プレゼントは私が選びますね
+wavs/yuuka/114.wav|0|出かけませんか
+wavs/yuuka/115.wav|0|高いものじゃなくてもいいので
+wavs/yuuka/116.wav|0|今日で先生が生まれたからやっくん
+wavs/yuuka/117.wav|0|学園都市の人口に比例してみると
+wavs/yuuka/118.wav|0|数学でも
+wavs/yuuka/119.wav|0|初めが肝心です
+wavs/yuuka/120.wav|0|今年の始まりは
+wavs/yuuka/121.wav|0|一緒に始めましょうか
+wavs/yuuka/122.wav|0|今は先生がいてくれて
+wavs/yuuka/123.wav|0|楽しいですね

filelists/yuuka_train.txt.cleaned

@@ -0,0 +1,108 @@
+wavs/yuuka/1.wav|0|bɯ↑rɯɯaaka↓ibɯ.
+wavs/yuuka/2.wav|0|jo↑oke↓Nwa kɯ↑ri↓a sa↑rema↓ʃIta.
+wavs/yuuka/3.wav|0|wa↑taʃi↓taʧiwa kyo↓o ko↑no ʃɯ↑NkaNwa.
+wavs/yuuka/4.wav|0|ʃo↑ogeN sɯ↑rɯ ko↑to↓ni na↓rɯdeʃoo.
+wavs/yuuka/5.wav|0|kyo↓omo ze↑Nryokɯde a↑na↓tao a↑ʃi↓sUto ʃi↑ma↓sUne.
+wavs/yuuka/6.wav|0|yo↓okoso.
+wavs/yuuka/7.wav|0|se↑Nse↓e.
+wavs/yuuka/8.wav|0|i↓makara ha↑Nsee↓kaio ha↑jimema↓sU.
+wavs/yuuka/9.wav|0|do↓koni i↑kɯ N↓desUka.
+wavs/yuuka/10.wav|0|se↑Nse↓e.
+wavs/yuuka/11.wav|0|mo↑osUko↓ʃi ga↑Nba↓Qte kɯ↑dasa↓i.
+wavs/yuuka/12.wav|0|i↓ma.
+wavs/yuuka/13.wav|0|se↑Nse↓eno ko↑odooni ʦɯ↓ite i↑ita↓i ko↑to↓ga.
+wavs/yuuka/14.wav|0|so↑no ɯ↑ʧino hI↑to↓ʦɯwa.
+wavs/yuuka/15.wav|0|to↑kɯbeʦɯna ko↑toba↓dakedo.
+wavs/yuuka/16.wav|0|de↓mo sa↑ikiNwa.
+wavs/yuuka/17.wav|0|se↑Nseeraʃi↓kɯ na↓Qta ki↑ga ʃi↑ma↓sU.
+wavs/yuuka/18.wav|0|wa↑taʃino o↑kage.
+wavs/yuuka/19.wav|0|ko↑rekarano mo↑kUhyooto.
+wavs/yuuka/20.wav|0|ho↑okooseeo so↑kUtee ʃI↑te kɯ↑dasa↓i.
+wavs/yuuka/21.wav|0|ba↓ka.
+wavs/yuuka/22.wav|0|o↑tonana N↓desUkara ʃi↑Qka↓rito o↑tonaraʃi↓kɯ.
+wavs/yuuka/23.wav|0|ke↑ekakɯ↓tekina ʃo↑ohio ʃI↑te kɯ↑dasa↓i.
+wavs/yuuka/24.wav|0|o↑ko↓zɯkaio mo↑raQte.
+wavs/yuuka/25.wav|0|ʦɯ↑gi↓wa mo↓o.
+wavs/yuuka/26.wav|0|ko↑Nbinide ko↑onyɯɯ ʃI↑ta.
+wavs/yuuka/27.wav|0|ʃi↑NkaNno ma↑Ngako↓onyɯɯde.
+wavs/yuuka/28.wav|0|o↑toto↓ino yo↓rɯ.
+wavs/yuuka/29.wav|0|se↑eto↓taʧino mo↑haNto na↓rɯbekI kyo↑oikUʃada.
+wavs/yuuka/30.wav|0|ʃi↑Njiraremase↓N.
+wavs/yuuka/31.wav|0|sa↑iteede↓sU.
+wavs/yuuka/32.wav|0|kU↑sɯriQte i↓ɯ na↑maeno.
+wavs/yuuka/48.wav|0|ka↑wanakya.
+wavs/yuuka/49.wav|0|ta↓iryokɯmo ji↑kaNmo.
+wavs/yuuka/50.wav|0|ta↑kUsaN ʦU↑kaQʧaQta ki↑ga ʃi↑ma↓sUkedo.
+wavs/yuuka/51.wav|0|se↑Nse↓eto i↑rɯto.
+wavs/yuuka/52.wav|0|to↑kɯ ʃI↑ta ki↓bɯN.
+wavs/yuuka/53.wav|0|ke↑esaNga de↑kimase↓N.
+wavs/yuuka/54.wav|0|ko↑no ki↑moʧigadesUyo.
+wavs/yuuka/55.wav|0|ke↑esaNdakeo ka↑Nga↓etetakedo.
+wavs/yuuka/56.wav|0|i↓mawa.
+wavs/yuuka/57.wav|0|ta↓da ko↑no ʃɯ↑NkaNo ka↑Njite i↑ta↓i.
+wavs/yuuka/58.wav|0|se↑ʦɯmee de↑ki↓nai ka↑Nkakɯda↓kedo.
+wavs/yuuka/59.wav|0|ki↑raijanai.
+wavs/yuuka/60.wav|0|go↑ori↓tekina se↑Ntakɯne.
+wavs/yuuka/61.wav|0|ho↑oʃɯɯwa de↓rɯ N↓desU.
+wavs/yuuka/62.wav|0|ya↑Qpa↓ri se↑Nse↓ewa ka↑ʃIko↓idesUne.
+wavs/yuuka/63.wav|0|te↑kino i↓ʧio ka↑kɯniN.
+wavs/yuuka/64.wav|0|te↑kino i↓ʧio ka↑kɯniN.
+wavs/yuuka/65.wav|0|ʧi↓kiN.
+wavs/yuuka/66.wav|0|ke↓Nryokɯ.
+wavs/yuuka/67.wav|0|ma↓da o↑waranai↓wayo.
+wavs/yuuka/68.wav|0|ka↓rewa ta↓ʦɯne.
+wavs/yuuka/69.wav|0|ko↑ogekiga wa↑taʃini me↑eʧɯɯ sɯ↑rɯ ka↑kɯriʦɯwa.
+wavs/yuuka/70.wav|0|ki↑wa↓mete hI↑kɯ↓i.
+wavs/yuuka/71.wav|0|ta↓Nnarɯ ke↑esaN ke↑Qkani sɯ↑gi↓naiwa.
+wavs/yuuka/72.wav|0|ʃo↓oriwa.
+wavs/yuuka/73.wav|0|ʃo↑omee sɯ↑rɯ.
+wavs/yuuka/74.wav|0|ke↑esaNno ka↑Npeki.
+wavs/yuuka/75.wav|0|o↑ʧIʦɯite i↑koo.
+wavs/yuuka/76.wav|0|wa↑taʃi↓taʧino ʃo↑oriʦɯwa ka↓nari ta↑ka↓i.
+wavs/yuuka/77.wav|0|a↑ri↓gatoo.
+wavs/yuuka/78.wav|0|wa↑taʃino ke↑esaNwa ka↑Npekiyo.
+wavs/yuuka/79.wav|0|ka↑kɯre↓rɯwa.
+wavs/yuuka/80.wav|0|ʃi↑eNo o↑negai.
+wavs/yuuka/81.wav|0|mɯ↑dayo.
+wavs/yuuka/82.wav|0|ko↑koma↓dewa ke↑esaN↓doorine.
+wavs/yuuka/83.wav|0|ke↑esaNga a↓Qte i↑re↓ba.
+wavs/yuuka/84.wav|0|sɯ↓gɯni ʦɯ↑gi↓no mo↑kUhyooga mi↑e↓te kɯ↓rɯ ha↑zɯ.
+wavs/yuuka/85.wav|0|ʧo↑odo i↓iyone.
+wavs/yuuka/86.wav|0|ji↑Nʦɯɯga.
+wavs/yuuka/87.wav|0|o↑osɯgi.
+wavs/yuuka/88.wav|0|i↑kɯ↓wayo.
+wavs/yuuka/89.wav|0|ke↑esaNdo↓ori.
+wavs/yuuka/90.wav|0|ga↑Npeki.
+wavs/yuuka/91.wav|0|sɯ↑ɯgakɯga ʃi↓Njiʦɯo mi↑ʧibi↓kɯ.
+wavs/yuuka/92.wav|0|t a ka↓ʧiyo.
+wavs/yuuka/93.wav|0|sɯ↑ma↓atoni ʧo↓ori.
+wavs/yuuka/94.wav|0|ɯ↓Nga yo↓kaQta.
+wavs/yuuka/95.wav|0|i↓ine.
+wavs/yuuka/96.wav|0|ke↑esaNdo↓oridesU.
+wavs/yuuka/97.wav|0|se↑Qkeeno ho↓oga ma↑ʧiga↓QtemaʃItakara.
+wavs/yuuka/98.wav|0|ke↑esaN↓kiga.
+wavs/yuuka/99.wav|0|ko↑ʃoo ʃI↑te i↑ta↓towa.
+wavs/yuuka/100.wav|0|se↑nakao mi↑se↓rɯ ʧa↓Nsɯ.
+wavs/yuuka/101.wav|0|ma↓QtemaʃIta.
+wavs/yuuka/102.wav|0|ʃɯ↓ɯʧide ka↑Njirarerɯ ke↑Qkato i↑ɯ no↑wa.
+wavs/yuuka/103.wav|0|hI↑too do↓kidokI sa↑sema↓sUne.
+wavs/yuuka/104.wav|0|ʦɯ↑gi↓no ni↓Nmɯwa.
+wavs/yuuka/105.wav|0|ki↑re↓jio kI↑tae↓te kɯ↑dasa↓ine.
+wavs/yuuka/106.wav|0|ko↑Nna hyo↑oge↓Nwa.
+wavs/yuuka/107.wav|0|a↑mari sU↑ki↓janaikedo.
+wavs/yuuka/108.wav|0|i↓mano wa↑taʃi.
+wavs/yuuka/109.wav|0|so↑kUtee fɯ↑noode↓sU.
+wavs/yuuka/110.wav|0|ko↑rewa.
+wavs/yuuka/111.wav|0|mo↑nosɯgo↓i pɯ↑rasɯni na↓rɯ ya↓ʦɯda.
+wavs/yuuka/112.wav|0|kyo↓owa wa↑taʃino ta↑Njoo↓bidesU.
+wavs/yuuka/113.wav|0|pɯ↑re↓zeNtowa wa↑taʃiga e↑rabima↓sUne.
+wavs/yuuka/114.wav|0|de↑kakemase↓Nka.
+wavs/yuuka/115.wav|0|ta↑ka↓i mo↑no↓janakUtemo i↓inode.
+wavs/yuuka/116.wav|0|kyo↓ode se↑Nse↓ega ɯ↑mareta↓kara ya↓QkɯN.
+wavs/yuuka/117.wav|0|ga↑kɯeN↓toʃino ji↑Nkooni hi↑ree ʃI↑te mi↓rɯto.
+wavs/yuuka/118.wav|0|sɯ↑ɯgakɯdemo.
+wavs/yuuka/119.wav|0|ha↑jimega ka↑NjiNde↓sU.
+wavs/yuuka/120.wav|0|ko↑toʃino ha↑jimariwa.
+wavs/yuuka/121.wav|0|i↑Qʃoni ha↑jimemaʃo↓oka.
+wavs/yuuka/122.wav|0|i↓mawa se↑Nse↓ega i↑te kɯ↑rete.
+wavs/yuuka/123.wav|0|ta↑noʃi↓idesUne.

filelists/yuuka_val.txt

@@ -0,0 +1,15 @@
+wavs/yuuka/33.wav|0|スマホゲームのガチャ
+wavs/yuuka/34.wav|0|切り戻し
+wavs/yuuka/35.wav|0|それでも
+wavs/yuuka/36.wav|0|衝動買いは禁止
+wavs/yuuka/37.wav|0|消費はちゃんと
+wavs/yuuka/38.wav|0|計画的に
+wavs/yuuka/39.wav|0|少し休んで行きますか
+wavs/yuuka/40.wav|0|私も
+wavs/yuuka/41.wav|0|なんだか今日は
+wavs/yuuka/42.wav|0|楽しいことがありそうな気がする
+wavs/yuuka/43.wav|0|確率的に
+wavs/yuuka/44.wav|0|このままだと破産する
+wavs/yuuka/45.wav|0|出勤を減らさないと
+wavs/yuuka/46.wav|0|もしこれを購入したら
+wavs/yuuka/47.wav|0|それは辛いけど

filelists/yuuka_val.txt.cleaned

@@ -0,0 +1,15 @@
+wavs/yuuka/33.wav|0|sɯ↑mahoge↓emɯno ga↓ʧa.
+wavs/yuuka/34.wav|0|ki↑rimodo↓ʃi.
+wavs/yuuka/35.wav|0|so↑rede↓mo.
+wavs/yuuka/36.wav|0|ʃo↑odoogaiwa ki↑Nʃi.
+wavs/yuuka/37.wav|0|ʃo↑ohiwa ʧa↑Nto.
+wavs/yuuka/38.wav|0|ke↑ekakɯ↓tekini.
+wavs/yuuka/39.wav|0|sU↑ko↓ʃi ya↑sɯ↓Nde i↑kima↓sUka.
+wavs/yuuka/40.wav|0|wa↑taʃimo.
+wavs/yuuka/41.wav|0|na↓Ndaka kyo↓owa.
+wavs/yuuka/42.wav|0|ta↑noʃi↓i ko↑to↓ga a↑ri↓soona ki↑ga sɯ↑rɯ.
+wavs/yuuka/43.wav|0|ka↑kɯriʦɯ↓tekini.
+wavs/yuuka/44.wav|0|ko↑no ma↑ma↓dato ha↑saN sɯ↑rɯ.
+wavs/yuuka/45.wav|0|ʃɯ↑QkiNo he↑rasanaito.
+wavs/yuuka/46.wav|0|mo↓ʃI ko↑reo ko↑onyɯɯ ʃI↑ta↓ra.
+wavs/yuuka/47.wav|0|so↑rewa ʦɯ↑ra↓ikedo.

inference.ipynb

@@ -0,0 +1,253 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "%matplotlib inline\n",
+    "import matplotlib.pyplot as plt\n",
+    "import IPython.display as ipd\n",
+    "\n",
+    "import os\n",
+    "import json\n",
+    "import math\n",
+    "import torch\n",
+    "from torch import nn\n",
+    "from torch.nn import functional as F\n",
+    "from torch.utils.data import DataLoader\n",
+    "\n",
+    "import commons\n",
+    "import utils\n",
+    "from data_utils import TextAudioLoader, TextAudioCollate, TextAudioSpeakerLoader, TextAudioSpeakerCollate\n",
+    "from models import SynthesizerTrn\n",
+    "from text.symbols import symbols\n",
+    "from text import text_to_sequence\n",
+    "\n",
+    "from scipy.io.wavfile import write\n",
+    "\n",
+    "\n",
+    "def get_text(text, hps):\n",
+    "    text_norm = text_to_sequence(text, hps.data.text_cleaners)\n",
+    "    if hps.data.add_blank:\n",
+    "        text_norm = commons.intersperse(text_norm, 0)\n",
+    "    text_norm = torch.LongTensor(text_norm)\n",
+    "    return text_norm"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "pycharm": {
+     "name": "#%% md\n"
+    }
+   },
+   "source": [
+    "## Single Speaker"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "hps = utils.get_hparams_from_file(\"configs/XXX.json\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "net_g = SynthesizerTrn(\n",
+    "    len(symbols),\n",
+    "    hps.data.filter_length // 2 + 1,\n",
+    "    hps.train.segment_size // hps.data.hop_length,\n",
+    "    **hps.model).cuda()\n",
+    "_ = net_g.eval()\n",
+    "\n",
+    "_ = utils.load_checkpoint(\"/path/to/model.pth\", net_g, None)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "stn_tst = get_text(\"こんにちは\", hps)\n",
+    "with torch.no_grad():\n",
+    "    x_tst = stn_tst.cuda().unsqueeze(0)\n",
+    "    x_tst_lengths = torch.LongTensor([stn_tst.size(0)]).cuda()\n",
+    "    audio = net_g.infer(x_tst, x_tst_lengths, noise_scale=.667, noise_scale_w=0.8, length_scale=1)[0][0,0].data.cpu().float().numpy()\n",
+    "ipd.display(ipd.Audio(audio, rate=hps.data.sampling_rate, normalize=False))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "pycharm": {
+     "name": "#%% md\n"
+    }
+   },
+   "source": [
+    "## Multiple Speakers"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "hps = utils.get_hparams_from_file(\"./configs/XXX.json\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "net_g = SynthesizerTrn(\n",
+    "    len(symbols),\n",
+    "    hps.data.filter_length // 2 + 1,\n",
+    "    hps.train.segment_size // hps.data.hop_length,\n",
+    "    n_speakers=hps.data.n_speakers,\n",
+    "    **hps.model).cuda()\n",
+    "_ = net_g.eval()\n",
+    "\n",
+    "_ = utils.load_checkpoint(\"/path/to/model.pth\", net_g, None)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "stn_tst = get_text(\"こんにちは\", hps)\n",
+    "with torch.no_grad():\n",
+    "    x_tst = stn_tst.cuda().unsqueeze(0)\n",
+    "    x_tst_lengths = torch.LongTensor([stn_tst.size(0)]).cuda()\n",
+    "    sid = torch.LongTensor([4]).cuda()\n",
+    "    audio = net_g.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=.667, noise_scale_w=0.8, length_scale=1)[0][0,0].data.cpu().float().numpy()\n",
+    "ipd.display(ipd.Audio(audio, rate=hps.data.sampling_rate, normalize=False))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "pycharm": {
+     "name": "#%% md\n"
+    }
+   },
+   "source": [
+    "### Voice Conversion"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "dataset = TextAudioSpeakerLoader(hps.data.validation_files, hps.data)\n",
+    "collate_fn = TextAudioSpeakerCollate()\n",
+    "loader = DataLoader(dataset, num_workers=8, shuffle=False,\n",
+    "    batch_size=1, pin_memory=True,\n",
+    "    drop_last=True, collate_fn=collate_fn)\n",
+    "data_list = list(loader)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "with torch.no_grad():\n",
+    "    x, x_lengths, spec, spec_lengths, y, y_lengths, sid_src = [x.cuda() for x in data_list[0]]\n",
+    "    sid_tgt1 = torch.LongTensor([1]).cuda()\n",
+    "    sid_tgt2 = torch.LongTensor([2]).cuda()\n",
+    "    sid_tgt3 = torch.LongTensor([4]).cuda()\n",
+    "    audio1 = net_g.voice_conversion(spec, spec_lengths, sid_src=sid_src, sid_tgt=sid_tgt1)[0][0,0].data.cpu().float().numpy()\n",
+    "    audio2 = net_g.voice_conversion(spec, spec_lengths, sid_src=sid_src, sid_tgt=sid_tgt2)[0][0,0].data.cpu().float().numpy()\n",
+    "    audio3 = net_g.voice_conversion(spec, spec_lengths, sid_src=sid_src, sid_tgt=sid_tgt3)[0][0,0].data.cpu().float().numpy()\n",
+    "print(\"Original SID: %d\" % sid_src.item())\n",
+    "ipd.display(ipd.Audio(y[0].cpu().numpy(), rate=hps.data.sampling_rate, normalize=False))\n",
+    "print(\"Converted SID: %d\" % sid_tgt1.item())\n",
+    "ipd.display(ipd.Audio(audio1, rate=hps.data.sampling_rate, normalize=False))\n",
+    "print(\"Converted SID: %d\" % sid_tgt2.item())\n",
+    "ipd.display(ipd.Audio(audio2, rate=hps.data.sampling_rate, normalize=False))\n",
+    "print(\"Converted SID: %d\" % sid_tgt3.item())\n",
+    "ipd.display(ipd.Audio(audio3, rate=hps.data.sampling_rate, normalize=False))"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3.7.9 64-bit",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.9"
+  },
+  "vscode": {
+   "interpreter": {
+    "hash": "c15292341d300295ca9f634d04c483f667a0c1d5ee0c309c2ac4e312cce8b8df"
+   }
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

losses.py

@@ -0,0 +1,61 @@
+import torch 
+from torch.nn import functional as F
+
+import commons
+
+
+def feature_loss(fmap_r, fmap_g):
+  loss = 0
+  for dr, dg in zip(fmap_r, fmap_g):
+    for rl, gl in zip(dr, dg):
+      rl = rl.float().detach()
+      gl = gl.float()
+      loss += torch.mean(torch.abs(rl - gl))
+
+  return loss * 2 
+
+
+def discriminator_loss(disc_real_outputs, disc_generated_outputs):
+  loss = 0
+  r_losses = []
+  g_losses = []
+  for dr, dg in zip(disc_real_outputs, disc_generated_outputs):
+    dr = dr.float()
+    dg = dg.float()
+    r_loss = torch.mean((1-dr)**2)
+    g_loss = torch.mean(dg**2)
+    loss += (r_loss + g_loss)
+    r_losses.append(r_loss.item())
+    g_losses.append(g_loss.item())
+
+  return loss, r_losses, g_losses
+
+
+def generator_loss(disc_outputs):
+  loss = 0
+  gen_losses = []
+  for dg in disc_outputs:
+    dg = dg.float()
+    l = torch.mean((1-dg)**2)
+    gen_losses.append(l)
+    loss += l
+
+  return loss, gen_losses
+
+
+def kl_loss(z_p, logs_q, m_p, logs_p, z_mask):
+  """
+  z_p, logs_q: [b, h, t_t]
+  m_p, logs_p: [b, h, t_t]
+  """
+  z_p = z_p.float()
+  logs_q = logs_q.float()
+  m_p = m_p.float()
+  logs_p = logs_p.float()
+  z_mask = z_mask.float()
+
+  kl = logs_p - logs_q - 0.5
+  kl += 0.5 * ((z_p - m_p)**2) * torch.exp(-2. * logs_p)
+  kl = torch.sum(kl * z_mask)
+  l = kl / torch.sum(z_mask)
+  return l

mel_processing.py

@@ -0,0 +1,112 @@
+import math
+import os
+import random
+import torch
+from torch import nn
+import torch.nn.functional as F
+import torch.utils.data
+import numpy as np
+import librosa
+import librosa.util as librosa_util
+from librosa.util import normalize, pad_center, tiny
+from scipy.signal import get_window
+from scipy.io.wavfile import read
+from librosa.filters import mel as librosa_mel_fn
+
+MAX_WAV_VALUE = 32768.0
+
+
+def dynamic_range_compression_torch(x, C=1, clip_val=1e-5):
+    """
+    PARAMS
+    ------
+    C: compression factor
+    """
+    return torch.log(torch.clamp(x, min=clip_val) * C)
+
+
+def dynamic_range_decompression_torch(x, C=1):
+    """
+    PARAMS
+    ------
+    C: compression factor used to compress
+    """
+    return torch.exp(x) / C
+
+
+def spectral_normalize_torch(magnitudes):
+    output = dynamic_range_compression_torch(magnitudes)
+    return output
+
+
+def spectral_de_normalize_torch(magnitudes):
+    output = dynamic_range_decompression_torch(magnitudes)
+    return output
+
+
+mel_basis = {}
+hann_window = {}
+
+
+def spectrogram_torch(y, n_fft, sampling_rate, hop_size, win_size, center=False):
+    if torch.min(y) < -1.:
+        print('min value is ', torch.min(y))
+    if torch.max(y) > 1.:
+        print('max value is ', torch.max(y))
+
+    global hann_window
+    dtype_device = str(y.dtype) + '_' + str(y.device)
+    wnsize_dtype_device = str(win_size) + '_' + dtype_device
+    if wnsize_dtype_device not in hann_window:
+        hann_window[wnsize_dtype_device] = torch.hann_window(win_size).to(dtype=y.dtype, device=y.device)
+
+    y = torch.nn.functional.pad(y.unsqueeze(1), (int((n_fft-hop_size)/2), int((n_fft-hop_size)/2)), mode='reflect')
+    y = y.squeeze(1)
+
+    spec = torch.stft(y, n_fft, hop_length=hop_size, win_length=win_size, window=hann_window[wnsize_dtype_device],
+                      center=center, pad_mode='reflect', normalized=False, onesided=True)
+
+    spec = torch.sqrt(spec.pow(2).sum(-1) + 1e-6)
+    return spec
+
+
+def spec_to_mel_torch(spec, n_fft, num_mels, sampling_rate, fmin, fmax):
+    global mel_basis
+    dtype_device = str(spec.dtype) + '_' + str(spec.device)
+    fmax_dtype_device = str(fmax) + '_' + dtype_device
+    if fmax_dtype_device not in mel_basis:
+        mel = librosa_mel_fn(sampling_rate, n_fft, num_mels, fmin, fmax)
+        mel_basis[fmax_dtype_device] = torch.from_numpy(mel).to(dtype=spec.dtype, device=spec.device)
+    spec = torch.matmul(mel_basis[fmax_dtype_device], spec)
+    spec = spectral_normalize_torch(spec)
+    return spec
+
+
+def mel_spectrogram_torch(y, n_fft, num_mels, sampling_rate, hop_size, win_size, fmin, fmax, center=False):
+    if torch.min(y) < -1.:
+        print('min value is ', torch.min(y))
+    if torch.max(y) > 1.:
+        print('max value is ', torch.max(y))
+
+    global mel_basis, hann_window
+    dtype_device = str(y.dtype) + '_' + str(y.device)
+    fmax_dtype_device = str(fmax) + '_' + dtype_device
+    wnsize_dtype_device = str(win_size) + '_' + dtype_device
+    if fmax_dtype_device not in mel_basis:
+        mel = librosa_mel_fn(sampling_rate, n_fft, num_mels, fmin, fmax)
+        mel_basis[fmax_dtype_device] = torch.from_numpy(mel).to(dtype=y.dtype, device=y.device)
+    if wnsize_dtype_device not in hann_window:
+        hann_window[wnsize_dtype_device] = torch.hann_window(win_size).to(dtype=y.dtype, device=y.device)
+
+    y = torch.nn.functional.pad(y.unsqueeze(1), (int((n_fft-hop_size)/2), int((n_fft-hop_size)/2)), mode='reflect')
+    y = y.squeeze(1)
+
+    spec = torch.stft(y, n_fft, hop_length=hop_size, win_length=win_size, window=hann_window[wnsize_dtype_device],
+                      center=center, pad_mode='reflect', normalized=False, onesided=True)
+
+    spec = torch.sqrt(spec.pow(2).sum(-1) + 1e-6)
+
+    spec = torch.matmul(mel_basis[fmax_dtype_device], spec)
+    spec = spectral_normalize_torch(spec)
+
+    return spec

models.py

@@ -0,0 +1,534 @@
+import copy
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+import commons
+import modules
+import attentions
+import monotonic_align
+
+from torch.nn import Conv1d, ConvTranspose1d, AvgPool1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm, spectral_norm
+from commons import init_weights, get_padding
+
+
+class StochasticDurationPredictor(nn.Module):
+  def __init__(self, in_channels, filter_channels, kernel_size, p_dropout, n_flows=4, gin_channels=0):
+    super().__init__()
+    filter_channels = in_channels # it needs to be removed from future version.
+    self.in_channels = in_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.n_flows = n_flows
+    self.gin_channels = gin_channels
+
+    self.log_flow = modules.Log()
+    self.flows = nn.ModuleList()
+    self.flows.append(modules.ElementwiseAffine(2))
+    for i in range(n_flows):
+      self.flows.append(modules.ConvFlow(2, filter_channels, kernel_size, n_layers=3))
+      self.flows.append(modules.Flip())
+
+    self.post_pre = nn.Conv1d(1, filter_channels, 1)
+    self.post_proj = nn.Conv1d(filter_channels, filter_channels, 1)
+    self.post_convs = modules.DDSConv(filter_channels, kernel_size, n_layers=3, p_dropout=p_dropout)
+    self.post_flows = nn.ModuleList()
+    self.post_flows.append(modules.ElementwiseAffine(2))
+    for i in range(4):
+      self.post_flows.append(modules.ConvFlow(2, filter_channels, kernel_size, n_layers=3))
+      self.post_flows.append(modules.Flip())
+
+    self.pre = nn.Conv1d(in_channels, filter_channels, 1)
+    self.proj = nn.Conv1d(filter_channels, filter_channels, 1)
+    self.convs = modules.DDSConv(filter_channels, kernel_size, n_layers=3, p_dropout=p_dropout)
+    if gin_channels != 0:
+      self.cond = nn.Conv1d(gin_channels, filter_channels, 1)
+
+  def forward(self, x, x_mask, w=None, g=None, reverse=False, noise_scale=1.0):
+    x = torch.detach(x)
+    x = self.pre(x)
+    if g is not None:
+      g = torch.detach(g)
+      x = x + self.cond(g)
+    x = self.convs(x, x_mask)
+    x = self.proj(x) * x_mask
+
+    if not reverse:
+      flows = self.flows
+      assert w is not None
+
+      logdet_tot_q = 0 
+      h_w = self.post_pre(w)
+      h_w = self.post_convs(h_w, x_mask)
+      h_w = self.post_proj(h_w) * x_mask
+      e_q = torch.randn(w.size(0), 2, w.size(2)).to(device=x.device, dtype=x.dtype) * x_mask
+      z_q = e_q
+      for flow in self.post_flows:
+        z_q, logdet_q = flow(z_q, x_mask, g=(x + h_w))
+        logdet_tot_q += logdet_q
+      z_u, z1 = torch.split(z_q, [1, 1], 1) 
+      u = torch.sigmoid(z_u) * x_mask
+      z0 = (w - u) * x_mask
+      logdet_tot_q += torch.sum((F.logsigmoid(z_u) + F.logsigmoid(-z_u)) * x_mask, [1,2])
+      logq = torch.sum(-0.5 * (math.log(2*math.pi) + (e_q**2)) * x_mask, [1,2]) - logdet_tot_q
+
+      logdet_tot = 0
+      z0, logdet = self.log_flow(z0, x_mask)
+      logdet_tot += logdet
+      z = torch.cat([z0, z1], 1)
+      for flow in flows:
+        z, logdet = flow(z, x_mask, g=x, reverse=reverse)
+        logdet_tot = logdet_tot + logdet
+      nll = torch.sum(0.5 * (math.log(2*math.pi) + (z**2)) * x_mask, [1,2]) - logdet_tot
+      return nll + logq # [b]
+    else:
+      flows = list(reversed(self.flows))
+      flows = flows[:-2] + [flows[-1]] # remove a useless vflow
+      z = torch.randn(x.size(0), 2, x.size(2)).to(device=x.device, dtype=x.dtype) * noise_scale
+      for flow in flows:
+        z = flow(z, x_mask, g=x, reverse=reverse)
+      z0, z1 = torch.split(z, [1, 1], 1)
+      logw = z0
+      return logw
+
+
+class DurationPredictor(nn.Module):
+  def __init__(self, in_channels, filter_channels, kernel_size, p_dropout, gin_channels=0):
+    super().__init__()
+
+    self.in_channels = in_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.gin_channels = gin_channels
+
+    self.drop = nn.Dropout(p_dropout)
+    self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size, padding=kernel_size//2)
+    self.norm_1 = modules.LayerNorm(filter_channels)
+    self.conv_2 = nn.Conv1d(filter_channels, filter_channels, kernel_size, padding=kernel_size//2)
+    self.norm_2 = modules.LayerNorm(filter_channels)
+    self.proj = nn.Conv1d(filter_channels, 1, 1)
+
+    if gin_channels != 0:
+      self.cond = nn.Conv1d(gin_channels, in_channels, 1)
+
+  def forward(self, x, x_mask, g=None):
+    x = torch.detach(x)
+    if g is not None:
+      g = torch.detach(g)
+      x = x + self.cond(g)
+    x = self.conv_1(x * x_mask)
+    x = torch.relu(x)
+    x = self.norm_1(x)
+    x = self.drop(x)
+    x = self.conv_2(x * x_mask)
+    x = torch.relu(x)
+    x = self.norm_2(x)
+    x = self.drop(x)
+    x = self.proj(x * x_mask)
+    return x * x_mask
+
+
+class TextEncoder(nn.Module):
+  def __init__(self,
+      n_vocab,
+      out_channels,
+      hidden_channels,
+      filter_channels,
+      n_heads,
+      n_layers,
+      kernel_size,
+      p_dropout):
+    super().__init__()
+    self.n_vocab = n_vocab
+    self.out_channels = out_channels
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+
+    self.emb = nn.Embedding(n_vocab, hidden_channels)
+    nn.init.normal_(self.emb.weight, 0.0, hidden_channels**-0.5)
+
+    self.encoder = attentions.Encoder(
+      hidden_channels,
+      filter_channels,
+      n_heads,
+      n_layers,
+      kernel_size,
+      p_dropout)
+    self.proj= nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+  def forward(self, x, x_lengths):
+    x = self.emb(x) * math.sqrt(self.hidden_channels) # [b, t, h]
+    x = torch.transpose(x, 1, -1) # [b, h, t]
+    x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(x.dtype)
+
+    x = self.encoder(x * x_mask, x_mask)
+    stats = self.proj(x) * x_mask
+
+    m, logs = torch.split(stats, self.out_channels, dim=1)
+    return x, m, logs, x_mask
+
+
+class ResidualCouplingBlock(nn.Module):
+  def __init__(self,
+      channels,
+      hidden_channels,
+      kernel_size,
+      dilation_rate,
+      n_layers,
+      n_flows=4,
+      gin_channels=0):
+    super().__init__()
+    self.channels = channels
+    self.hidden_channels = hidden_channels
+    self.kernel_size = kernel_size
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.n_flows = n_flows
+    self.gin_channels = gin_channels
+
+    self.flows = nn.ModuleList()
+    for i in range(n_flows):
+      self.flows.append(modules.ResidualCouplingLayer(channels, hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=gin_channels, mean_only=True))
+      self.flows.append(modules.Flip())
+
+  def forward(self, x, x_mask, g=None, reverse=False):
+    if not reverse:
+      for flow in self.flows:
+        x, _ = flow(x, x_mask, g=g, reverse=reverse)
+    else:
+      for flow in reversed(self.flows):
+        x = flow(x, x_mask, g=g, reverse=reverse)
+    return x
+
+
+class PosteriorEncoder(nn.Module):
+  def __init__(self,
+      in_channels,
+      out_channels,
+      hidden_channels,
+      kernel_size,
+      dilation_rate,
+      n_layers,
+      gin_channels=0):
+    super().__init__()
+    self.in_channels = in_channels
+    self.out_channels = out_channels
+    self.hidden_channels = hidden_channels
+    self.kernel_size = kernel_size
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.gin_channels = gin_channels
+
+    self.pre = nn.Conv1d(in_channels, hidden_channels, 1)
+    self.enc = modules.WN(hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=gin_channels)
+    self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+  def forward(self, x, x_lengths, g=None):
+    x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(x.dtype)
+    x = self.pre(x) * x_mask
+    x = self.enc(x, x_mask, g=g)
+    stats = self.proj(x) * x_mask
+    m, logs = torch.split(stats, self.out_channels, dim=1)
+    z = (m + torch.randn_like(m) * torch.exp(logs)) * x_mask
+    return z, m, logs, x_mask
+
+
+class Generator(torch.nn.Module):
+    def __init__(self, initial_channel, resblock, resblock_kernel_sizes, resblock_dilation_sizes, upsample_rates, upsample_initial_channel, upsample_kernel_sizes, gin_channels=0):
+        super(Generator, self).__init__()
+        self.num_kernels = len(resblock_kernel_sizes)
+        self.num_upsamples = len(upsample_rates)
+        self.conv_pre = Conv1d(initial_channel, upsample_initial_channel, 7, 1, padding=3)
+        resblock = modules.ResBlock1 if resblock == '1' else modules.ResBlock2
+
+        self.ups = nn.ModuleList()
+        for i, (u, k) in enumerate(zip(upsample_rates, upsample_kernel_sizes)):
+            self.ups.append(weight_norm(
+                ConvTranspose1d(upsample_initial_channel//(2**i), upsample_initial_channel//(2**(i+1)),
+                                k, u, padding=(k-u)//2)))
+
+        self.resblocks = nn.ModuleList()
+        for i in range(len(self.ups)):
+            ch = upsample_initial_channel//(2**(i+1))
+            for j, (k, d) in enumerate(zip(resblock_kernel_sizes, resblock_dilation_sizes)):
+                self.resblocks.append(resblock(ch, k, d))
+
+        self.conv_post = Conv1d(ch, 1, 7, 1, padding=3, bias=False)
+        self.ups.apply(init_weights)
+
+        if gin_channels != 0:
+            self.cond = nn.Conv1d(gin_channels, upsample_initial_channel, 1)
+
+    def forward(self, x, g=None):
+        x = self.conv_pre(x)
+        if g is not None:
+          x = x + self.cond(g)
+
+        for i in range(self.num_upsamples):
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            x = self.ups[i](x)
+            xs = None
+            for j in range(self.num_kernels):
+                if xs is None:
+                    xs = self.resblocks[i*self.num_kernels+j](x)
+                else:
+                    xs += self.resblocks[i*self.num_kernels+j](x)
+            x = xs / self.num_kernels
+        x = F.leaky_relu(x)
+        x = self.conv_post(x)
+        x = torch.tanh(x)
+
+        return x
+
+    def remove_weight_norm(self):
+        print('Removing weight norm...')
+        for l in self.ups:
+            remove_weight_norm(l)
+        for l in self.resblocks:
+            l.remove_weight_norm()
+
+
+class DiscriminatorP(torch.nn.Module):
+    def __init__(self, period, kernel_size=5, stride=3, use_spectral_norm=False):
+        super(DiscriminatorP, self).__init__()
+        self.period = period
+        self.use_spectral_norm = use_spectral_norm
+        norm_f = weight_norm if use_spectral_norm == False else spectral_norm
+        self.convs = nn.ModuleList([
+            norm_f(Conv2d(1, 32, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(32, 128, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(128, 512, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(512, 1024, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(1024, 1024, (kernel_size, 1), 1, padding=(get_padding(kernel_size, 1), 0))),
+        ])
+        self.conv_post = norm_f(Conv2d(1024, 1, (3, 1), 1, padding=(1, 0)))
+
+    def forward(self, x):
+        fmap = []
+
+        # 1d to 2d
+        b, c, t = x.shape
+        if t % self.period != 0: # pad first
+            n_pad = self.period - (t % self.period)
+            x = F.pad(x, (0, n_pad), "reflect")
+            t = t + n_pad
+        x = x.view(b, c, t // self.period, self.period)
+
+        for l in self.convs:
+            x = l(x)
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            fmap.append(x)
+        x = self.conv_post(x)
+        fmap.append(x)
+        x = torch.flatten(x, 1, -1)
+
+        return x, fmap
+
+
+class DiscriminatorS(torch.nn.Module):
+    def __init__(self, use_spectral_norm=False):
+        super(DiscriminatorS, self).__init__()
+        norm_f = weight_norm if use_spectral_norm == False else spectral_norm
+        self.convs = nn.ModuleList([
+            norm_f(Conv1d(1, 16, 15, 1, padding=7)),
+            norm_f(Conv1d(16, 64, 41, 4, groups=4, padding=20)),
+            norm_f(Conv1d(64, 256, 41, 4, groups=16, padding=20)),
+            norm_f(Conv1d(256, 1024, 41, 4, groups=64, padding=20)),
+            norm_f(Conv1d(1024, 1024, 41, 4, groups=256, padding=20)),
+            norm_f(Conv1d(1024, 1024, 5, 1, padding=2)),
+        ])
+        self.conv_post = norm_f(Conv1d(1024, 1, 3, 1, padding=1))
+
+    def forward(self, x):
+        fmap = []
+
+        for l in self.convs:
+            x = l(x)
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            fmap.append(x)
+        x = self.conv_post(x)
+        fmap.append(x)
+        x = torch.flatten(x, 1, -1)
+
+        return x, fmap
+
+
+class MultiPeriodDiscriminator(torch.nn.Module):
+    def __init__(self, use_spectral_norm=False):
+        super(MultiPeriodDiscriminator, self).__init__()
+        periods = [2,3,5,7,11]
+
+        discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
+        discs = discs + [DiscriminatorP(i, use_spectral_norm=use_spectral_norm) for i in periods]
+        self.discriminators = nn.ModuleList(discs)
+
+    def forward(self, y, y_hat):
+        y_d_rs = []
+        y_d_gs = []
+        fmap_rs = []
+        fmap_gs = []
+        for i, d in enumerate(self.discriminators):
+            y_d_r, fmap_r = d(y)
+            y_d_g, fmap_g = d(y_hat)
+            y_d_rs.append(y_d_r)
+            y_d_gs.append(y_d_g)
+            fmap_rs.append(fmap_r)
+            fmap_gs.append(fmap_g)
+
+        return y_d_rs, y_d_gs, fmap_rs, fmap_gs
+
+
+
+class SynthesizerTrn(nn.Module):
+  """
+  Synthesizer for Training
+  """
+
+  def __init__(self, 
+    n_vocab,
+    spec_channels,
+    segment_size,
+    inter_channels,
+    hidden_channels,
+    filter_channels,
+    n_heads,
+    n_layers,
+    kernel_size,
+    p_dropout,
+    resblock, 
+    resblock_kernel_sizes, 
+    resblock_dilation_sizes, 
+    upsample_rates, 
+    upsample_initial_channel, 
+    upsample_kernel_sizes,
+    n_speakers=0,
+    gin_channels=0,
+    use_sdp=True,
+    **kwargs):
+
+    super().__init__()
+    self.n_vocab = n_vocab
+    self.spec_channels = spec_channels
+    self.inter_channels = inter_channels
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.resblock = resblock
+    self.resblock_kernel_sizes = resblock_kernel_sizes
+    self.resblock_dilation_sizes = resblock_dilation_sizes
+    self.upsample_rates = upsample_rates
+    self.upsample_initial_channel = upsample_initial_channel
+    self.upsample_kernel_sizes = upsample_kernel_sizes
+    self.segment_size = segment_size
+    self.n_speakers = n_speakers
+    self.gin_channels = gin_channels
+
+    self.use_sdp = use_sdp
+
+    self.enc_p = TextEncoder(n_vocab,
+        inter_channels,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size,
+        p_dropout)
+    self.dec = Generator(inter_channels, resblock, resblock_kernel_sizes, resblock_dilation_sizes, upsample_rates, upsample_initial_channel, upsample_kernel_sizes, gin_channels=gin_channels)
+    self.enc_q = PosteriorEncoder(spec_channels, inter_channels, hidden_channels, 5, 1, 16, gin_channels=gin_channels)
+    self.flow = ResidualCouplingBlock(inter_channels, hidden_channels, 5, 1, 4, gin_channels=gin_channels)
+
+    if use_sdp:
+      self.dp = StochasticDurationPredictor(hidden_channels, 192, 3, 0.5, 4, gin_channels=gin_channels)
+    else:
+      self.dp = DurationPredictor(hidden_channels, 256, 3, 0.5, gin_channels=gin_channels)
+
+    if n_speakers > 1:
+      self.emb_g = nn.Embedding(n_speakers, gin_channels)
+
+  def forward(self, x, x_lengths, y, y_lengths, sid=None):
+
+    x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths)
+    if self.n_speakers > 0:
+      g = self.emb_g(sid).unsqueeze(-1) # [b, h, 1]
+    else:
+      g = None
+
+    z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
+    z_p = self.flow(z, y_mask, g=g)
+
+    with torch.no_grad():
+      # negative cross-entropy
+      s_p_sq_r = torch.exp(-2 * logs_p) # [b, d, t]
+      neg_cent1 = torch.sum(-0.5 * math.log(2 * math.pi) - logs_p, [1], keepdim=True) # [b, 1, t_s]
+      neg_cent2 = torch.matmul(-0.5 * (z_p ** 2).transpose(1, 2), s_p_sq_r) # [b, t_t, d] x [b, d, t_s] = [b, t_t, t_s]
+      neg_cent3 = torch.matmul(z_p.transpose(1, 2), (m_p * s_p_sq_r)) # [b, t_t, d] x [b, d, t_s] = [b, t_t, t_s]
+      neg_cent4 = torch.sum(-0.5 * (m_p ** 2) * s_p_sq_r, [1], keepdim=True) # [b, 1, t_s]
+      neg_cent = neg_cent1 + neg_cent2 + neg_cent3 + neg_cent4
+
+      attn_mask = torch.unsqueeze(x_mask, 2) * torch.unsqueeze(y_mask, -1)
+      attn = monotonic_align.maximum_path(neg_cent, attn_mask.squeeze(1)).unsqueeze(1).detach()
+
+    w = attn.sum(2)
+    if self.use_sdp:
+      l_length = self.dp(x, x_mask, w, g=g)
+      l_length = l_length / torch.sum(x_mask)
+    else:
+      logw_ = torch.log(w + 1e-6) * x_mask
+      logw = self.dp(x, x_mask, g=g)
+      l_length = torch.sum((logw - logw_)**2, [1,2]) / torch.sum(x_mask) # for averaging 
+
+    # expand prior
+    m_p = torch.matmul(attn.squeeze(1), m_p.transpose(1, 2)).transpose(1, 2)
+    logs_p = torch.matmul(attn.squeeze(1), logs_p.transpose(1, 2)).transpose(1, 2)
+
+    z_slice, ids_slice = commons.rand_slice_segments(z, y_lengths, self.segment_size)
+    o = self.dec(z_slice, g=g)
+    return o, l_length, attn, ids_slice, x_mask, y_mask, (z, z_p, m_p, logs_p, m_q, logs_q)
+
+  def infer(self, x, x_lengths, sid=None, noise_scale=1, length_scale=1, noise_scale_w=1., max_len=None):
+    x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths)
+    if self.n_speakers > 0:
+      g = self.emb_g(sid).unsqueeze(-1) # [b, h, 1]
+    else:
+      g = None
+
+    if self.use_sdp:
+      logw = self.dp(x, x_mask, g=g, reverse=True, noise_scale=noise_scale_w)
+    else:
+      logw = self.dp(x, x_mask, g=g)
+    w = torch.exp(logw) * x_mask * length_scale
+    w_ceil = torch.ceil(w)
+    y_lengths = torch.clamp_min(torch.sum(w_ceil, [1, 2]), 1).long()
+    y_mask = torch.unsqueeze(commons.sequence_mask(y_lengths, None), 1).to(x_mask.dtype)
+    attn_mask = torch.unsqueeze(x_mask, 2) * torch.unsqueeze(y_mask, -1)
+    attn = commons.generate_path(w_ceil, attn_mask)
+
+    m_p = torch.matmul(attn.squeeze(1), m_p.transpose(1, 2)).transpose(1, 2) # [b, t', t], [b, t, d] -> [b, d, t']
+    logs_p = torch.matmul(attn.squeeze(1), logs_p.transpose(1, 2)).transpose(1, 2) # [b, t', t], [b, t, d] -> [b, d, t']
+
+    z_p = m_p + torch.randn_like(m_p) * torch.exp(logs_p) * noise_scale
+    z = self.flow(z_p, y_mask, g=g, reverse=True)
+    o = self.dec((z * y_mask)[:,:,:max_len], g=g)
+    return o, attn, y_mask, (z, z_p, m_p, logs_p)
+
+  def voice_conversion(self, y, y_lengths, sid_src, sid_tgt):
+    assert self.n_speakers > 0, "n_speakers have to be larger than 0."
+    g_src = self.emb_g(sid_src).unsqueeze(-1)
+    g_tgt = self.emb_g(sid_tgt).unsqueeze(-1)
+    z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g_src)
+    z_p = self.flow(z, y_mask, g=g_src)
+    z_hat = self.flow(z_p, y_mask, g=g_tgt, reverse=True)
+    o_hat = self.dec(z_hat * y_mask, g=g_tgt)
+    return o_hat, y_mask, (z, z_p, z_hat)
+

models/model_info.json

@@ -0,0 +1 @@
+{"Yuuka": {"name_en": "Yuuka", "name_zh": "\u65e9\u6fd1\u4f18\u9999", "cover": "models/Yuuka/cover.png", "sid": 0, "example": "\u5148\u751f\u3002\u4eca\u65e5\u3082\u5168\u529b\u3067\u3042\u306a\u305f\u3092\u30a2\u30b7\u30b9\u30c8\u3057\u307e\u3059\u306d\u3002", "language": "JP", "type": "single", "model_path": "models\\Yuuka\\Yuuka.pth"}, "Mika": {"name_en": "Mika", "name_zh": "\u5723\u56ed\u672a\u82b1", "cover": "models\\Mika\\cover.png", "sid": "0", "example": "\u304a\u304b\u3048\u308a\u3001\u5148\u751f\uff01\u3061\u3083\u30fc\u3093\u3068\u3044\u3044\u5b50\u3067\u304a\u7559\u5b88\u756a\u3057\u3066\u305f\u3088\u3002", "language": "JP", "type": "single", "model_path": "models\\Mika\\Mika.pth"}}

models/parappa/config.json

@@ -0,0 +1,94 @@
+{
+  "train": {
+    "log_interval": 100,
+    "eval_interval": 500,
+    "seed": 109652,
+    "epochs": 20001,
+    "learning_rate": 0.00015,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 30,
+    "fp16_run": false,
+    "bf16_run": false,
+    "lr_decay": 0.999875,
+    "segment_size": 10240,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "use_sr": true,
+    "max_speclen": 512,
+    "port": "8001",
+    "keep_ckpts": 3
+  },
+  "data": {
+    "training_files": "filelists/44k/train.txt",
+    "validation_files": "filelists/44k/val.txt",
+    "max_wav_value": 32768.0,
+    "sampling_rate": 44100,
+    "filter_length": 2048,
+    "hop_length": 512,
+    "win_length": 2048,
+    "n_mel_channels": 80,
+    "mel_fmin": 0.0,
+    "mel_fmax": 22050
+  },
+  "model": {
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.1,
+    "resblock": "1",
+    "resblock_kernel_sizes": [
+      3,
+      7,
+      11
+    ],
+    "resblock_dilation_sizes": [
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ]
+    ],
+    "upsample_rates": [
+      8,
+      8,
+      2,
+      2,
+      2
+    ],
+    "upsample_initial_channel": 512,
+    "upsample_kernel_sizes": [
+      16,
+      16,
+      4,
+      4,
+      4
+    ],
+    "n_layers_q": 3,
+    "use_spectral_norm": false,
+    "gin_channels": 256,
+    "ssl_dim": 256,
+    "n_speakers": 200
+  },
+  "spk": {
+    "parappa": 0
+  }
+}

models/parappa/path.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a712fbabc95b65036407a9a94d85c40630c43cd70193fff3735ba73468dc538
+size 542789469

modules.py

@@ -0,0 +1,390 @@
+import copy
+import math
+import numpy as np
+import scipy
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+from torch.nn import Conv1d, ConvTranspose1d, AvgPool1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm
+
+import commons
+from commons import init_weights, get_padding
+from transforms import piecewise_rational_quadratic_transform
+
+
+LRELU_SLOPE = 0.1
+
+
+class LayerNorm(nn.Module):
+  def __init__(self, channels, eps=1e-5):
+    super().__init__()
+    self.channels = channels
+    self.eps = eps
+
+    self.gamma = nn.Parameter(torch.ones(channels))
+    self.beta = nn.Parameter(torch.zeros(channels))
+
+  def forward(self, x):
+    x = x.transpose(1, -1)
+    x = F.layer_norm(x, (self.channels,), self.gamma, self.beta, self.eps)
+    return x.transpose(1, -1)
+
+ 
+class ConvReluNorm(nn.Module):
+  def __init__(self, in_channels, hidden_channels, out_channels, kernel_size, n_layers, p_dropout):
+    super().__init__()
+    self.in_channels = in_channels
+    self.hidden_channels = hidden_channels
+    self.out_channels = out_channels
+    self.kernel_size = kernel_size
+    self.n_layers = n_layers
+    self.p_dropout = p_dropout
+    assert n_layers > 1, "Number of layers should be larger than 0."
+
+    self.conv_layers = nn.ModuleList()
+    self.norm_layers = nn.ModuleList()
+    self.conv_layers.append(nn.Conv1d(in_channels, hidden_channels, kernel_size, padding=kernel_size//2))
+    self.norm_layers.append(LayerNorm(hidden_channels))
+    self.relu_drop = nn.Sequential(
+        nn.ReLU(),
+        nn.Dropout(p_dropout))
+    for _ in range(n_layers-1):
+      self.conv_layers.append(nn.Conv1d(hidden_channels, hidden_channels, kernel_size, padding=kernel_size//2))
+      self.norm_layers.append(LayerNorm(hidden_channels))
+    self.proj = nn.Conv1d(hidden_channels, out_channels, 1)
+    self.proj.weight.data.zero_()
+    self.proj.bias.data.zero_()
+
+  def forward(self, x, x_mask):
+    x_org = x
+    for i in range(self.n_layers):
+      x = self.conv_layers[i](x * x_mask)
+      x = self.norm_layers[i](x)
+      x = self.relu_drop(x)
+    x = x_org + self.proj(x)
+    return x * x_mask
+
+
+class DDSConv(nn.Module):
+  """
+  Dialted and Depth-Separable Convolution
+  """
+  def __init__(self, channels, kernel_size, n_layers, p_dropout=0.):
+    super().__init__()
+    self.channels = channels
+    self.kernel_size = kernel_size
+    self.n_layers = n_layers
+    self.p_dropout = p_dropout
+
+    self.drop = nn.Dropout(p_dropout)
+    self.convs_sep = nn.ModuleList()
+    self.convs_1x1 = nn.ModuleList()
+    self.norms_1 = nn.ModuleList()
+    self.norms_2 = nn.ModuleList()
+    for i in range(n_layers):
+      dilation = kernel_size ** i
+      padding = (kernel_size * dilation - dilation) // 2
+      self.convs_sep.append(nn.Conv1d(channels, channels, kernel_size, 
+          groups=channels, dilation=dilation, padding=padding
+      ))
+      self.convs_1x1.append(nn.Conv1d(channels, channels, 1))
+      self.norms_1.append(LayerNorm(channels))
+      self.norms_2.append(LayerNorm(channels))
+
+  def forward(self, x, x_mask, g=None):
+    if g is not None:
+      x = x + g
+    for i in range(self.n_layers):
+      y = self.convs_sep[i](x * x_mask)
+      y = self.norms_1[i](y)
+      y = F.gelu(y)
+      y = self.convs_1x1[i](y)
+      y = self.norms_2[i](y)
+      y = F.gelu(y)
+      y = self.drop(y)
+      x = x + y
+    return x * x_mask
+
+
+class WN(torch.nn.Module):
+  def __init__(self, hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=0, p_dropout=0):
+    super(WN, self).__init__()
+    assert(kernel_size % 2 == 1)
+    self.hidden_channels =hidden_channels
+    self.kernel_size = kernel_size,
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.gin_channels = gin_channels
+    self.p_dropout = p_dropout
+
+    self.in_layers = torch.nn.ModuleList()
+    self.res_skip_layers = torch.nn.ModuleList()
+    self.drop = nn.Dropout(p_dropout)
+
+    if gin_channels != 0:
+      cond_layer = torch.nn.Conv1d(gin_channels, 2*hidden_channels*n_layers, 1)
+      self.cond_layer = torch.nn.utils.weight_norm(cond_layer, name='weight')
+
+    for i in range(n_layers):
+      dilation = dilation_rate ** i
+      padding = int((kernel_size * dilation - dilation) / 2)
+      in_layer = torch.nn.Conv1d(hidden_channels, 2*hidden_channels, kernel_size,
+                                 dilation=dilation, padding=padding)
+      in_layer = torch.nn.utils.weight_norm(in_layer, name='weight')
+      self.in_layers.append(in_layer)
+
+      # last one is not necessary
+      if i < n_layers - 1:
+        res_skip_channels = 2 * hidden_channels
+      else:
+        res_skip_channels = hidden_channels
+
+      res_skip_layer = torch.nn.Conv1d(hidden_channels, res_skip_channels, 1)
+      res_skip_layer = torch.nn.utils.weight_norm(res_skip_layer, name='weight')
+      self.res_skip_layers.append(res_skip_layer)
+
+  def forward(self, x, x_mask, g=None, **kwargs):
+    output = torch.zeros_like(x)
+    n_channels_tensor = torch.IntTensor([self.hidden_channels])
+
+    if g is not None:
+      g = self.cond_layer(g)
+
+    for i in range(self.n_layers):
+      x_in = self.in_layers[i](x)
+      if g is not None:
+        cond_offset = i * 2 * self.hidden_channels
+        g_l = g[:,cond_offset:cond_offset+2*self.hidden_channels,:]
+      else:
+        g_l = torch.zeros_like(x_in)
+
+      acts = commons.fused_add_tanh_sigmoid_multiply(
+          x_in,
+          g_l,
+          n_channels_tensor)
+      acts = self.drop(acts)
+
+      res_skip_acts = self.res_skip_layers[i](acts)
+      if i < self.n_layers - 1:
+        res_acts = res_skip_acts[:,:self.hidden_channels,:]
+        x = (x + res_acts) * x_mask
+        output = output + res_skip_acts[:,self.hidden_channels:,:]
+      else:
+        output = output + res_skip_acts
+    return output * x_mask
+
+  def remove_weight_norm(self):
+    if self.gin_channels != 0:
+      torch.nn.utils.remove_weight_norm(self.cond_layer)
+    for l in self.in_layers:
+      torch.nn.utils.remove_weight_norm(l)
+    for l in self.res_skip_layers:
+     torch.nn.utils.remove_weight_norm(l)
+
+
+class ResBlock1(torch.nn.Module):
+    def __init__(self, channels, kernel_size=3, dilation=(1, 3, 5)):
+        super(ResBlock1, self).__init__()
+        self.convs1 = nn.ModuleList([
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[0],
+                               padding=get_padding(kernel_size, dilation[0]))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[1],
+                               padding=get_padding(kernel_size, dilation[1]))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[2],
+                               padding=get_padding(kernel_size, dilation[2])))
+        ])
+        self.convs1.apply(init_weights)
+
+        self.convs2 = nn.ModuleList([
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
+                               padding=get_padding(kernel_size, 1))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
+                               padding=get_padding(kernel_size, 1))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
+                               padding=get_padding(kernel_size, 1)))
+        ])
+        self.convs2.apply(init_weights)
+
+    def forward(self, x, x_mask=None):
+        for c1, c2 in zip(self.convs1, self.convs2):
+            xt = F.leaky_relu(x, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c1(xt)
+            xt = F.leaky_relu(xt, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c2(xt)
+            x = xt + x
+        if x_mask is not None:
+            x = x * x_mask
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.convs1:
+            remove_weight_norm(l)
+        for l in self.convs2:
+            remove_weight_norm(l)
+
+
+class ResBlock2(torch.nn.Module):
+    def __init__(self, channels, kernel_size=3, dilation=(1, 3)):
+        super(ResBlock2, self).__init__()
+        self.convs = nn.ModuleList([
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[0],
+                               padding=get_padding(kernel_size, dilation[0]))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[1],
+                               padding=get_padding(kernel_size, dilation[1])))
+        ])
+        self.convs.apply(init_weights)
+
+    def forward(self, x, x_mask=None):
+        for c in self.convs:
+            xt = F.leaky_relu(x, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c(xt)
+            x = xt + x
+        if x_mask is not None:
+            x = x * x_mask
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.convs:
+            remove_weight_norm(l)
+
+
+class Log(nn.Module):
+  def forward(self, x, x_mask, reverse=False, **kwargs):
+    if not reverse:
+      y = torch.log(torch.clamp_min(x, 1e-5)) * x_mask
+      logdet = torch.sum(-y, [1, 2])
+      return y, logdet
+    else:
+      x = torch.exp(x) * x_mask
+      return x
+    
+
+class Flip(nn.Module):
+  def forward(self, x, *args, reverse=False, **kwargs):
+    x = torch.flip(x, [1])
+    if not reverse:
+      logdet = torch.zeros(x.size(0)).to(dtype=x.dtype, device=x.device)
+      return x, logdet
+    else:
+      return x
+
+
+class ElementwiseAffine(nn.Module):
+  def __init__(self, channels):
+    super().__init__()
+    self.channels = channels
+    self.m = nn.Parameter(torch.zeros(channels,1))
+    self.logs = nn.Parameter(torch.zeros(channels,1))
+
+  def forward(self, x, x_mask, reverse=False, **kwargs):
+    if not reverse:
+      y = self.m + torch.exp(self.logs) * x
+      y = y * x_mask
+      logdet = torch.sum(self.logs * x_mask, [1,2])
+      return y, logdet
+    else:
+      x = (x - self.m) * torch.exp(-self.logs) * x_mask
+      return x
+
+
+class ResidualCouplingLayer(nn.Module):
+  def __init__(self,
+      channels,
+      hidden_channels,
+      kernel_size,
+      dilation_rate,
+      n_layers,
+      p_dropout=0,
+      gin_channels=0,
+      mean_only=False):
+    assert channels % 2 == 0, "channels should be divisible by 2"
+    super().__init__()
+    self.channels = channels
+    self.hidden_channels = hidden_channels
+    self.kernel_size = kernel_size
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.half_channels = channels // 2
+    self.mean_only = mean_only
+
+    self.pre = nn.Conv1d(self.half_channels, hidden_channels, 1)
+    self.enc = WN(hidden_channels, kernel_size, dilation_rate, n_layers, p_dropout=p_dropout, gin_channels=gin_channels)
+    self.post = nn.Conv1d(hidden_channels, self.half_channels * (2 - mean_only), 1)
+    self.post.weight.data.zero_()
+    self.post.bias.data.zero_()
+
+  def forward(self, x, x_mask, g=None, reverse=False):
+    x0, x1 = torch.split(x, [self.half_channels]*2, 1)
+    h = self.pre(x0) * x_mask
+    h = self.enc(h, x_mask, g=g)
+    stats = self.post(h) * x_mask
+    if not self.mean_only:
+      m, logs = torch.split(stats, [self.half_channels]*2, 1)
+    else:
+      m = stats
+      logs = torch.zeros_like(m)
+
+    if not reverse:
+      x1 = m + x1 * torch.exp(logs) * x_mask
+      x = torch.cat([x0, x1], 1)
+      logdet = torch.sum(logs, [1,2])
+      return x, logdet
+    else:
+      x1 = (x1 - m) * torch.exp(-logs) * x_mask
+      x = torch.cat([x0, x1], 1)
+      return x
+
+
+class ConvFlow(nn.Module):
+  def __init__(self, in_channels, filter_channels, kernel_size, n_layers, num_bins=10, tail_bound=5.0):
+    super().__init__()
+    self.in_channels = in_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.n_layers = n_layers
+    self.num_bins = num_bins
+    self.tail_bound = tail_bound
+    self.half_channels = in_channels // 2
+
+    self.pre = nn.Conv1d(self.half_channels, filter_channels, 1)
+    self.convs = DDSConv(filter_channels, kernel_size, n_layers, p_dropout=0.)
+    self.proj = nn.Conv1d(filter_channels, self.half_channels * (num_bins * 3 - 1), 1)
+    self.proj.weight.data.zero_()
+    self.proj.bias.data.zero_()
+
+  def forward(self, x, x_mask, g=None, reverse=False):
+    x0, x1 = torch.split(x, [self.half_channels]*2, 1)
+    h = self.pre(x0)
+    h = self.convs(h, x_mask, g=g)
+    h = self.proj(h) * x_mask
+
+    b, c, t = x0.shape
+    h = h.reshape(b, c, -1, t).permute(0, 1, 3, 2) # [b, cx?, t] -> [b, c, t, ?]
+
+    unnormalized_widths = h[..., :self.num_bins] / math.sqrt(self.filter_channels)
+    unnormalized_heights = h[..., self.num_bins:2*self.num_bins] / math.sqrt(self.filter_channels)
+    unnormalized_derivatives = h[..., 2 * self.num_bins:]
+
+    x1, logabsdet = piecewise_rational_quadratic_transform(x1,
+        unnormalized_widths,
+        unnormalized_heights,
+        unnormalized_derivatives,
+        inverse=reverse,
+        tails='linear',
+        tail_bound=self.tail_bound
+    )
+
+    x = torch.cat([x0, x1], 1) * x_mask
+    logdet = torch.sum(logabsdet * x_mask, [1,2])
+    if not reverse:
+        return x, logdet
+    else:
+        return x

monotonic_align/__init__.py

@@ -0,0 +1,19 @@
+import numpy as np
+import torch
+from .monotonic_align.core import maximum_path_c
+
+
+def maximum_path(neg_cent, mask):
+  """ Cython optimized version.
+  neg_cent: [b, t_t, t_s]
+  mask: [b, t_t, t_s]
+  """
+  device = neg_cent.device
+  dtype = neg_cent.dtype
+  neg_cent = neg_cent.data.cpu().numpy().astype(np.float32)
+  path = np.zeros(neg_cent.shape, dtype=np.int32)
+
+  t_t_max = mask.sum(1)[:, 0].data.cpu().numpy().astype(np.int32)
+  t_s_max = mask.sum(2)[:, 0].data.cpu().numpy().astype(np.int32)
+  maximum_path_c(path, neg_cent, t_t_max, t_s_max)
+  return torch.from_numpy(path).to(device=device, dtype=dtype)

monotonic_align/core.pyx

@@ -0,0 +1,42 @@
+cimport cython
+from cython.parallel import prange
+
+
+@cython.boundscheck(False)
+@cython.wraparound(False)
+cdef void maximum_path_each(int[:,::1] path, float[:,::1] value, int t_y, int t_x, float max_neg_val=-1e9) nogil:
+  cdef int x
+  cdef int y
+  cdef float v_prev
+  cdef float v_cur
+  cdef float tmp
+  cdef int index = t_x - 1
+
+  for y in range(t_y):
+    for x in range(max(0, t_x + y - t_y), min(t_x, y + 1)):
+      if x == y:
+        v_cur = max_neg_val
+      else:
+        v_cur = value[y-1, x]
+      if x == 0:
+        if y == 0:
+          v_prev = 0.
+        else:
+          v_prev = max_neg_val
+      else:
+        v_prev = value[y-1, x-1]
+      value[y, x] += max(v_prev, v_cur)
+
+  for y in range(t_y - 1, -1, -1):
+    path[y, index] = 1
+    if index != 0 and (index == y or value[y-1, index] < value[y-1, index-1]):
+      index = index - 1
+
+
+@cython.boundscheck(False)
+@cython.wraparound(False)
+cpdef void maximum_path_c(int[:,:,::1] paths, float[:,:,::1] values, int[::1] t_ys, int[::1] t_xs) nogil:
+  cdef int b = paths.shape[0]
+  cdef int i
+  for i in prange(b, nogil=True):
+    maximum_path_each(paths[i], values[i], t_ys[i], t_xs[i])

monotonic_align/setup.py

@@ -0,0 +1,9 @@
+from distutils.core import setup
+from Cython.Build import cythonize
+import numpy
+
+setup(
+  name = 'monotonic_align',
+  ext_modules = cythonize("core.pyx"),
+  include_dirs=[numpy.get_include()]
+)

preprocess.py

@@ -0,0 +1,25 @@
+import argparse
+import text
+from utils import load_filepaths_and_text
+
+if __name__ == '__main__':
+  parser = argparse.ArgumentParser()
+  parser.add_argument("--out_extension", default="cleaned")
+  parser.add_argument("--text_index", default=2, type=int)
+  parser.add_argument("--filelists", nargs="+", default=["filelists/yuuka_train.txt", "filelists/yuuka_val.txt"])
+  parser.add_argument("--text_cleaners", nargs="+", default=["japanese_cleaners"])
+
+  args = parser.parse_args()
+    
+
+  for filelist in args.filelists:
+    print("START:", filelist)
+    filepaths_and_text = load_filepaths_and_text(filelist)
+    for i in range(len(filepaths_and_text)):
+      original_text = filepaths_and_text[i][args.text_index]
+      cleaned_text = text._clean_text(original_text, args.text_cleaners)
+      filepaths_and_text[i][args.text_index] = cleaned_text
+
+    new_filelist = filelist + "." + args.out_extension
+    with open(new_filelist, "w", encoding="utf-8") as f:
+      f.writelines(["|".join(x) + "\n" for x in filepaths_and_text])

requirements.txt

@@ -0,0 +1,13 @@
+Cython
+librosa
+matplotlib
+numpy
+scipy
+tensorboard
+torch
+torchvision
+unidecode
+pyopenjtalk
+protobuf
+tqdm
+gradio

text/LICENSE

@@ -0,0 +1,19 @@
+Copyright (c) 2017 Keith Ito
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in
+all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+THE SOFTWARE.

text/__init__.py

@@ -0,0 +1,56 @@
+""" from https://github.com/keithito/tacotron """
+from text import cleaners
+from text.symbols import symbols
+
+
+# Mappings from symbol to numeric ID and vice versa:
+_symbol_to_id = {s: i for i, s in enumerate(symbols)}
+_id_to_symbol = {i: s for i, s in enumerate(symbols)}
+
+
+def text_to_sequence(text, cleaner_names):
+  '''Converts a string of text to a sequence of IDs corresponding to the symbols in the text.
+    Args:
+      text: string to convert to a sequence
+      cleaner_names: names of the cleaner functions to run the text through
+    Returns:
+      List of integers corresponding to the symbols in the text
+  '''
+  sequence = []
+
+  clean_text = _clean_text(text, cleaner_names)
+  for symbol in clean_text:
+    if symbol not in _symbol_to_id.keys():
+      continue
+    symbol_id = _symbol_to_id[symbol]
+    sequence += [symbol_id]
+  return sequence
+
+
+def cleaned_text_to_sequence(cleaned_text):
+  '''Converts a string of text to a sequence of IDs corresponding to the symbols in the text.
+    Args:
+      text: string to convert to a sequence
+    Returns:
+      List of integers corresponding to the symbols in the text
+  '''
+  sequence = [_symbol_to_id[symbol] for symbol in cleaned_text if symbol in _symbol_to_id.keys()]
+  return sequence
+
+
+def sequence_to_text(sequence):
+  '''Converts a sequence of IDs back to a string'''
+  result = ''
+  for symbol_id in sequence:
+    s = _id_to_symbol[symbol_id]
+    result += s
+  return result
+
+
+def _clean_text(text, cleaner_names):
+  for name in cleaner_names:
+    cleaner = getattr(cleaners, name)
+    if not cleaner:
+      raise Exception('Unknown cleaner: %s' % name)
+    text = cleaner(text)
+  return text