init

Dockerfile

@@ -0,0 +1,22 @@
+FROM python:3.9.16-slim-bullseye
+
+RUN mkdir -p /app
+WORKDIR /app
+
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update && \
+    apt install build-essential -yq && \
+    apt-get clean && \
+    apt-get purge -y --auto-remove -o APT::AutoRemove::RecommendsImportant=false && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY requirements.txt /app
+RUN pip install -r requirements.txt
+
+COPY . /app
+
+EXPOSE 7860
+
+CMD [ "uvicorn" , "/app/app.py" , "--host" , "0.0.0.0" , "--port" , "7860" ]
+

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2022 CjangCjengh
+
+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.

Model/Nene_Nanami_Rong_Tang/1374_epochs.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edfb6b428c398fab83a85b5ae41e13cb5a9f7be12692129e8a880d4553701f7b
+size 158888013

Model/Nene_Nanami_Rong_Tang/config.json

@@ -0,0 +1,35 @@
+{
+  "train": {
+    "segment_size": 8192
+  },
+  "data": {
+    "text_cleaners":["zh_ja_mixture_cleaners"],
+    "max_wav_value": 32768.0,
+    "sampling_rate": 22050,
+    "filter_length": 1024,
+    "hop_length": 256,
+    "win_length": 1024,
+    "add_blank": true,
+    "n_speakers": 5
+  },
+  "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": ["\u7dbe\u5730\u5be7\u3005", "\u5728\u539f\u4e03\u6d77", "\u5c0f\u8338", "\u5510\u4e50\u541f"],
+  "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", " "]
+}

Model/Zero_no_tsukaima/1158_epochs.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da9a595f45f51dee33168c1e32bba38fe402fe3d8857331f5798a2407b6b2a86
+size 158902605

Model/Zero_no_tsukaima/config.json

@@ -0,0 +1,35 @@
+{
+  "train": {
+    "segment_size": 8192
+  },
+  "data": {
+    "text_cleaners":["japanese_cleaners2"],
+    "max_wav_value": 32768.0,
+    "sampling_rate": 22050,
+    "filter_length": 1024,
+    "hop_length": 256,
+    "win_length": 1024,
+    "add_blank": true,
+    "n_speakers": 26
+  },
+  "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": ["\u30eb\u30a4\u30ba", "\u30c6\u30a3\u30d5\u30a1\u30cb\u30a2", "\u30a4\u30eb\u30af\u30af\u30a5", "\u30a2\u30f3\u30ea\u30a8\u30c3\u30bf", "\u30bf\u30d0\u30b5", "\u30b7\u30a8\u30b9\u30bf", "\u30cf\u30eb\u30ca", "\u5c11\u5973\u30ea\u30b7\u30e5", "\u30ea\u30b7\u30e5", "\u30a2\u30ad\u30ca", "\u30af\u30ea\u30b9", "\u30ab\u30c8\u30ec\u30a2", "\u30a8\u30ec\u30aa\u30ce\u30fc\u30eb", "\u30e2\u30f3\u30e2\u30e9\u30f3\u30b7\u30fc", "\u30ea\u30fc\u30f4\u30eb", "\u30ad\u30e5\u30eb\u30b1", "\u30a6\u30a7\u30b6\u30ea\u30fc", "\u30b5\u30a4\u30c8", "\u30ae\u30fc\u30b7\u30e5", "\u30b3\u30eb\u30d9\u30fc\u30eb", "\u30aa\u30b9\u30de\u30f3", "\u30c7\u30eb\u30d5\u30ea\u30f3\u30ac\u30fc", "\u30c6\u30af\u30b9\u30c8", "\u30c0\u30f3\u30d7\u30ea\u30e1", "\u30ac\u30ec\u30c3\u30c8", "\u30b9\u30ab\u30ed\u30f3"],
+  "symbols": ["_", ",", ".", "!", "?", "-", "~", "\u2026", "A", "E", "I", "N", "O", "Q", "U", "a", "b", "d", "e", "f", "g", "h", "i", "j", "k", "m", "n", "o", "p", "r", "s", "t", "u", "v", "w", "y", "z", "\u0283", "\u02a7", "\u02a6", "\u2193", "\u2191", " "]
+}

Model/g/G_953000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c54396a7a9027952e4d72fceb7e1da1f003d108837138927c2054a33eda0292
+size 479276657

Model/g/config.json

@@ -0,0 +1,55 @@
+{
+  "train": {
+    "log_interval": 200,
+    "eval_interval": 1000,
+    "seed": 1234,
+    "epochs": 10000,
+    "learning_rate": 2e-4,
+    "betas": [0.8, 0.99],
+    "eps": 1e-9,
+    "batch_size": 64,
+    "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/uma_genshin_genshinjp_bh3_train.txt.cleaned",
+    "validation_files":"filelists/uma_genshin_genshinjp_bh3_val.txt.cleaned",
+    "text_cleaners":["zh_ja_mixture_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", " "]
+}

README.md

@@ -1,11 +1,231 @@
----
-title: MoeGoe Simple API
-emoji: 🌍
-colorFrom: yellow
-colorTo: purple
-sdk: docker
-pinned: false
-license: mit
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# MoeGoe-Simple-API
+
+
+
+MoeGoe-Simple-API 是一个易部署的api，可以通过api的方式调用语音合成，可用于聊天机器人等，目前支持的功能有语音合成和语音转换。
+
+- 可导入模型
+- 支持加载多模型，可以将多个模型合并为一个新的id对应角色模型的映射表
+<details><summary>点击预览返回的映射表</summary><pre><code>
+[
+	{
+		"0": "綾地寧々"
+	},
+	{
+		"1": "在原七海"
+	},
+	{
+		"2": "小茸"
+	},
+	{
+		"3": "唐乐吟"
+	},
+	{
+		"...": "..."
+	},
+	{
+		"196": "ルイズ"
+	},
+	{
+		"197": "ティファニア"
+	}
+]
+</code></pre></details>
+
+# 如何部署
+
+## Docker部署
+
+### docker镜像拉取脚本
+
+```
+bash -c "$(wget -O- https://gist.githubusercontent.com/Artrajz/b2c02499f91c3a51b8b48f1a3c9a7ead/raw/e3033f1b222868b4b0f1b522e52e18217460ff91/moegoe-simple-api-installer-latest.sh)"
+```
+
+镜像大小为5g，所以拉取会比较慢，拉取成功后由于没有导入vits模型所以无法使用，需要按以下步骤导入模型
+
+### 下载VITS模型
+
+VITS模型放入`/usr/local/moegoe-simple-api/Model`文件夹中，模型文件夹中要有.pth和config.json文件
+
+<details><summary>点击查看模型目录结构</summary><pre><code>
+├─Model
+│  ├─g
+│  │      config.json
+│  │      G_953000.pth
+│  │
+│  ├─Nene_Nanami_Rong_Tang
+│  │      1374_epochs.pth
+│  │      config.json
+│  │
+│  └─Zero_no_tsukaima
+│          1158_epochs.pth
+│          config.json
+</code></pre></details>
+
+### 修改模型路径
+
+在 `/usr/local/moegoe-simple-api/config.py` 中修改模型路径
+
+<details><summary>点击查看config.py模型路径填写示例</summary><pre><code>
+vits模型路径填写方法，MODEL_LIST中的每一行是
+[ABS_PATH+"/Model/{模型文件夹}/{.pth模型}", ABS_PATH+"/Model/{模型文件夹}/config.json"],
+也可以写相对路径或绝对路径，由于windows和linux路径写法不同，用上面的写法或绝对路径最稳妥
+示例：
+MODEL_LIST = [
+    [ABS_PATH+"/Model/Nene_Nanami_Rong_Tang/1374_epochs.pth", ABS_PATH+"/Model/Nene_Nanami_Rong_Tang/config.json"],
+    [ABS_PATH+"/Model/Zero_no_tsukaima/1158_epochs.pth", ABS_PATH+"/Model/Zero_no_tsukaima/config.json"],
+    [ABS_PATH+"/Model/g/G_953000.pth", ABS_PATH+"/Model/g/config.json"],
+]
+</code></pre></details>
+
+
+### 开始使用！
+
+终端输入`docker compose up -d`
+
+或再次执行拉取脚本
+
+### 镜像更新方法
+
+重新执行docker镜像拉取脚本即可
+
+
+## 直接部署
+
+1. 下载VITS模型并放入`Model`文件夹中
+2. 在 `config.py` 中修改模型路径
+3. 安装python依赖（建议用conda虚拟环境） `pip install -r requirements.txt`
+4. 开始使用！`python app.py`
+
+# 参数
+
+## 语音合成voice
+
+| Name          | Parameter | Is must | Default | Value        | Instruction                                      |
+| ------------- | --------- | ------- | ------- | ------------ | ------------------------------------------------ |
+| 合成文本      | text      | true    |         | text         |                                                  |
+| 角色id        | id        | false   | 0       | (number)     |                                                  |
+| 音频格式      | format    | false   | wav     | wav,ogg,silk | silk支持tx系语音                                 |
+| 文本语言      | lang      | false   | mix     | zh,ja,mix    | 当lang=mix时，文本应该用[ZH] 或 [JA] 包裹        |
+| 语音长度/语速 | speed     | false   | 1.0     | (number)     | 调节语音长度，相当于调节语速，该数值越大语速越慢 |
+
+## 语音转换voice conversion
+
+| Name       | Parameter   | Is must | Default | Value      | Instruction            |
+| ---------- | ----------- | ------- | ------- | ---------- | ---------------------- |
+| 上传音频   | upload      | true    |         | audio file | 只支持wav和ogg         |
+| 源角色id   | original_id | true    |         | (number)   | 上传文件所使用的角色id |
+| 目标角色id | target_id   | true    |         | (number)   | 要转换的目标角色id     |
+
+# 调用方法
+
+## GET
+
+- GET/POST http://127.0.0.1:23456/voice/speakers
+
+  返回id对应角色的映射表（json格式）
+
+- GET http://127.0.0.1/voice?text=[JA]text[JA][ZH]text[ZH]&id=0&format=wav&lang=mix
+
+  返回wav音频文件
+
+- GET http://127.0.0.1/voice?text=[JA]text[JA][ZH]text[ZH]&id=0&format=ogg&lang=mix
+
+  返回ogg音频文件
+
+- GET http://127.0.0.1/voice?text=text&lang=zh
+
+  设定语言为zh，则文本无需[ZH]包裹
+
+- GET http://127.0.0.1/voice?text=text&lang=ja
+
+  设定语言为ja，则文本无需[JA]包裹
+  
+- GET http://127.0.0.1/voice?text=text&id=142&format=wav&lang=zh&speed=1.4
+
+  文本为text，角色id为142，音频格式为wav，文本语言为zh，语速为1.4
+
+## POST
+
+- python
+
+```python
+import re
+import requests
+import json
+import os
+import random
+import string
+from requests_toolbelt.multipart.encoder import MultipartEncoder
+
+abs_path = os.path.dirname(__file__)
+
+def voice_speakers():
+    url = "http://127.0.0.1:23456/voice/speakers"
+
+    res = requests.post(url=url)
+    json = res.json()
+    for i in json:
+        print(i)
+    
+def voice():
+    post_json = json.dumps({
+        "text":"需要合成的文本",
+        "id":142,
+        "format":"wav",
+        "lang":"zh",
+        "speed":1.4,
+        })
+    headers={'content-type':'application/json'}
+    url = "http://127.0.0.1:23456/voice"
+
+    res = requests.post(url=url,data=post_json,headers=headers)
+    fname = re.findall("filename=(.+)", res.headers["Content-Disposition"])[0]
+    path = f"{abs_path}/{fname}"
+    with open(path, "wb") as f:
+        f.write(res.content)
+
+def voice_conversion(upload_name):
+    upload_path = f'{abs_path}/{upload_name}'
+    upload_type = f'audio/{upload_name.split(".")[1]}' #wav,ogg
+    
+    fields = {
+        "upload": (upload_name, open(upload_path,'rb'),upload_type),
+        "original_id": "142",
+        "target_id": "92",
+    }
+    boundary = '----VoiceConversionFormBoundary' \
+               + ''.join(random.sample(string.ascii_letters + string.digits, 16))
+    m = MultipartEncoder(fields=fields, boundary=boundary)
+    
+    headers = {"Content-Type": m.content_type}
+    url = "http://127.0.0.1:23456/voice/conversion"
+
+    res = requests.post(url=url,data=m,headers=headers)
+    fname = re.findall("filename=(.+)", res.headers["Content-Disposition"])[0]
+    path = f"{abs_path}/{fname}"
+    with open(path, "wb") as f:
+        f.write(res.content)
+```
+
+# 可能遇到的问题
+
+~~本人遇到过的问题~~
+
+### 运行后服务器无响应
+
+可能是内存不足，可以尝试减少模型加载数量。
+
+### 模型推理时服务器无响应
+
+可能是同时处理多个推理任务导致CPU堵塞，可以尝试在*voice.py*中取消以下**两行**代码的注释，意思是让pytorch只使用1个物理CPU核心，防止一个任务抢占过多CPU资源。
+
+```python
+import torch
+torch.set_num_threads(1)
+```
+
+# 鸣谢
+
+ 该项目基于[CjangCjengh](https://github.com/CjangCjengh)的[MoeGoe](https://github.com/CjangCjengh/MoeGoe)

app.py

@@ -0,0 +1,115 @@
+import os
+
+import logging
+import uuid
+
+from flask import Flask, request, send_file, jsonify
+from werkzeug.utils import secure_filename
+from flask_apscheduler import APScheduler
+
+from utils import clean_folder, merge_model
+
+app = Flask(__name__)
+app.config.from_pyfile("config.py")
+
+scheduler = APScheduler()
+scheduler.init_app(app)
+scheduler.start()
+
+logging.getLogger('numba').setLevel(logging.WARNING)
+
+voice_obj, voice_speakers = merge_model(app.config["MODEL_LIST"])
+
+if not os.path.exists(app.config['UPLOAD_FOLDER']):
+    try:
+        os.mkdir(app.config['UPLOAD_FOLDER'])
+    except:
+        pass
+
+
+@app.route('/')
+@app.route('/voice/')
+def index():
+    return "usage:https://github.com/Artrajz/MoeGoe-Simple-API#readme"
+
+
+@app.route('/voice/speakers', methods=["GET", "POST"])
+def voice_speakers_api():
+    speakers_list = voice_speakers
+    return jsonify(speakers_list)
+
+
+@app.route('/voice', methods=["GET", "POST"])
+def voice_api():
+    if request.method == "GET":
+        text = request.args.get("text")
+        speaker_id = int(request.args.get("id", 0))
+        format = request.args.get("format", "wav")
+        lang = request.args.get("lang", "mix")
+        speed = float(request.args.get("speed", 1.0))
+    elif request.method == "POST":
+        json_data = request.json
+        text = json_data["text"]
+        speaker_id = int(json_data["id"])
+        format = json_data["format"]
+        lang = json_data["lang"]
+        speed = float(json_data["speed"])
+
+    if lang.upper() == "ZH":
+        text = f"[ZH]{text}[ZH]"
+    elif lang.upper() == "JA":
+        text = f"[JA]{text}[JA]"
+
+    real_id = voice_obj[speaker_id][0]
+    real_obj = voice_obj[speaker_id][1]
+
+    output, file_type, fname = real_obj.generate(text, real_id, format, speed)
+
+    return send_file(path_or_file=output, mimetype=file_type, download_name=fname)
+
+
+@app.route('/voice/conversion', methods=["GET", "POST"])
+def voice_conversion_api():
+    if request.method == "GET":
+        return jsonify("method should be POST")
+    if request.method == "POST":
+        # json_data = request.json
+        voice = request.files['upload']
+        original_id = int(request.form["original_id"])
+        target_id = int(request.form["target_id"])
+
+        form = {}
+
+        format = voice.filename.split(".")[1]
+
+        fname = secure_filename(str(uuid.uuid1()) + "." + voice.filename.split(".")[1])
+        voice.save(os.path.join(app.config['UPLOAD_FOLDER'], fname))
+
+        real_original_id = int(voice_obj[original_id][0])
+        real_target_id = int(voice_obj[target_id][0])
+        real_obj = voice_obj[original_id][1]
+        real_target_obj = voice_obj[target_id][1]
+
+        if voice_obj[original_id][2] != voice_obj[target_id][2]:
+            form["status"] = "error"
+            form["message"] = "speaker IDs are in diffrent Model!"
+            return form
+
+        output = real_obj.voice_conversion(os.path.join(app.config['UPLOAD_FOLDER'], fname),
+                                           real_original_id, real_target_id, format)
+        file_type = f"audio/{format}"
+
+        return send_file(path_or_file=output, mimetype=file_type, download_name=fname)
+        # return output
+
+
+# 定时清理临时文件，每小时清一次
+@scheduler.task('interval', id='随便写', seconds=3600, misfire_grace_time=900)
+def clean_task():
+    clean_folder(app.config["UPLOAD_FOLDER"])
+    clean_folder(app.config["SILK_OUT_PATH"])
+
+
+if __name__ == '__main__':
+    app.run(host='0.0.0.0', port=app.config["PORT"])  # 如果对外开放用这个,docker部署也用这个
+    # app.run(host='127.0.0.1', port=app.config["PORT"], debug=True)  # 本地运行、调试

attentions.py

@@ -0,0 +1,300 @@
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+import commons
+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 +1,96 @@
+import torch
+from torch.nn import functional as F
+import torch.jit
+
+
+def script_method(fn, _rcb=None):
+  return fn
+
+
+def script(obj, optimize=True, _frames_up=0, _rcb=None):
+  return obj
+
+
+torch.jit.script_method = script_method
+torch.jit.script = script
+
+
+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 intersperse(lst, item):
+  result = [item] * (len(lst) * 2 + 1)
+  result[1::2] = lst
+  return result
+
+
+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 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 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

config.py

@@ -0,0 +1,32 @@
+import os
+import sys
+
+JSON_AS_ASCII = False
+MAX_CONTENT_LENGTH = 5242880
+
+# 端口
+PORT = 7860
+# 项目的绝对路径
+ABS_PATH = os.path.join(os.path.dirname(os.path.realpath(sys.argv[0])))
+# 上传文件的临时路径，非必要不要动
+UPLOAD_FOLDER = ABS_PATH + "/upload"
+# silk文件输出的临时路径，非必要不要动
+SILK_OUT_PATH = ABS_PATH + "/out_silk"
+
+'''
+vits模型路径填写方法，MODEL_LIST中的每一行是
+[ABS_PATH+"/Model/{模型文件夹}/{.pth模型}", ABS_PATH+"/Model/{模型文件夹}/config.json"],
+也可以写相对路径或绝对路径，由于windows和linux路径写法不同，用上面的写法或绝对路径最稳妥
+示例：
+MODEL_LIST = [
+    [ABS_PATH+"/Model/Nene_Nanami_Rong_Tang/1374_epochs.pth", ABS_PATH+"/Model/Nene_Nanami_Rong_Tang/config.json"],
+    [ABS_PATH+"/Model/Zero_no_tsukaima/1158_epochs.pth", ABS_PATH+"/Model/Zero_no_tsukaima/config.json"],
+    [ABS_PATH+"/Model/g/G_953000.pth", ABS_PATH+"/Model/g/config.json"],
+]
+'''
+# 模型加载列表
+MODEL_LIST = [
+    [ABS_PATH+"/Model/Nene_Nanami_Rong_Tang/1374_epochs.pth", ABS_PATH+"/Model/Nene_Nanami_Rong_Tang/config.json"],
+    [ABS_PATH+"/Model/Zero_no_tsukaima/1158_epochs.pth", ABS_PATH+"/Model/Zero_no_tsukaima/config.json"],
+    [ABS_PATH+"/Model/g/G_953000.pth", ABS_PATH+"/Model/g/config.json"],
+]

docker-compose.yaml

@@ -0,0 +1,12 @@
+version: '3.4'
+services:
+  moegoe:
+    image: artrajz/moegoe-simple-api:latest
+    restart: always
+    ports:
+      - 23456:23456
+    environment:
+      LANG: 'C.UTF-8'
+    volumes:
+      - ./Model:/app/Model # 挂载模型文件夹
+      - ./config.py:/app/config.py # 挂载配置文件

hubert_model.py

@@ -0,0 +1,221 @@
+import copy
+from typing import Optional, Tuple
+import random
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.nn.modules.utils import consume_prefix_in_state_dict_if_present
+
+class Hubert(nn.Module):
+    def __init__(self, num_label_embeddings: int = 100, mask: bool = True):
+        super().__init__()
+        self._mask = mask
+        self.feature_extractor = FeatureExtractor()
+        self.feature_projection = FeatureProjection()
+        self.positional_embedding = PositionalConvEmbedding()
+        self.norm = nn.LayerNorm(768)
+        self.dropout = nn.Dropout(0.1)
+        self.encoder = TransformerEncoder(
+            nn.TransformerEncoderLayer(
+                768, 12, 3072, activation="gelu", batch_first=True
+            ),
+            12,
+        )
+        self.proj = nn.Linear(768, 256)
+
+        self.masked_spec_embed = nn.Parameter(torch.FloatTensor(768).uniform_())
+        self.label_embedding = nn.Embedding(num_label_embeddings, 256)
+
+    def mask(self, x: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        mask = None
+        if self.training and self._mask:
+            mask = _compute_mask((x.size(0), x.size(1)), 0.8, 10, x.device, 2)
+            x[mask] = self.masked_spec_embed.to(x.dtype)
+        return x, mask
+
+    def encode(
+        self, x: torch.Tensor, layer: Optional[int] = None
+    ) -> Tuple[torch.Tensor, torch.Tensor]:
+        x = self.feature_extractor(x)
+        x = self.feature_projection(x.transpose(1, 2))
+        x, mask = self.mask(x)
+        x = x + self.positional_embedding(x)
+        x = self.dropout(self.norm(x))
+        x = self.encoder(x, output_layer=layer)
+        return x, mask
+
+    def logits(self, x: torch.Tensor) -> torch.Tensor:
+        logits = torch.cosine_similarity(
+            x.unsqueeze(2),
+            self.label_embedding.weight.unsqueeze(0).unsqueeze(0),
+            dim=-1,
+        )
+        return logits / 0.1
+
+    def forward(self, x: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        x, mask = self.encode(x)
+        x = self.proj(x)
+        logits = self.logits(x)
+        return logits, mask
+
+
+class HubertSoft(Hubert):
+    def __init__(self):
+        super().__init__()
+
+    @torch.inference_mode()
+    def units(self, wav: torch.Tensor) -> torch.Tensor:
+        wav = F.pad(wav, ((400 - 320) // 2, (400 - 320) // 2))
+        x, _ = self.encode(wav)
+        return self.proj(x)
+
+
+class FeatureExtractor(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.conv0 = nn.Conv1d(1, 512, 10, 5, bias=False)
+        self.norm0 = nn.GroupNorm(512, 512)
+        self.conv1 = nn.Conv1d(512, 512, 3, 2, bias=False)
+        self.conv2 = nn.Conv1d(512, 512, 3, 2, bias=False)
+        self.conv3 = nn.Conv1d(512, 512, 3, 2, bias=False)
+        self.conv4 = nn.Conv1d(512, 512, 3, 2, bias=False)
+        self.conv5 = nn.Conv1d(512, 512, 2, 2, bias=False)
+        self.conv6 = nn.Conv1d(512, 512, 2, 2, bias=False)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = F.gelu(self.norm0(self.conv0(x)))
+        x = F.gelu(self.conv1(x))
+        x = F.gelu(self.conv2(x))
+        x = F.gelu(self.conv3(x))
+        x = F.gelu(self.conv4(x))
+        x = F.gelu(self.conv5(x))
+        x = F.gelu(self.conv6(x))
+        return x
+
+
+class FeatureProjection(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.norm = nn.LayerNorm(512)
+        self.projection = nn.Linear(512, 768)
+        self.dropout = nn.Dropout(0.1)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = self.norm(x)
+        x = self.projection(x)
+        x = self.dropout(x)
+        return x
+
+
+class PositionalConvEmbedding(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.conv = nn.Conv1d(
+            768,
+            768,
+            kernel_size=128,
+            padding=128 // 2,
+            groups=16,
+        )
+        self.conv = nn.utils.weight_norm(self.conv, name="weight", dim=2)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = self.conv(x.transpose(1, 2))
+        x = F.gelu(x[:, :, :-1])
+        return x.transpose(1, 2)
+
+
+class TransformerEncoder(nn.Module):
+    def __init__(
+        self, encoder_layer: nn.TransformerEncoderLayer, num_layers: int
+    ) -> None:
+        super(TransformerEncoder, self).__init__()
+        self.layers = nn.ModuleList(
+            [copy.deepcopy(encoder_layer) for _ in range(num_layers)]
+        )
+        self.num_layers = num_layers
+
+    def forward(
+        self,
+        src: torch.Tensor,
+        mask: torch.Tensor = None,
+        src_key_padding_mask: torch.Tensor = None,
+        output_layer: Optional[int] = None,
+    ) -> torch.Tensor:
+        output = src
+        for layer in self.layers[:output_layer]:
+            output = layer(
+                output, src_mask=mask, src_key_padding_mask=src_key_padding_mask
+            )
+        return output
+
+
+def _compute_mask(
+    shape: Tuple[int, int],
+    mask_prob: float,
+    mask_length: int,
+    device: torch.device,
+    min_masks: int = 0,
+) -> torch.Tensor:
+    batch_size, sequence_length = shape
+
+    if mask_length < 1:
+        raise ValueError("`mask_length` has to be bigger than 0.")
+
+    if mask_length > sequence_length:
+        raise ValueError(
+            f"`mask_length` has to be smaller than `sequence_length`, but got `mask_length`: {mask_length} and `sequence_length`: {sequence_length}`"
+        )
+
+    # compute number of masked spans in batch
+    num_masked_spans = int(mask_prob * sequence_length / mask_length + random.random())
+    num_masked_spans = max(num_masked_spans, min_masks)
+
+    # make sure num masked indices <= sequence_length
+    if num_masked_spans * mask_length > sequence_length:
+        num_masked_spans = sequence_length // mask_length
+
+    # SpecAugment mask to fill
+    mask = torch.zeros((batch_size, sequence_length), device=device, dtype=torch.bool)
+
+    # uniform distribution to sample from, make sure that offset samples are < sequence_length
+    uniform_dist = torch.ones(
+        (batch_size, sequence_length - (mask_length - 1)), device=device
+    )
+
+    # get random indices to mask
+    mask_indices = torch.multinomial(uniform_dist, num_masked_spans)
+
+    # expand masked indices to masked spans
+    mask_indices = (
+        mask_indices.unsqueeze(dim=-1)
+        .expand((batch_size, num_masked_spans, mask_length))
+        .reshape(batch_size, num_masked_spans * mask_length)
+    )
+    offsets = (
+        torch.arange(mask_length, device=device)[None, None, :]
+        .expand((batch_size, num_masked_spans, mask_length))
+        .reshape(batch_size, num_masked_spans * mask_length)
+    )
+    mask_idxs = mask_indices + offsets
+
+    # scatter indices to mask
+    mask = mask.scatter(1, mask_idxs, True)
+
+    return mask
+
+
+def hubert_soft(
+    path: str
+) -> HubertSoft:
+    r"""HuBERT-Soft from `"A Comparison of Discrete and Soft Speech Units for Improved Voice Conversion"`.
+    Args:
+        path (str): path of a pretrained model
+    """
+    hubert = HubertSoft()
+    checkpoint = torch.load(path)
+    consume_prefix_in_state_dict_if_present(checkpoint, "module.")
+    hubert.load_state_dict(checkpoint)
+    hubert.eval()
+    return hubert

mel_processing.py

@@ -0,0 +1,101 @@
+import torch
+import torch.utils.data
+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, return_complex=False)
+
+    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 +1,404 @@
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+import commons
+import modules
+import attentions
+
+from torch.nn import Conv1d, ConvTranspose1d
+from torch.nn.utils import weight_norm
+from commons import init_weights
+
+
+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,
+      emotion_embedding):
+    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.emotion_embedding = emotion_embedding
+    
+    if self.n_vocab!=0:
+      self.emb = nn.Embedding(n_vocab, hidden_channels)
+      if emotion_embedding:
+        self.emo_proj = nn.Linear(1024, 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, emotion_embedding=None):
+    if self.n_vocab!=0:
+      x = self.emb(x) * math.sqrt(self.hidden_channels) # [b, t, h]
+    if emotion_embedding is not None:
+      x = x + self.emo_proj(emotion_embedding.unsqueeze(1))
+    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
+
+
+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,
+    emotion_embedding=False,
+    **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,
+        emotion_embedding)
+    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 infer(self, x, x_lengths, sid=None, noise_scale=1, length_scale=1, noise_scale_w=1., max_len=None, emotion_embedding=None):
+    x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths, emotion_embedding)
+    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)
+

modules.py

@@ -0,0 +1,387 @@
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+from torch.nn import Conv1d
+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):
+  """
+  Dilated 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

requirements.txt

@@ -0,0 +1,23 @@
+numba
+librosa
+numpy==1.23.3
+scipy
+torch
+unidecode
+openjtalk>=0.3.0.dev2
+jamo
+pypinyin
+jieba
+protobuf
+cn2an
+inflect
+eng_to_ipa
+ko_pron
+indic_transliteration
+num_thai
+opencc
+audonnx
+flask
+av
+pilk
+flask_apscheduler

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 +1,32 @@
+""" from https://github.com/keithito/tacotron """
+from text import cleaners
+
+
+def text_to_sequence(text, symbols, 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
+  '''
+  _symbol_to_id = {s: i for i, s in enumerate(symbols)}
+
+  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 _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

text/cantonese.py

@@ -0,0 +1,59 @@
+import re
+import cn2an
+import opencc
+
+
+converter = opencc.OpenCC('jyutjyu')
+
+# List of (Latin alphabet, ipa) pairs:
+_latin_to_ipa = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('A', 'ei˥'),
+    ('B', 'biː˥'),
+    ('C', 'siː˥'),
+    ('D', 'tiː˥'),
+    ('E', 'iː˥'),
+    ('F', 'e˥fuː˨˩'),
+    ('G', 'tsiː˥'),
+    ('H', 'ɪk̚˥tsʰyː˨˩'),
+    ('I', 'ɐi˥'),
+    ('J', 'tsei˥'),
+    ('K', 'kʰei˥'),
+    ('L', 'e˥llou˨˩'),
+    ('M', 'ɛːm˥'),
+    ('N', 'ɛːn˥'),
+    ('O', 'ou˥'),
+    ('P', 'pʰiː˥'),
+    ('Q', 'kʰiːu˥'),
+    ('R', 'aː˥lou˨˩'),
+    ('S', 'ɛː˥siː˨˩'),
+    ('T', 'tʰiː˥'),
+    ('U', 'juː˥'),
+    ('V', 'wiː˥'),
+    ('W', 'tʊk̚˥piː˥juː˥'),
+    ('X', 'ɪk̚˥siː˨˩'),
+    ('Y', 'waːi˥'),
+    ('Z', 'iː˨sɛːt̚˥')
+]]
+
+
+def number_to_cantonese(text):
+    return re.sub(r'\d+(?:\.?\d+)?', lambda x: cn2an.an2cn(x.group()), text)
+
+
+def latin_to_ipa(text):
+    for regex, replacement in _latin_to_ipa:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def cantonese_to_ipa(text):
+    text = number_to_cantonese(text.upper())
+    text = converter.convert(text).replace('-','').replace('$',' ')
+    text = re.sub(r'[A-Z]', lambda x: latin_to_ipa(x.group())+' ', text)
+    text = re.sub(r'[、；：]', '，', text)
+    text = re.sub(r'\s*，\s*', ', ', text)
+    text = re.sub(r'\s*。\s*', '. ', text)
+    text = re.sub(r'\s*？\s*', '? ', text)
+    text = re.sub(r'\s*！\s*', '! ', text)
+    text = re.sub(r'\s*$', '', text)
+    return text

text/cleaners.py

@@ -0,0 +1,145 @@
+import re
+
+
+def japanese_cleaners(text):
+    from text.japanese import japanese_to_romaji_with_accent
+    text = japanese_to_romaji_with_accent(text)
+    text = re.sub(r'([A-Za-z])$', r'\1.', text)
+    return text
+
+
+def japanese_cleaners2(text):
+    return japanese_cleaners(text).replace('ts', 'ʦ').replace('...', '…')
+
+
+def korean_cleaners(text):
+    '''Pipeline for Korean text'''
+    from text.korean import latin_to_hangul, number_to_hangul, divide_hangul
+    text = latin_to_hangul(text)
+    text = number_to_hangul(text)
+    text = divide_hangul(text)
+    text = re.sub(r'([\u3131-\u3163])$', r'\1.', text)
+    return text
+
+
+def chinese_cleaners(text):
+    '''Pipeline for Chinese text'''
+    from text.mandarin import number_to_chinese, chinese_to_bopomofo, latin_to_bopomofo
+    text = number_to_chinese(text)
+    text = chinese_to_bopomofo(text)
+    text = latin_to_bopomofo(text)
+    text = re.sub(r'([ˉˊˇˋ˙])$', r'\1。', text)
+    return text
+
+
+def zh_ja_mixture_cleaners(text):
+    from text.mandarin import chinese_to_romaji
+    from text.japanese import japanese_to_romaji_with_accent
+    text = re.sub(r'\[ZH\](.*?)\[ZH\]',
+                  lambda x: chinese_to_romaji(x.group(1))+' ', text)
+    text = re.sub(r'\[JA\](.*?)\[JA\]', lambda x: japanese_to_romaji_with_accent(
+        x.group(1)).replace('ts', 'ʦ').replace('u', 'ɯ').replace('...', '…')+' ', text)
+    text = re.sub(r'\s+$', '', text)
+    text = re.sub(r'([^\.,!\?\-…~])$', r'\1.', text)
+    return text
+
+
+def sanskrit_cleaners(text):
+    text = text.replace('॥', '।').replace('ॐ', 'ओम्')
+    text = re.sub(r'([^।])$', r'\1।', text)
+    return text
+
+
+def cjks_cleaners(text):
+    from text.mandarin import chinese_to_lazy_ipa
+    from text.japanese import japanese_to_ipa
+    from text.korean import korean_to_lazy_ipa
+    from text.sanskrit import devanagari_to_ipa
+    from text.english import english_to_lazy_ipa
+    text = re.sub(r'\[ZH\](.*?)\[ZH\]',
+                  lambda x: chinese_to_lazy_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[JA\](.*?)\[JA\]',
+                  lambda x: japanese_to_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[KO\](.*?)\[KO\]',
+                  lambda x: korean_to_lazy_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[SA\](.*?)\[SA\]',
+                  lambda x: devanagari_to_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[EN\](.*?)\[EN\]',
+                  lambda x: english_to_lazy_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\s+$', '', text)
+    text = re.sub(r'([^\.,!\?\-…~])$', r'\1.', text)
+    return text
+
+
+def cjke_cleaners(text):
+    from text.mandarin import chinese_to_lazy_ipa
+    from text.japanese import japanese_to_ipa
+    from text.korean import korean_to_ipa
+    from text.english import english_to_ipa2
+    text = re.sub(r'\[ZH\](.*?)\[ZH\]', lambda x: chinese_to_lazy_ipa(x.group(1)).replace(
+        'ʧ', 'tʃ').replace('ʦ', 'ts').replace('ɥan', 'ɥæn')+' ', text)
+    text = re.sub(r'\[JA\](.*?)\[JA\]', lambda x: japanese_to_ipa(x.group(1)).replace('ʧ', 'tʃ').replace(
+        'ʦ', 'ts').replace('ɥan', 'ɥæn').replace('ʥ', 'dz')+' ', text)
+    text = re.sub(r'\[KO\](.*?)\[KO\]',
+                  lambda x: korean_to_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[EN\](.*?)\[EN\]', lambda x: english_to_ipa2(x.group(1)).replace('ɑ', 'a').replace(
+        'ɔ', 'o').replace('ɛ', 'e').replace('ɪ', 'i').replace('ʊ', 'u')+' ', text)
+    text = re.sub(r'\s+$', '', text)
+    text = re.sub(r'([^\.,!\?\-…~])$', r'\1.', text)
+    return text
+
+
+def cjke_cleaners2(text):
+    from text.mandarin import chinese_to_ipa
+    from text.japanese import japanese_to_ipa2
+    from text.korean import korean_to_ipa
+    from text.english import english_to_ipa2
+    text = re.sub(r'\[ZH\](.*?)\[ZH\]',
+                  lambda x: chinese_to_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[JA\](.*?)\[JA\]',
+                  lambda x: japanese_to_ipa2(x.group(1))+' ', text)
+    text = re.sub(r'\[KO\](.*?)\[KO\]',
+                  lambda x: korean_to_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[EN\](.*?)\[EN\]',
+                  lambda x: english_to_ipa2(x.group(1))+' ', text)
+    text = re.sub(r'\s+$', '', text)
+    text = re.sub(r'([^\.,!\?\-…~])$', r'\1.', text)
+    return text
+
+
+def thai_cleaners(text):
+    from text.thai import num_to_thai, latin_to_thai
+    text = num_to_thai(text)
+    text = latin_to_thai(text)
+    return text
+
+
+def shanghainese_cleaners(text):
+    from text.shanghainese import shanghainese_to_ipa
+    text = shanghainese_to_ipa(text)
+    text = re.sub(r'([^\.,!\?\-…~])$', r'\1.', text)
+    return text
+
+
+def chinese_dialect_cleaners(text):
+    from text.mandarin import chinese_to_ipa2
+    from text.japanese import japanese_to_ipa3
+    from text.shanghainese import shanghainese_to_ipa
+    from text.cantonese import cantonese_to_ipa
+    from text.english import english_to_lazy_ipa2
+    from text.ngu_dialect import ngu_dialect_to_ipa
+    text = re.sub(r'\[ZH\](.*?)\[ZH\]',
+                  lambda x: chinese_to_ipa2(x.group(1))+' ', text)
+    text = re.sub(r'\[JA\](.*?)\[JA\]',
+                  lambda x: japanese_to_ipa3(x.group(1)).replace('Q', 'ʔ')+' ', text)
+    text = re.sub(r'\[SH\](.*?)\[SH\]', lambda x: shanghainese_to_ipa(x.group(1)).replace('1', '˥˧').replace('5',
+                  '˧˧˦').replace('6', '˩˩˧').replace('7', '˥').replace('8', '˩˨').replace('ᴀ', 'ɐ').replace('ᴇ', 'e')+' ', text)
+    text = re.sub(r'\[GD\](.*?)\[GD\]',
+                  lambda x: cantonese_to_ipa(x.group(1))+' ', text)
+    text = re.sub(r'\[EN\](.*?)\[EN\]',
+                  lambda x: english_to_lazy_ipa2(x.group(1))+' ', text)
+    text = re.sub(r'\[([A-Z]{2})\](.*?)\[\1\]', lambda x: ngu_dialect_to_ipa(x.group(2), x.group(
+        1)).replace('ʣ', 'dz').replace('ʥ', 'dʑ').replace('ʦ', 'ts').replace('ʨ', 'tɕ')+' ', text)
+    text = re.sub(r'\s+$', '', text)
+    text = re.sub(r'([^\.,!\?\-…~])$', r'\1.', text)
+    return text

text/english.py

@@ -0,0 +1,188 @@
+""" from https://github.com/keithito/tacotron """
+
+'''
+Cleaners are transformations that run over the input text at both training and eval time.
+
+Cleaners can be selected by passing a comma-delimited list of cleaner names as the "cleaners"
+hyperparameter. Some cleaners are English-specific. You'll typically want to use:
+  1. "english_cleaners" for English text
+  2. "transliteration_cleaners" for non-English text that can be transliterated to ASCII using
+     the Unidecode library (https://pypi.python.org/pypi/Unidecode)
+  3. "basic_cleaners" if you do not want to transliterate (in this case, you should also update
+     the symbols in symbols.py to match your data).
+'''
+
+
+# Regular expression matching whitespace:
+
+
+import re
+import inflect
+from unidecode import unidecode
+import eng_to_ipa as ipa
+_inflect = inflect.engine()
+_comma_number_re = re.compile(r'([0-9][0-9\,]+[0-9])')
+_decimal_number_re = re.compile(r'([0-9]+\.[0-9]+)')
+_pounds_re = re.compile(r'£([0-9\,]*[0-9]+)')
+_dollars_re = re.compile(r'\$([0-9\.\,]*[0-9]+)')
+_ordinal_re = re.compile(r'[0-9]+(st|nd|rd|th)')
+_number_re = re.compile(r'[0-9]+')
+
+# List of (regular expression, replacement) pairs for abbreviations:
+_abbreviations = [(re.compile('\\b%s\\.' % x[0], re.IGNORECASE), x[1]) for x in [
+    ('mrs', 'misess'),
+    ('mr', 'mister'),
+    ('dr', 'doctor'),
+    ('st', 'saint'),
+    ('co', 'company'),
+    ('jr', 'junior'),
+    ('maj', 'major'),
+    ('gen', 'general'),
+    ('drs', 'doctors'),
+    ('rev', 'reverend'),
+    ('lt', 'lieutenant'),
+    ('hon', 'honorable'),
+    ('sgt', 'sergeant'),
+    ('capt', 'captain'),
+    ('esq', 'esquire'),
+    ('ltd', 'limited'),
+    ('col', 'colonel'),
+    ('ft', 'fort'),
+]]
+
+
+# List of (ipa, lazy ipa) pairs:
+_lazy_ipa = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('r', 'ɹ'),
+    ('æ', 'e'),
+    ('ɑ', 'a'),
+    ('ɔ', 'o'),
+    ('ð', 'z'),
+    ('θ', 's'),
+    ('ɛ', 'e'),
+    ('ɪ', 'i'),
+    ('ʊ', 'u'),
+    ('ʒ', 'ʥ'),
+    ('ʤ', 'ʥ'),
+    ('ˈ', '↓'),
+]]
+
+# List of (ipa, lazy ipa2) pairs:
+_lazy_ipa2 = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('r', 'ɹ'),
+    ('ð', 'z'),
+    ('θ', 's'),
+    ('ʒ', 'ʑ'),
+    ('ʤ', 'dʑ'),
+    ('ˈ', '↓'),
+]]
+
+# List of (ipa, ipa2) pairs
+_ipa_to_ipa2 = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('r', 'ɹ'),
+    ('ʤ', 'dʒ'),
+    ('ʧ', 'tʃ')
+]]
+
+
+def expand_abbreviations(text):
+    for regex, replacement in _abbreviations:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def collapse_whitespace(text):
+    return re.sub(r'\s+', ' ', text)
+
+
+def _remove_commas(m):
+    return m.group(1).replace(',', '')
+
+
+def _expand_decimal_point(m):
+    return m.group(1).replace('.', ' point ')
+
+
+def _expand_dollars(m):
+    match = m.group(1)
+    parts = match.split('.')
+    if len(parts) > 2:
+        return match + ' dollars'  # Unexpected format
+    dollars = int(parts[0]) if parts[0] else 0
+    cents = int(parts[1]) if len(parts) > 1 and parts[1] else 0
+    if dollars and cents:
+        dollar_unit = 'dollar' if dollars == 1 else 'dollars'
+        cent_unit = 'cent' if cents == 1 else 'cents'
+        return '%s %s, %s %s' % (dollars, dollar_unit, cents, cent_unit)
+    elif dollars:
+        dollar_unit = 'dollar' if dollars == 1 else 'dollars'
+        return '%s %s' % (dollars, dollar_unit)
+    elif cents:
+        cent_unit = 'cent' if cents == 1 else 'cents'
+        return '%s %s' % (cents, cent_unit)
+    else:
+        return 'zero dollars'
+
+
+def _expand_ordinal(m):
+    return _inflect.number_to_words(m.group(0))
+
+
+def _expand_number(m):
+    num = int(m.group(0))
+    if num > 1000 and num < 3000:
+        if num == 2000:
+            return 'two thousand'
+        elif num > 2000 and num < 2010:
+            return 'two thousand ' + _inflect.number_to_words(num % 100)
+        elif num % 100 == 0:
+            return _inflect.number_to_words(num // 100) + ' hundred'
+        else:
+            return _inflect.number_to_words(num, andword='', zero='oh', group=2).replace(', ', ' ')
+    else:
+        return _inflect.number_to_words(num, andword='')
+
+
+def normalize_numbers(text):
+    text = re.sub(_comma_number_re, _remove_commas, text)
+    text = re.sub(_pounds_re, r'\1 pounds', text)
+    text = re.sub(_dollars_re, _expand_dollars, text)
+    text = re.sub(_decimal_number_re, _expand_decimal_point, text)
+    text = re.sub(_ordinal_re, _expand_ordinal, text)
+    text = re.sub(_number_re, _expand_number, text)
+    return text
+
+
+def mark_dark_l(text):
+    return re.sub(r'l([^aeiouæɑɔəɛɪʊ ]*(?: |$))', lambda x: 'ɫ'+x.group(1), text)
+
+
+def english_to_ipa(text):
+    text = unidecode(text).lower()
+    text = expand_abbreviations(text)
+    text = normalize_numbers(text)
+    phonemes = ipa.convert(text)
+    phonemes = collapse_whitespace(phonemes)
+    return phonemes
+
+
+def english_to_lazy_ipa(text):
+    text = english_to_ipa(text)
+    for regex, replacement in _lazy_ipa:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def english_to_ipa2(text):
+    text = english_to_ipa(text)
+    text = mark_dark_l(text)
+    for regex, replacement in _ipa_to_ipa2:
+        text = re.sub(regex, replacement, text)
+    return text.replace('...', '…')
+
+
+def english_to_lazy_ipa2(text):
+    text = english_to_ipa(text)
+    for regex, replacement in _lazy_ipa2:
+        text = re.sub(regex, replacement, text)
+    return text

text/japanese.py

@@ -0,0 +1,153 @@
+import re
+from unidecode import unidecode
+import pyopenjtalk
+
+
+# Regular expression matching Japanese without punctuation marks:
+_japanese_characters = re.compile(
+    r'[A-Za-z\d\u3005\u3040-\u30ff\u4e00-\u9fff\uff11-\uff19\uff21-\uff3a\uff41-\uff5a\uff66-\uff9d]')
+
+# Regular expression matching non-Japanese characters or punctuation marks:
+_japanese_marks = re.compile(
+    r'[^A-Za-z\d\u3005\u3040-\u30ff\u4e00-\u9fff\uff11-\uff19\uff21-\uff3a\uff41-\uff5a\uff66-\uff9d]')
+
+# List of (symbol, Japanese) pairs for marks:
+_symbols_to_japanese = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('％', 'パーセント')
+]]
+
+# List of (romaji, ipa) pairs for marks:
+_romaji_to_ipa = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('ts', 'ʦ'),
+    ('u', 'ɯ'),
+    ('j', 'ʥ'),
+    ('y', 'j'),
+    ('ni', 'n^i'),
+    ('nj', 'n^'),
+    ('hi', 'çi'),
+    ('hj', 'ç'),
+    ('f', 'ɸ'),
+    ('I', 'i*'),
+    ('U', 'ɯ*'),
+    ('r', 'ɾ')
+]]
+
+# List of (romaji, ipa2) pairs for marks:
+_romaji_to_ipa2 = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('u', 'ɯ'),
+    ('ʧ', 'tʃ'),
+    ('j', 'dʑ'),
+    ('y', 'j'),
+    ('ni', 'n^i'),
+    ('nj', 'n^'),
+    ('hi', 'çi'),
+    ('hj', 'ç'),
+    ('f', 'ɸ'),
+    ('I', 'i*'),
+    ('U', 'ɯ*'),
+    ('r', 'ɾ')
+]]
+
+# List of (consonant, sokuon) pairs:
+_real_sokuon = [(re.compile('%s' % x[0]), x[1]) for x in [
+    (r'Q([↑↓]*[kg])', r'k#\1'),
+    (r'Q([↑↓]*[tdjʧ])', r't#\1'),
+    (r'Q([↑↓]*[sʃ])', r's\1'),
+    (r'Q([↑↓]*[pb])', r'p#\1')
+]]
+
+# List of (consonant, hatsuon) pairs:
+_real_hatsuon = [(re.compile('%s' % x[0]), x[1]) for x in [
+    (r'N([↑↓]*[pbm])', r'm\1'),
+    (r'N([↑↓]*[ʧʥj])', r'n^\1'),
+    (r'N([↑↓]*[tdn])', r'n\1'),
+    (r'N([↑↓]*[kg])', r'ŋ\1')
+]]
+
+
+def symbols_to_japanese(text):
+    for regex, replacement in _symbols_to_japanese:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def japanese_to_romaji_with_accent(text):
+    '''Reference https://r9y9.github.io/ttslearn/latest/notebooks/ch10_Recipe-Tacotron.html'''
+    text = symbols_to_japanese(text)
+    sentences = re.split(_japanese_marks, text)
+    marks = re.findall(_japanese_marks, text)
+    text = ''
+    for i, sentence in enumerate(sentences):
+        if re.match(_japanese_characters, sentence):
+            if text != '':
+                text += ' '
+            labels = pyopenjtalk.extract_fullcontext(sentence)
+            for n, label in enumerate(labels):
+                phoneme = re.search(r'\-([^\+]*)\+', label).group(1)
+                if phoneme not in ['sil', 'pau']:
+                    text += phoneme.replace('ch', 'ʧ').replace('sh',
+                                                               'ʃ').replace('cl', 'Q')
+                else:
+                    continue
+                # n_moras = int(re.search(r'/F:(\d+)_', label).group(1))
+                a1 = int(re.search(r"/A:(\-?[0-9]+)\+", label).group(1))
+                a2 = int(re.search(r"\+(\d+)\+", label).group(1))
+                a3 = int(re.search(r"\+(\d+)/", label).group(1))
+                if re.search(r'\-([^\+]*)\+', labels[n + 1]).group(1) in ['sil', 'pau']:
+                    a2_next = -1
+                else:
+                    a2_next = int(
+                        re.search(r"\+(\d+)\+", labels[n + 1]).group(1))
+                # Accent phrase boundary
+                if a3 == 1 and a2_next == 1:
+                    text += ' '
+                # Falling
+                elif a1 == 0 and a2_next == a2 + 1:
+                    text += '↓'
+                # Rising
+                elif a2 == 1 and a2_next == 2:
+                    text += '↑'
+        if i < len(marks):
+            text += unidecode(marks[i]).replace(' ', '')
+    return text
+
+
+def get_real_sokuon(text):
+    for regex, replacement in _real_sokuon:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def get_real_hatsuon(text):
+    for regex, replacement in _real_hatsuon:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def japanese_to_ipa(text):
+    text = japanese_to_romaji_with_accent(text).replace('...', '…')
+    text = re.sub(
+        r'([aiueo])\1+', lambda x: x.group(0)[0]+'ː'*(len(x.group(0))-1), text)
+    text = get_real_sokuon(text)
+    text = get_real_hatsuon(text)
+    for regex, replacement in _romaji_to_ipa:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def japanese_to_ipa2(text):
+    text = japanese_to_romaji_with_accent(text).replace('...', '…')
+    text = get_real_sokuon(text)
+    text = get_real_hatsuon(text)
+    for regex, replacement in _romaji_to_ipa2:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def japanese_to_ipa3(text):
+    text = japanese_to_ipa2(text).replace('n^', 'ȵ').replace(
+        'ʃ', 'ɕ').replace('*', '\u0325').replace('#', '\u031a')
+    text = re.sub(
+        r'([aiɯeo])\1+', lambda x: x.group(0)[0]+'ː'*(len(x.group(0))-1), text)
+    text = re.sub(r'((?:^|\s)(?:ts|tɕ|[kpt]))', r'\1ʰ', text)
+    return text

text/korean.py

@@ -0,0 +1,210 @@
+import re
+from jamo import h2j, j2hcj
+import ko_pron
+
+
+# This is a list of Korean classifiers preceded by pure Korean numerals.
+_korean_classifiers = '군데 권 개 그루 닢 대 두 마리 모 모금 뭇 발 발짝 방 번 벌 보루 살 수 술 시 쌈 움큼 정 짝 채 척 첩 축 켤레 톨 통'
+
+# List of (hangul, hangul divided) pairs:
+_hangul_divided = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('ㄳ', 'ㄱㅅ'),
+    ('ㄵ', 'ㄴㅈ'),
+    ('ㄶ', 'ㄴㅎ'),
+    ('ㄺ', 'ㄹㄱ'),
+    ('ㄻ', 'ㄹㅁ'),
+    ('ㄼ', 'ㄹㅂ'),
+    ('ㄽ', 'ㄹㅅ'),
+    ('ㄾ', 'ㄹㅌ'),
+    ('ㄿ', 'ㄹㅍ'),
+    ('ㅀ', 'ㄹㅎ'),
+    ('ㅄ', 'ㅂㅅ'),
+    ('ㅘ', 'ㅗㅏ'),
+    ('ㅙ', 'ㅗㅐ'),
+    ('ㅚ', 'ㅗㅣ'),
+    ('ㅝ', 'ㅜㅓ'),
+    ('ㅞ', 'ㅜㅔ'),
+    ('ㅟ', 'ㅜㅣ'),
+    ('ㅢ', 'ㅡㅣ'),
+    ('ㅑ', 'ㅣㅏ'),
+    ('ㅒ', 'ㅣㅐ'),
+    ('ㅕ', 'ㅣㅓ'),
+    ('ㅖ', 'ㅣㅔ'),
+    ('ㅛ', 'ㅣㅗ'),
+    ('ㅠ', 'ㅣㅜ')
+]]
+
+# List of (Latin alphabet, hangul) pairs:
+_latin_to_hangul = [(re.compile('%s' % x[0], re.IGNORECASE), x[1]) for x in [
+    ('a', '에이'),
+    ('b', '비'),
+    ('c', '시'),
+    ('d', '디'),
+    ('e', '이'),
+    ('f', '에프'),
+    ('g', '지'),
+    ('h', '에이치'),
+    ('i', '아이'),
+    ('j', '제이'),
+    ('k', '케이'),
+    ('l', '엘'),
+    ('m', '엠'),
+    ('n', '엔'),
+    ('o', '오'),
+    ('p', '피'),
+    ('q', '큐'),
+    ('r', '아르'),
+    ('s', '에스'),
+    ('t', '티'),
+    ('u', '유'),
+    ('v', '브이'),
+    ('w', '더블유'),
+    ('x', '엑스'),
+    ('y', '와이'),
+    ('z', '제트')
+]]
+
+# List of (ipa, lazy ipa) pairs:
+_ipa_to_lazy_ipa = [(re.compile('%s' % x[0], re.IGNORECASE), x[1]) for x in [
+    ('t͡ɕ','ʧ'),
+    ('d͡ʑ','ʥ'),
+    ('ɲ','n^'),
+    ('ɕ','ʃ'),
+    ('ʷ','w'),
+    ('ɭ','l`'),
+    ('ʎ','ɾ'),
+    ('ɣ','ŋ'),
+    ('ɰ','ɯ'),
+    ('ʝ','j'),
+    ('ʌ','ə'),
+    ('ɡ','g'),
+    ('\u031a','#'),
+    ('\u0348','='),
+    ('\u031e',''),
+    ('\u0320',''),
+    ('\u0339','')
+]]
+
+
+def latin_to_hangul(text):
+    for regex, replacement in _latin_to_hangul:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def divide_hangul(text):
+    text = j2hcj(h2j(text))
+    for regex, replacement in _hangul_divided:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def hangul_number(num, sino=True):
+    '''Reference https://github.com/Kyubyong/g2pK'''
+    num = re.sub(',', '', num)
+
+    if num == '0':
+        return '영'
+    if not sino and num == '20':
+        return '스무'
+
+    digits = '123456789'
+    names = '일이삼사오육칠팔구'
+    digit2name = {d: n for d, n in zip(digits, names)}
+
+    modifiers = '한 두 세 네 다섯 여섯 일곱 여덟 아홉'
+    decimals = '열 스물 서른 마흔 쉰 예순 일흔 여든 아흔'
+    digit2mod = {d: mod for d, mod in zip(digits, modifiers.split())}
+    digit2dec = {d: dec for d, dec in zip(digits, decimals.split())}
+
+    spelledout = []
+    for i, digit in enumerate(num):
+        i = len(num) - i - 1
+        if sino:
+            if i == 0:
+                name = digit2name.get(digit, '')
+            elif i == 1:
+                name = digit2name.get(digit, '') + '십'
+                name = name.replace('일십', '십')
+        else:
+            if i == 0:
+                name = digit2mod.get(digit, '')
+            elif i == 1:
+                name = digit2dec.get(digit, '')
+        if digit == '0':
+            if i % 4 == 0:
+                last_three = spelledout[-min(3, len(spelledout)):]
+                if ''.join(last_three) == '':
+                    spelledout.append('')
+                    continue
+            else:
+                spelledout.append('')
+                continue
+        if i == 2:
+            name = digit2name.get(digit, '') + '백'
+            name = name.replace('일백', '백')
+        elif i == 3:
+            name = digit2name.get(digit, '') + '천'
+            name = name.replace('일천', '천')
+        elif i == 4:
+            name = digit2name.get(digit, '') + '만'
+            name = name.replace('일만', '만')
+        elif i == 5:
+            name = digit2name.get(digit, '') + '십'
+            name = name.replace('일십', '십')
+        elif i == 6:
+            name = digit2name.get(digit, '') + '백'
+            name = name.replace('일백', '백')
+        elif i == 7:
+            name = digit2name.get(digit, '') + '천'
+            name = name.replace('일천', '천')
+        elif i == 8:
+            name = digit2name.get(digit, '') + '억'
+        elif i == 9:
+            name = digit2name.get(digit, '') + '십'
+        elif i == 10:
+            name = digit2name.get(digit, '') + '백'
+        elif i == 11:
+            name = digit2name.get(digit, '') + '천'
+        elif i == 12:
+            name = digit2name.get(digit, '') + '조'
+        elif i == 13:
+            name = digit2name.get(digit, '') + '십'
+        elif i == 14:
+            name = digit2name.get(digit, '') + '백'
+        elif i == 15:
+            name = digit2name.get(digit, '') + '천'
+        spelledout.append(name)
+    return ''.join(elem for elem in spelledout)
+
+
+def number_to_hangul(text):
+    '''Reference https://github.com/Kyubyong/g2pK'''
+    tokens = set(re.findall(r'(\d[\d,]*)([\uac00-\ud71f]+)', text))
+    for token in tokens:
+        num, classifier = token
+        if classifier[:2] in _korean_classifiers or classifier[0] in _korean_classifiers:
+            spelledout = hangul_number(num, sino=False)
+        else:
+            spelledout = hangul_number(num, sino=True)
+        text = text.replace(f'{num}{classifier}', f'{spelledout}{classifier}')
+    # digit by digit for remaining digits
+    digits = '0123456789'
+    names = '영일이삼사오육칠팔구'
+    for d, n in zip(digits, names):
+        text = text.replace(d, n)
+    return text
+
+
+def korean_to_lazy_ipa(text):
+    text = latin_to_hangul(text)
+    text = number_to_hangul(text)
+    text=re.sub('[\uac00-\ud7af]+',lambda x:ko_pron.romanise(x.group(0),'ipa').split('] ~ [')[0],text)
+    for regex, replacement in _ipa_to_lazy_ipa:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def korean_to_ipa(text):
+    text = korean_to_lazy_ipa(text)
+    return text.replace('ʧ','tʃ').replace('ʥ','dʑ')

text/mandarin.py

@@ -0,0 +1,330 @@
+import os
+import sys
+import re
+from pypinyin import lazy_pinyin, BOPOMOFO
+import jieba
+import cn2an
+import logging
+
+logging.getLogger('jieba').setLevel(logging.WARNING)
+jieba.set_dictionary(os.path.dirname(os.path.realpath(sys.argv[0]))+'/jieba/dict.txt')
+jieba.initialize()
+
+
+# List of (Latin alphabet, bopomofo) pairs:
+_latin_to_bopomofo = [(re.compile('%s' % x[0], re.IGNORECASE), x[1]) for x in [
+    ('a', 'ㄟˉ'),
+    ('b', 'ㄅㄧˋ'),
+    ('c', 'ㄙㄧˉ'),
+    ('d', 'ㄉㄧˋ'),
+    ('e', 'ㄧˋ'),
+    ('f', 'ㄝˊㄈㄨˋ'),
+    ('g', 'ㄐㄧˋ'),
+    ('h', 'ㄝˇㄑㄩˋ'),
+    ('i', 'ㄞˋ'),
+    ('j', 'ㄐㄟˋ'),
+    ('k', 'ㄎㄟˋ'),
+    ('l', 'ㄝˊㄛˋ'),
+    ('m', 'ㄝˊㄇㄨˋ'),
+    ('n', 'ㄣˉ'),
+    ('o', 'ㄡˉ'),
+    ('p', 'ㄆㄧˉ'),
+    ('q', 'ㄎㄧㄡˉ'),
+    ('r', 'ㄚˋ'),
+    ('s', 'ㄝˊㄙˋ'),
+    ('t', 'ㄊㄧˋ'),
+    ('u', 'ㄧㄡˉ'),
+    ('v', 'ㄨㄧˉ'),
+    ('w', 'ㄉㄚˋㄅㄨˋㄌㄧㄡˋ'),
+    ('x', 'ㄝˉㄎㄨˋㄙˋ'),
+    ('y', 'ㄨㄞˋ'),
+    ('z', 'ㄗㄟˋ')
+]]
+
+# List of (bopomofo, romaji) pairs:
+_bopomofo_to_romaji = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('ㄅㄛ', 'p⁼wo'),
+    ('ㄆㄛ', 'pʰwo'),
+    ('ㄇㄛ', 'mwo'),
+    ('ㄈㄛ', 'fwo'),
+    ('ㄅ', 'p⁼'),
+    ('ㄆ', 'pʰ'),
+    ('ㄇ', 'm'),
+    ('ㄈ', 'f'),
+    ('ㄉ', 't⁼'),
+    ('ㄊ', 'tʰ'),
+    ('ㄋ', 'n'),
+    ('ㄌ', 'l'),
+    ('ㄍ', 'k⁼'),
+    ('ㄎ', 'kʰ'),
+    ('ㄏ', 'h'),
+    ('ㄐ', 'ʧ⁼'),
+    ('ㄑ', 'ʧʰ'),
+    ('ㄒ', 'ʃ'),
+    ('ㄓ', 'ʦ`⁼'),
+    ('ㄔ', 'ʦ`ʰ'),
+    ('ㄕ', 's`'),
+    ('ㄖ', 'ɹ`'),
+    ('ㄗ', 'ʦ⁼'),
+    ('ㄘ', 'ʦʰ'),
+    ('ㄙ', 's'),
+    ('ㄚ', 'a'),
+    ('ㄛ', 'o'),
+    ('ㄜ', 'ə'),
+    ('ㄝ', 'e'),
+    ('ㄞ', 'ai'),
+    ('ㄟ', 'ei'),
+    ('ㄠ', 'au'),
+    ('ㄡ', 'ou'),
+    ('ㄧㄢ', 'yeNN'),
+    ('ㄢ', 'aNN'),
+    ('ㄧㄣ', 'iNN'),
+    ('ㄣ', 'əNN'),
+    ('ㄤ', 'aNg'),
+    ('ㄧㄥ', 'iNg'),
+    ('ㄨㄥ', 'uNg'),
+    ('ㄩㄥ', 'yuNg'),
+    ('ㄥ', 'əNg'),
+    ('ㄦ', 'əɻ'),
+    ('ㄧ', 'i'),
+    ('ㄨ', 'u'),
+    ('ㄩ', 'ɥ'),
+    ('ˉ', '→'),
+    ('ˊ', '↑'),
+    ('ˇ', '↓↑'),
+    ('ˋ', '↓'),
+    ('˙', ''),
+    ('，', ','),
+    ('。', '.'),
+    ('！', '!'),
+    ('？', '?'),
+    ('—', '-')
+]]
+
+# List of (romaji, ipa) pairs:
+_romaji_to_ipa = [(re.compile('%s' % x[0], re.IGNORECASE), x[1]) for x in [
+    ('ʃy', 'ʃ'),
+    ('ʧʰy', 'ʧʰ'),
+    ('ʧ⁼y', 'ʧ⁼'),
+    ('NN', 'n'),
+    ('Ng', 'ŋ'),
+    ('y', 'j'),
+    ('h', 'x')
+]]
+
+# List of (bopomofo, ipa) pairs:
+_bopomofo_to_ipa = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('ㄅㄛ', 'p⁼wo'),
+    ('ㄆㄛ', 'pʰwo'),
+    ('ㄇㄛ', 'mwo'),
+    ('ㄈㄛ', 'fwo'),
+    ('ㄅ', 'p⁼'),
+    ('ㄆ', 'pʰ'),
+    ('ㄇ', 'm'),
+    ('ㄈ', 'f'),
+    ('ㄉ', 't⁼'),
+    ('ㄊ', 'tʰ'),
+    ('ㄋ', 'n'),
+    ('ㄌ', 'l'),
+    ('ㄍ', 'k⁼'),
+    ('ㄎ', 'kʰ'),
+    ('ㄏ', 'x'),
+    ('ㄐ', 'tʃ⁼'),
+    ('ㄑ', 'tʃʰ'),
+    ('ㄒ', 'ʃ'),
+    ('ㄓ', 'ts`⁼'),
+    ('ㄔ', 'ts`ʰ'),
+    ('ㄕ', 's`'),
+    ('ㄖ', 'ɹ`'),
+    ('ㄗ', 'ts⁼'),
+    ('ㄘ', 'tsʰ'),
+    ('ㄙ', 's'),
+    ('ㄚ', 'a'),
+    ('ㄛ', 'o'),
+    ('ㄜ', 'ə'),
+    ('ㄝ', 'ɛ'),
+    ('ㄞ', 'aɪ'),
+    ('ㄟ', 'eɪ'),
+    ('ㄠ', 'ɑʊ'),
+    ('ㄡ', 'oʊ'),
+    ('ㄧㄢ', 'jɛn'),
+    ('ㄩㄢ', 'ɥæn'),
+    ('ㄢ', 'an'),
+    ('ㄧㄣ', 'in'),
+    ('ㄩㄣ', 'ɥn'),
+    ('ㄣ', 'ən'),
+    ('ㄤ', 'ɑŋ'),
+    ('ㄧㄥ', 'iŋ'),
+    ('ㄨㄥ', 'ʊŋ'),
+    ('ㄩㄥ', 'jʊŋ'),
+    ('ㄥ', 'əŋ'),
+    ('ㄦ', 'əɻ'),
+    ('ㄧ', 'i'),
+    ('ㄨ', 'u'),
+    ('ㄩ', 'ɥ'),
+    ('ˉ', '→'),
+    ('ˊ', '↑'),
+    ('ˇ', '↓↑'),
+    ('ˋ', '↓'),
+    ('˙', ''),
+    ('，', ','),
+    ('。', '.'),
+    ('！', '!'),
+    ('？', '?'),
+    ('—', '-')
+]]
+
+# List of (bopomofo, ipa2) pairs:
+_bopomofo_to_ipa2 = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('ㄅㄛ', 'pwo'),
+    ('ㄆㄛ', 'pʰwo'),
+    ('ㄇㄛ', 'mwo'),
+    ('ㄈㄛ', 'fwo'),
+    ('ㄅ', 'p'),
+    ('ㄆ', 'pʰ'),
+    ('ㄇ', 'm'),
+    ('ㄈ', 'f'),
+    ('ㄉ', 't'),
+    ('ㄊ', 'tʰ'),
+    ('ㄋ', 'n'),
+    ('ㄌ', 'l'),
+    ('ㄍ', 'k'),
+    ('ㄎ', 'kʰ'),
+    ('ㄏ', 'h'),
+    ('ㄐ', 'tɕ'),
+    ('ㄑ', 'tɕʰ'),
+    ('ㄒ', 'ɕ'),
+    ('ㄓ', 'tʂ'),
+    ('ㄔ', 'tʂʰ'),
+    ('ㄕ', 'ʂ'),
+    ('ㄖ', 'ɻ'),
+    ('ㄗ', 'ts'),
+    ('ㄘ', 'tsʰ'),
+    ('ㄙ', 's'),
+    ('ㄚ', 'a'),
+    ('ㄛ', 'o'),
+    ('ㄜ', 'ɤ'),
+    ('ㄝ', 'ɛ'),
+    ('ㄞ', 'aɪ'),
+    ('ㄟ', 'eɪ'),
+    ('ㄠ', 'ɑʊ'),
+    ('ㄡ', 'oʊ'),
+    ('ㄧㄢ', 'jɛn'),
+    ('ㄩㄢ', 'yæn'),
+    ('ㄢ', 'an'),
+    ('ㄧㄣ', 'in'),
+    ('ㄩㄣ', 'yn'),
+    ('ㄣ', 'ən'),
+    ('ㄤ', 'ɑŋ'),
+    ('ㄧㄥ', 'iŋ'),
+    ('ㄨㄥ', 'ʊŋ'),
+    ('ㄩㄥ', 'jʊŋ'),
+    ('ㄥ', 'ɤŋ'),
+    ('ㄦ', 'əɻ'),
+    ('ㄧ', 'i'),
+    ('ㄨ', 'u'),
+    ('ㄩ', 'y'),
+    ('ˉ', '˥'),
+    ('ˊ', '˧˥'),
+    ('ˇ', '˨˩˦'),
+    ('ˋ', '˥˩'),
+    ('˙', ''),
+    ('，', ','),
+    ('。', '.'),
+    ('！', '!'),
+    ('？', '?'),
+    ('—', '-')
+]]
+
+
+def number_to_chinese(text):
+    numbers = re.findall(r'\d+(?:\.?\d+)?', text)
+    for number in numbers:
+        text = text.replace(number, cn2an.an2cn(number), 1)
+    return text
+
+
+def chinese_to_bopomofo(text):
+    text = text.replace('、', '，').replace('；', '，').replace('：', '，')
+    words = jieba.lcut(text, cut_all=False)
+    text = ''
+    for word in words:
+        bopomofos = lazy_pinyin(word, BOPOMOFO)
+        if not re.search('[\u4e00-\u9fff]', word):
+            text += word
+            continue
+        for i in range(len(bopomofos)):
+            bopomofos[i] = re.sub(r'([\u3105-\u3129])$', r'\1ˉ', bopomofos[i])
+        if text != '':
+            text += ' '
+        text += ''.join(bopomofos)
+    return text
+
+
+def latin_to_bopomofo(text):
+    for regex, replacement in _latin_to_bopomofo:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def bopomofo_to_romaji(text):
+    for regex, replacement in _bopomofo_to_romaji:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def bopomofo_to_ipa(text):
+    for regex, replacement in _bopomofo_to_ipa:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def bopomofo_to_ipa2(text):
+    for regex, replacement in _bopomofo_to_ipa2:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def chinese_to_romaji(text):
+    text = number_to_chinese(text)
+    text = chinese_to_bopomofo(text)
+    text = latin_to_bopomofo(text)
+    text = bopomofo_to_romaji(text)
+    text = re.sub('i([aoe])', r'y\1', text)
+    text = re.sub('u([aoəe])', r'w\1', text)
+    text = re.sub('([ʦsɹ]`[⁼ʰ]?)([→↓↑ ]+|$)',
+                  r'\1ɹ`\2', text).replace('ɻ', 'ɹ`')
+    text = re.sub('([ʦs][⁼ʰ]?)([→↓↑ ]+|$)', r'\1ɹ\2', text)
+    return text
+
+
+def chinese_to_lazy_ipa(text):
+    text = chinese_to_romaji(text)
+    for regex, replacement in _romaji_to_ipa:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def chinese_to_ipa(text):
+    text = number_to_chinese(text)
+    text = chinese_to_bopomofo(text)
+    text = latin_to_bopomofo(text)
+    text = bopomofo_to_ipa(text)
+    text = re.sub('i([aoe])', r'j\1', text)
+    text = re.sub('u([aoəe])', r'w\1', text)
+    text = re.sub('([sɹ]`[⁼ʰ]?)([→↓↑ ]+|$)',
+                  r'\1ɹ`\2', text).replace('ɻ', 'ɹ`')
+    text = re.sub('([s][⁼ʰ]?)([→↓↑ ]+|$)', r'\1ɹ\2', text)
+    return text
+
+
+def chinese_to_ipa2(text):
+    text = number_to_chinese(text)
+    text = chinese_to_bopomofo(text)
+    text = latin_to_bopomofo(text)
+    text = bopomofo_to_ipa2(text)
+    text = re.sub(r'i([aoe])', r'j\1', text)
+    text = re.sub(r'u([aoəe])', r'w\1', text)
+    text = re.sub(r'([ʂɹ]ʰ?)([˩˨˧˦˥ ]+|$)', r'\1ʅ\2', text)
+    text = re.sub(r'(sʰ?)([˩˨˧˦˥ ]+|$)', r'\1ɿ\2', text)
+    return text

text/ngu_dialect.py

@@ -0,0 +1,30 @@
+import re
+import opencc
+
+
+dialects = {'SZ': 'suzhou', 'WX': 'wuxi', 'CZ': 'changzhou', 'HZ': 'hangzhou',
+            'SX': 'shaoxing', 'NB': 'ningbo', 'JJ': 'jingjiang', 'YX': 'yixing',
+            'JD': 'jiading', 'ZR': 'zhenru', 'PH': 'pinghu', 'TX': 'tongxiang',
+            'JS': 'jiashan', 'HN': 'xiashi', 'LP': 'linping', 'XS': 'xiaoshan',
+            'FY': 'fuyang', 'RA': 'ruao', 'CX': 'cixi', 'SM': 'sanmen',
+            'TT': 'tiantai', 'WZ': 'wenzhou', 'SC': 'suichang', 'YB': 'youbu'}
+
+converters = {}
+
+for dialect in dialects.values():
+    try:
+        converters[dialect] = opencc.OpenCC(dialect)
+    except:
+        pass
+
+
+def ngu_dialect_to_ipa(text, dialect):
+    dialect = dialects[dialect]
+    text = converters[dialect].convert(text).replace('-','').replace('$',' ')
+    text = re.sub(r'[、；：]', '，', text)
+    text = re.sub(r'\s*，\s*', ', ', text)
+    text = re.sub(r'\s*。\s*', '. ', text)
+    text = re.sub(r'\s*？\s*', '? ', text)
+    text = re.sub(r'\s*！\s*', '! ', text)
+    text = re.sub(r'\s*$', '', text)
+    return text

text/sanskrit.py

@@ -0,0 +1,62 @@
+import re
+from indic_transliteration import sanscript
+
+
+# List of (iast, ipa) pairs:
+_iast_to_ipa = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('a', 'ə'),
+    ('ā', 'aː'),
+    ('ī', 'iː'),
+    ('ū', 'uː'),
+    ('ṛ', 'ɹ`'),
+    ('ṝ', 'ɹ`ː'),
+    ('ḷ', 'l`'),
+    ('ḹ', 'l`ː'),
+    ('e', 'eː'),
+    ('o', 'oː'),
+    ('k', 'k⁼'),
+    ('k⁼h', 'kʰ'),
+    ('g', 'g⁼'),
+    ('g⁼h', 'gʰ'),
+    ('ṅ', 'ŋ'),
+    ('c', 'ʧ⁼'),
+    ('ʧ⁼h', 'ʧʰ'),
+    ('j', 'ʥ⁼'),
+    ('ʥ⁼h', 'ʥʰ'),
+    ('ñ', 'n^'),
+    ('ṭ', 't`⁼'),
+    ('t`⁼h', 't`ʰ'),
+    ('ḍ', 'd`⁼'),
+    ('d`⁼h', 'd`ʰ'),
+    ('ṇ', 'n`'),
+    ('t', 't⁼'),
+    ('t⁼h', 'tʰ'),
+    ('d', 'd⁼'),
+    ('d⁼h', 'dʰ'),
+    ('p', 'p⁼'),
+    ('p⁼h', 'pʰ'),
+    ('b', 'b⁼'),
+    ('b⁼h', 'bʰ'),
+    ('y', 'j'),
+    ('ś', 'ʃ'),
+    ('ṣ', 's`'),
+    ('r', 'ɾ'),
+    ('l̤', 'l`'),
+    ('h', 'ɦ'),
+    ("'", ''),
+    ('~', '^'),
+    ('ṃ', '^')
+]]
+
+
+def devanagari_to_ipa(text):
+    text = text.replace('ॐ', 'ओम्')
+    text = re.sub(r'\s*।\s*$', '.', text)
+    text = re.sub(r'\s*।\s*', ', ', text)
+    text = re.sub(r'\s*॥', '.', text)
+    text = sanscript.transliterate(text, sanscript.DEVANAGARI, sanscript.IAST)
+    for regex, replacement in _iast_to_ipa:
+        text = re.sub(regex, replacement, text)
+    text = re.sub('(.)[`ː]*ḥ', lambda x: x.group(0)
+                  [:-1]+'h'+x.group(1)+'*', text)
+    return text

text/shanghainese.py

@@ -0,0 +1,64 @@
+import re
+import cn2an
+import opencc
+
+
+converter = opencc.OpenCC('zaonhe')
+
+# List of (Latin alphabet, ipa) pairs:
+_latin_to_ipa = [(re.compile('%s' % x[0]), x[1]) for x in [
+    ('A', 'ᴇ'),
+    ('B', 'bi'),
+    ('C', 'si'),
+    ('D', 'di'),
+    ('E', 'i'),
+    ('F', 'ᴇf'),
+    ('G', 'dʑi'),
+    ('H', 'ᴇtɕʰ'),
+    ('I', 'ᴀi'),
+    ('J', 'dʑᴇ'),
+    ('K', 'kʰᴇ'),
+    ('L', 'ᴇl'),
+    ('M', 'ᴇm'),
+    ('N', 'ᴇn'),
+    ('O', 'o'),
+    ('P', 'pʰi'),
+    ('Q', 'kʰiu'),
+    ('R', 'ᴀl'),
+    ('S', 'ᴇs'),
+    ('T', 'tʰi'),
+    ('U', 'ɦiu'),
+    ('V', 'vi'),
+    ('W', 'dᴀbɤliu'),
+    ('X', 'ᴇks'),
+    ('Y', 'uᴀi'),
+    ('Z', 'zᴇ')
+]]
+
+
+def _number_to_shanghainese(num):
+    num = cn2an.an2cn(num).replace('一十','十').replace('二十', '廿').replace('二', '两')
+    return re.sub(r'((?:^|[^三四五六七八九])十|廿)两', r'\1二', num)
+
+
+def number_to_shanghainese(text):
+    return re.sub(r'\d+(?:\.?\d+)?', lambda x: _number_to_shanghainese(x.group()), text)
+
+
+def latin_to_ipa(text):
+    for regex, replacement in _latin_to_ipa:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def shanghainese_to_ipa(text):
+    text = number_to_shanghainese(text.upper())
+    text = converter.convert(text).replace('-','').replace('$',' ')
+    text = re.sub(r'[A-Z]', lambda x: latin_to_ipa(x.group())+' ', text)
+    text = re.sub(r'[、；：]', '，', text)
+    text = re.sub(r'\s*，\s*', ', ', text)
+    text = re.sub(r'\s*。\s*', '. ', text)
+    text = re.sub(r'\s*？\s*', '? ', text)
+    text = re.sub(r'\s*！\s*', '! ', text)
+    text = re.sub(r'\s*$', '', text)
+    return text

text/thai.py

@@ -0,0 +1,44 @@
+import re
+from num_thai.thainumbers import NumThai
+
+
+num = NumThai()
+
+# List of (Latin alphabet, Thai) pairs:
+_latin_to_thai = [(re.compile('%s' % x[0], re.IGNORECASE), x[1]) for x in [
+    ('a', 'เอ'),
+    ('b','บี'),
+    ('c','ซี'),
+    ('d','ดี'),
+    ('e','อี'),
+    ('f','เอฟ'),
+    ('g','จี'),
+    ('h','เอช'),
+    ('i','ไอ'),
+    ('j','เจ'),
+    ('k','เค'),
+    ('l','แอล'),
+    ('m','เอ็ม'),
+    ('n','เอ็น'),
+    ('o','โอ'),
+    ('p','พี'),
+    ('q','คิว'),
+    ('r','แอร์'),
+    ('s','เอส'),
+    ('t','ที'),
+    ('u','ยู'),
+    ('v','วี'),
+    ('w','ดับเบิลยู'),
+    ('x','เอ็กซ์'),
+    ('y','วาย'),
+    ('z','ซี')
+]]
+
+
+def num_to_thai(text):
+    return re.sub(r'(?:\d+(?:,?\d+)?)+(?:\.\d+(?:,?\d+)?)?', lambda x: ''.join(num.NumberToTextThai(float(x.group(0).replace(',', '')))), text)
+
+def latin_to_thai(text):
+    for regex, replacement in _latin_to_thai:
+        text = re.sub(regex, replacement, text)
+    return text

transforms.py

@@ -0,0 +1,193 @@
+import torch
+from torch.nn import functional as F
+
+import numpy as np
+
+
+DEFAULT_MIN_BIN_WIDTH = 1e-3
+DEFAULT_MIN_BIN_HEIGHT = 1e-3
+DEFAULT_MIN_DERIVATIVE = 1e-3
+
+
+def piecewise_rational_quadratic_transform(inputs, 
+                                           unnormalized_widths,
+                                           unnormalized_heights,
+                                           unnormalized_derivatives,
+                                           inverse=False,
+                                           tails=None, 
+                                           tail_bound=1.,
+                                           min_bin_width=DEFAULT_MIN_BIN_WIDTH,
+                                           min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
+                                           min_derivative=DEFAULT_MIN_DERIVATIVE):
+
+    if tails is None:
+        spline_fn = rational_quadratic_spline
+        spline_kwargs = {}
+    else:
+        spline_fn = unconstrained_rational_quadratic_spline
+        spline_kwargs = {
+            'tails': tails,
+            'tail_bound': tail_bound
+        }
+
+    outputs, logabsdet = spline_fn(
+            inputs=inputs,
+            unnormalized_widths=unnormalized_widths,
+            unnormalized_heights=unnormalized_heights,
+            unnormalized_derivatives=unnormalized_derivatives,
+            inverse=inverse,
+            min_bin_width=min_bin_width,
+            min_bin_height=min_bin_height,
+            min_derivative=min_derivative,
+            **spline_kwargs
+    )
+    return outputs, logabsdet
+
+
+def searchsorted(bin_locations, inputs, eps=1e-6):
+    bin_locations[..., -1] += eps
+    return torch.sum(
+        inputs[..., None] >= bin_locations,
+        dim=-1
+    ) - 1
+
+
+def unconstrained_rational_quadratic_spline(inputs,
+                                            unnormalized_widths,
+                                            unnormalized_heights,
+                                            unnormalized_derivatives,
+                                            inverse=False,
+                                            tails='linear',
+                                            tail_bound=1.,
+                                            min_bin_width=DEFAULT_MIN_BIN_WIDTH,
+                                            min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
+                                            min_derivative=DEFAULT_MIN_DERIVATIVE):
+    inside_interval_mask = (inputs >= -tail_bound) & (inputs <= tail_bound)
+    outside_interval_mask = ~inside_interval_mask
+
+    outputs = torch.zeros_like(inputs)
+    logabsdet = torch.zeros_like(inputs)
+
+    if tails == 'linear':
+        unnormalized_derivatives = F.pad(unnormalized_derivatives, pad=(1, 1))
+        constant = np.log(np.exp(1 - min_derivative) - 1)
+        unnormalized_derivatives[..., 0] = constant
+        unnormalized_derivatives[..., -1] = constant
+
+        outputs[outside_interval_mask] = inputs[outside_interval_mask]
+        logabsdet[outside_interval_mask] = 0
+    else:
+        raise RuntimeError('{} tails are not implemented.'.format(tails))
+
+    outputs[inside_interval_mask], logabsdet[inside_interval_mask] = rational_quadratic_spline(
+        inputs=inputs[inside_interval_mask],
+        unnormalized_widths=unnormalized_widths[inside_interval_mask, :],
+        unnormalized_heights=unnormalized_heights[inside_interval_mask, :],
+        unnormalized_derivatives=unnormalized_derivatives[inside_interval_mask, :],
+        inverse=inverse,
+        left=-tail_bound, right=tail_bound, bottom=-tail_bound, top=tail_bound,
+        min_bin_width=min_bin_width,
+        min_bin_height=min_bin_height,
+        min_derivative=min_derivative
+    )
+
+    return outputs, logabsdet
+
+def rational_quadratic_spline(inputs,
+                              unnormalized_widths,
+                              unnormalized_heights,
+                              unnormalized_derivatives,
+                              inverse=False,
+                              left=0., right=1., bottom=0., top=1.,
+                              min_bin_width=DEFAULT_MIN_BIN_WIDTH,
+                              min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
+                              min_derivative=DEFAULT_MIN_DERIVATIVE):
+    if torch.min(inputs) < left or torch.max(inputs) > right:
+        raise ValueError('Input to a transform is not within its domain')
+
+    num_bins = unnormalized_widths.shape[-1]
+
+    if min_bin_width * num_bins > 1.0:
+        raise ValueError('Minimal bin width too large for the number of bins')
+    if min_bin_height * num_bins > 1.0:
+        raise ValueError('Minimal bin height too large for the number of bins')
+
+    widths = F.softmax(unnormalized_widths, dim=-1)
+    widths = min_bin_width + (1 - min_bin_width * num_bins) * widths
+    cumwidths = torch.cumsum(widths, dim=-1)
+    cumwidths = F.pad(cumwidths, pad=(1, 0), mode='constant', value=0.0)
+    cumwidths = (right - left) * cumwidths + left
+    cumwidths[..., 0] = left
+    cumwidths[..., -1] = right
+    widths = cumwidths[..., 1:] - cumwidths[..., :-1]
+
+    derivatives = min_derivative + F.softplus(unnormalized_derivatives)
+
+    heights = F.softmax(unnormalized_heights, dim=-1)
+    heights = min_bin_height + (1 - min_bin_height * num_bins) * heights
+    cumheights = torch.cumsum(heights, dim=-1)
+    cumheights = F.pad(cumheights, pad=(1, 0), mode='constant', value=0.0)
+    cumheights = (top - bottom) * cumheights + bottom
+    cumheights[..., 0] = bottom
+    cumheights[..., -1] = top
+    heights = cumheights[..., 1:] - cumheights[..., :-1]
+
+    if inverse:
+        bin_idx = searchsorted(cumheights, inputs)[..., None]
+    else:
+        bin_idx = searchsorted(cumwidths, inputs)[..., None]
+
+    input_cumwidths = cumwidths.gather(-1, bin_idx)[..., 0]
+    input_bin_widths = widths.gather(-1, bin_idx)[..., 0]
+
+    input_cumheights = cumheights.gather(-1, bin_idx)[..., 0]
+    delta = heights / widths
+    input_delta = delta.gather(-1, bin_idx)[..., 0]
+
+    input_derivatives = derivatives.gather(-1, bin_idx)[..., 0]
+    input_derivatives_plus_one = derivatives[..., 1:].gather(-1, bin_idx)[..., 0]
+
+    input_heights = heights.gather(-1, bin_idx)[..., 0]
+
+    if inverse:
+        a = (((inputs - input_cumheights) * (input_derivatives
+                                             + input_derivatives_plus_one
+                                             - 2 * input_delta)
+              + input_heights * (input_delta - input_derivatives)))
+        b = (input_heights * input_derivatives
+             - (inputs - input_cumheights) * (input_derivatives
+                                              + input_derivatives_plus_one
+                                              - 2 * input_delta))
+        c = - input_delta * (inputs - input_cumheights)
+
+        discriminant = b.pow(2) - 4 * a * c
+        assert (discriminant >= 0).all()
+
+        root = (2 * c) / (-b - torch.sqrt(discriminant))
+        outputs = root * input_bin_widths + input_cumwidths
+
+        theta_one_minus_theta = root * (1 - root)
+        denominator = input_delta + ((input_derivatives + input_derivatives_plus_one - 2 * input_delta)
+                                     * theta_one_minus_theta)
+        derivative_numerator = input_delta.pow(2) * (input_derivatives_plus_one * root.pow(2)
+                                                     + 2 * input_delta * theta_one_minus_theta
+                                                     + input_derivatives * (1 - root).pow(2))
+        logabsdet = torch.log(derivative_numerator) - 2 * torch.log(denominator)
+
+        return outputs, -logabsdet
+    else:
+        theta = (inputs - input_cumwidths) / input_bin_widths
+        theta_one_minus_theta = theta * (1 - theta)
+
+        numerator = input_heights * (input_delta * theta.pow(2)
+                                     + input_derivatives * theta_one_minus_theta)
+        denominator = input_delta + ((input_derivatives + input_derivatives_plus_one - 2 * input_delta)
+                                     * theta_one_minus_theta)
+        outputs = input_cumheights + numerator / denominator
+
+        derivative_numerator = input_delta.pow(2) * (input_derivatives_plus_one * theta.pow(2)
+                                                     + 2 * input_delta * theta_one_minus_theta
+                                                     + input_derivatives * (1 - theta).pow(2))
+        logabsdet = torch.log(derivative_numerator) - 2 * torch.log(denominator)
+
+        return outputs, logabsdet

utils.py

@@ -0,0 +1,171 @@
+import logging
+import os
+import sys
+from io import BytesIO
+from json import loads
+import av
+import pilk
+from flask import Flask
+from torch import load, FloatTensor
+from numpy import float32
+import librosa
+
+from voice import Voice
+
+app = Flask(__name__)
+app.config.from_pyfile("config.py")
+
+
+class HParams():
+    def __init__(self, **kwargs):
+        for k, v in kwargs.items():
+            if type(v) == dict:
+                v = HParams(**v)
+            self[k] = v
+
+    def keys(self):
+        return self.__dict__.keys()
+
+    def items(self):
+        return self.__dict__.items()
+
+    def values(self):
+        return self.__dict__.values()
+
+    def __len__(self):
+        return len(self.__dict__)
+
+    def __getitem__(self, key):
+        return getattr(self, key)
+
+    def __setitem__(self, key, value):
+        return setattr(self, key, value)
+
+    def __contains__(self, key):
+        return key in self.__dict__
+
+    def __repr__(self):
+        return self.__dict__.__repr__()
+
+
+def load_checkpoint(checkpoint_path, model):
+    checkpoint_dict = load(checkpoint_path, map_location='cpu')
+    iteration = checkpoint_dict['iteration']
+    saved_state_dict = checkpoint_dict['model']
+    if hasattr(model, 'module'):
+        state_dict = model.module.state_dict()
+    else:
+        state_dict = model.state_dict()
+    new_state_dict = {}
+    for k, v in state_dict.items():
+        try:
+            new_state_dict[k] = saved_state_dict[k]
+        except:
+            logging.info("%s is not in the checkpoint" % k)
+            new_state_dict[k] = v
+    if hasattr(model, 'module'):
+        model.module.load_state_dict(new_state_dict)
+    else:
+        model.load_state_dict(new_state_dict)
+    logging.info("Loaded checkpoint '{}' (iteration {})".format(
+        checkpoint_path, iteration))
+    return
+
+
+def get_hparams_from_file(config_path):
+    with open(config_path, "r") as f:
+        data = f.read()
+    config = loads(data)
+
+    hparams = HParams(**config)
+    return hparams
+
+
+def load_audio_to_torch(full_path, target_sampling_rate):
+    audio, sampling_rate = librosa.load(full_path, sr=target_sampling_rate, mono=True)
+    return FloatTensor(audio.astype(float32))
+
+
+def wav2ogg(input, output):
+    with av.open(input, 'rb') as i:
+        with av.open(output, 'wb', format='ogg') as o:
+            out_stream = o.add_stream('libvorbis')
+            for frame in i.decode(audio=0):
+                for p in out_stream.encode(frame):
+                    o.mux(p)
+
+            for p in out_stream.encode(None):
+                o.mux(p)
+
+
+# def wav2silk(input, output):
+#     with av.open(input) as in_wav:
+#         in_stream = in_wav.streams.audio[0]
+#         sample_rate = in_stream.codec_context.sample_rate
+#         with BytesIO() as pcm:
+#             with av.open(pcm, 'w', 's16le') as out_pcm:
+#                 out_stream = out_pcm.add_stream(
+#                     'pcm_s16le',
+#                     rate=sample_rate,
+#                     layout='mono'
+#                 )
+#                 for frame in in_wav.decode(in_stream):
+#                     frame.pts = None
+#                     for packet in out_stream.encode(frame):
+#                         out_pcm.mux(packet)
+#
+#             pilk.encode(out_pcm, output, pcm_rate=sample_rate, tencent=True)
+
+
+def to_pcm(in_path: str) -> tuple[str, int]:
+    out_path = os.path.splitext(in_path)[0] + '.pcm'
+    with av.open(in_path) as in_container:
+        in_stream = in_container.streams.audio[0]
+        sample_rate = in_stream.codec_context.sample_rate
+        with av.open(out_path, 'w', 's16le') as out_container:
+            out_stream = out_container.add_stream(
+                'pcm_s16le',
+                rate=sample_rate,
+                layout='mono'
+            )
+            try:
+                for frame in in_container.decode(in_stream):
+                    frame.pts = None
+                    for packet in out_stream.encode(frame):
+                        out_container.mux(packet)
+            except:
+                pass
+    return out_path, sample_rate
+
+
+def convert_to_silk(media_path: str) -> str:
+    pcm_path, sample_rate = to_pcm(media_path)
+    silk_path = os.path.splitext(pcm_path)[0] + '.silk'
+    pilk.encode(pcm_path, silk_path, pcm_rate=sample_rate, tencent=True)
+    os.remove(pcm_path)
+    return silk_path
+
+
+def clean_folder(folder_path):
+    for filename in os.listdir(folder_path):
+        file_path = os.path.join(folder_path, filename)
+        # 如果是文件，则删除文件
+        if os.path.isfile(file_path):
+            os.remove(file_path)
+
+
+def merge_model(merging_model):
+    voice_obj = []
+    voice_speakers = []
+    new_id = 0
+    out_path = os.path.dirname(os.path.realpath(sys.argv[0])) + "/out_silk"
+    for obj_id, i in enumerate(merging_model):
+        obj = Voice(i[0], i[1], out_path)
+
+        for id, name in enumerate(obj.return_speakers()):
+            voice_obj.append([int(id), obj, obj_id])
+            voice_speakers.append({new_id: name})
+
+            new_id += 1
+
+    return voice_obj, voice_speakers

voice.py

@@ -0,0 +1,231 @@
+import os
+
+import librosa
+from scipy.io.wavfile import write
+from mel_processing import spectrogram_torch
+from text import text_to_sequence, _clean_text
+from models import SynthesizerTrn
+import utils
+import commons
+import sys
+import re
+import numpy as np
+# import torch
+# torch.set_num_threads(1) #设置torch线程为1，防止多任务推理时服务崩溃，但flask仍然会使用多线程
+from torch import no_grad, LongTensor, inference_mode, FloatTensor
+import audonnx
+import uuid
+from io import BytesIO
+
+
+class Voice:
+    def __init__(self, model, config, out_path=None):
+        self.out_path = out_path
+        if not os.path.exists(self.out_path):
+            try:
+                os.mkdir(self.out_path)
+            except:
+                pass
+
+        self.hps_ms = utils.get_hparams_from_file(config)
+        self.n_speakers = self.hps_ms.data.n_speakers if 'n_speakers' in self.hps_ms.data.keys() else 0
+        self.n_symbols = len(self.hps_ms.symbols) if 'symbols' in self.hps_ms.keys() else 0
+        self.speakers = self.hps_ms.speakers if 'speakers' in self.hps_ms.keys() else ['0']
+        self.use_f0 = self.hps_ms.data.use_f0 if 'use_f0' in self.hps_ms.data.keys() else False
+        self.emotion_embedding = self.hps_ms.data.emotion_embedding if 'emotion_embedding' in self.hps_ms.data.keys() else False
+
+        self.net_g_ms = SynthesizerTrn(
+            self.n_symbols,
+            self.hps_ms.data.filter_length // 2 + 1,
+            self.hps_ms.train.segment_size // self.hps_ms.data.hop_length,
+            n_speakers=self.n_speakers,
+            emotion_embedding=self.emotion_embedding,
+            **self.hps_ms.model)
+        _ = self.net_g_ms.eval()
+        utils.load_checkpoint(model, self.net_g_ms)
+
+    def get_text(self, text, hps, cleaned=False):
+        if cleaned:
+            text_norm = text_to_sequence(text, hps.symbols, [])
+        else:
+            text_norm = text_to_sequence(text, hps.symbols, hps.data.text_cleaners)
+        if hps.data.add_blank:
+            text_norm = commons.intersperse(text_norm, 0)
+        text_norm = LongTensor(text_norm)
+        return text_norm
+
+    def get_label_value(self, text, label, default, warning_name='value'):
+        value = re.search(rf'\[{label}=(.+?)\]', text)
+        if value:
+            try:
+                text = re.sub(rf'\[{label}=(.+?)\]', '', text, 1)
+                value = float(value.group(1))
+            except:
+                print(f'Invalid {warning_name}!')
+                sys.exit(1)
+        else:
+            value = default
+        return value, text
+
+    def ex_return(self, text, escape=False):
+        if escape:
+            return text.encode('unicode_escape').decode()
+        else:
+            return text
+
+    def return_speakers(self, escape=False):
+        return self.speakers
+
+    def get_label(self, text, label):
+        if f'[{label}]' in text:
+            return True, text.replace(f'[{label}]', '')
+        else:
+            return False, text
+
+    def generate(self, text=None, speaker_id=None, format=None, speed=1, audio_path=None, target_id=None, escape=False,
+                 option=None, w2v2_folder=None):
+        if self.n_symbols != 0:
+            if not self.emotion_embedding:
+                length_scale, text = self.get_label_value(text, 'LENGTH', speed, 'length scale')
+                noise_scale, text = self.get_label_value(text, 'NOISE', 0.667, 'noise scale')
+                noise_scale_w, text = self.get_label_value(text, 'NOISEW', 0.8, 'deviation of noise')
+                cleaned, text = self.get_label(text, 'CLEANED')
+
+                stn_tst = self.get_text(text, self.hps_ms, cleaned=cleaned)
+                with no_grad():
+                    x_tst = stn_tst.unsqueeze(0)
+                    x_tst_lengths = LongTensor([stn_tst.size(0)])
+                    sid = LongTensor([speaker_id])
+                    audio = self.net_g_ms.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()
+
+            # else:
+            #     w2v2_model = audonnx.load(os.path.dirname(w2v2_folder))
+            #
+            #     if option == 'clean':
+            #         self.ex_print(_clean_text(
+            #             text, self.hps_ms.data.text_cleaners), escape)
+            #
+            #     length_scale, text = self.get_label_value(
+            #         text, 'LENGTH', 1, 'length scale')
+            #     noise_scale, text = self.get_label_value(
+            #         text, 'NOISE', 0.667, 'noise scale')
+            #     noise_scale_w, text = self.get_label_value(
+            #         text, 'NOISEW', 0.8, 'deviation of noise')
+            #     cleaned, text = self.get_label(text, 'CLEANED')
+            #
+            #     stn_tst = self.get_text(text, self.hps_ms, cleaned=cleaned)
+            #
+            #     emotion_reference = input('Path of an emotion reference: ')
+            #     if emotion_reference.endswith('.npy'):
+            #         emotion = np.load(emotion_reference)
+            #         emotion = FloatTensor(emotion).unsqueeze(0)
+            #     else:
+            #         audio16000, sampling_rate = librosa.load(
+            #             emotion_reference, sr=16000, mono=True)
+            #         emotion = w2v2_model(audio16000, sampling_rate)[
+            #             'hidden_states']
+            #         emotion_reference = re.sub(
+            #             r'\..*$', '', emotion_reference)
+            #         np.save(emotion_reference, emotion.squeeze(0))
+            #         emotion = FloatTensor(emotion)
+            #
+            #
+            #     with no_grad():
+            #         x_tst = stn_tst.unsqueeze(0)
+            #         x_tst_lengths = LongTensor([stn_tst.size(0)])
+            #         sid = LongTensor([speaker_id])
+            #         audio = self.net_g_ms.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale,
+            #                                     noise_scale_w=noise_scale_w,
+            #                                     length_scale=length_scale, emotion_embedding=emotion)[0][
+            #             0, 0].data.cpu().float().numpy()
+
+        # else:
+        # model = input('Path of a hubert-soft Model: ')
+        # from hubert_model import hubert_soft
+        # hubert = hubert_soft(model)
+
+        # if audio_path != '[VC]':
+        #     if self.use_f0:
+        #         audio, sampling_rate = librosa.load(
+        #             audio_path, sr=self.hps_ms.data.sampling_rate, mono=True)
+        #         audio16000 = librosa.resample(
+        #             audio, orig_sr=sampling_rate, target_sr=16000)
+        #     else:
+        #         audio16000, sampling_rate = librosa.load(
+        #             audio_path, sr=16000, mono=True)
+        #
+        #     out_path = "H:/git/MoeGoe-Simple-API/upload/hubert.wav"
+        #     length_scale, out_path = self.get_label_value(
+        #         out_path, 'LENGTH', 1, 'length scale')
+        #     noise_scale, out_path = self.get_label_value(
+        #         out_path, 'NOISE', 0.1, 'noise scale')
+        #     noise_scale_w, out_path = self.get_label_value(
+        #         out_path, 'NOISEW', 0.1, 'deviation of noise')
+        #
+        #     with inference_mode():
+        #         units = hubert.units(FloatTensor(audio16000).unsqueeze(
+        #             0).unsqueeze(0)).squeeze(0).numpy()
+        #         if self.use_f0:
+        #             f0_scale, out_path = self.get_label_value(
+        #                 out_path, 'F0', 1, 'f0 scale')
+        #             f0 = librosa.pyin(audio, sr=sampling_rate,
+        #                               fmin=librosa.note_to_hz('C0'),
+        #                               fmax=librosa.note_to_hz('C7'),
+        #                               frame_length=1780)[0]
+        #             target_length = len(units[:, 0])
+        #             f0 = np.nan_to_num(np.interp(np.arange(0, len(f0) * target_length, len(f0)) / target_length,
+        #                                          np.arange(0, len(f0)), f0)) * f0_scale
+        #             units[:, 0] = f0 / 10
+        #
+        #     stn_tst = FloatTensor(units)
+        #     with no_grad():
+        #         x_tst = stn_tst.unsqueeze(0)
+        #         x_tst_lengths = LongTensor([stn_tst.size(0)])
+        #         sid = LongTensor([target_id])
+        #         audio = self.net_g_ms.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.float().numpy()
+
+        with BytesIO() as f:
+            fname = str(uuid.uuid1())
+
+            if format == 'ogg':
+                write(f, self.hps_ms.data.sampling_rate, audio)
+                with BytesIO() as o:
+                    utils.wav2ogg(f, o)
+                    return BytesIO(o.getvalue()), "audio/ogg", fname + ".ogg"
+            elif format == 'silk':
+                file_path = self.out_path + "/" + fname + ".wav"
+                write(file_path, 24000, audio)
+                silk_path = utils.convert_to_silk(file_path)
+                os.remove(file_path)
+                return silk_path, "audio/silk", fname + ".silk"
+            else:
+                write(f, self.hps_ms.data.sampling_rate, audio)
+                return BytesIO(f.getvalue()), "audio/wav", fname + ".wav"
+
+    def voice_conversion(self, audio_path, original_id, target_id):
+
+        audio = utils.load_audio_to_torch(
+            audio_path, self.hps_ms.data.sampling_rate)
+
+        y = audio.unsqueeze(0)
+
+        spec = spectrogram_torch(y, self.hps_ms.data.filter_length,
+                                 self.hps_ms.data.sampling_rate, self.hps_ms.data.hop_length,
+                                 self.hps_ms.data.win_length,
+                                 center=False)
+        spec_lengths = LongTensor([spec.size(-1)])
+        sid_src = LongTensor([original_id])
+
+        with no_grad():
+            sid_tgt = LongTensor([target_id])
+            audio = self.net_g_ms.voice_conversion(spec, spec_lengths, sid_src=sid_src, sid_tgt=sid_tgt)[
+                0][0, 0].data.cpu().float().numpy()
+
+        with BytesIO() as f:
+            write(f, self.hps_ms.data.sampling_rate, audio)
+            return BytesIO(f.getvalue())