Upload folder using huggingface_hub

.gitignore

@@ -0,0 +1,162 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+.DS_Store

Data/configs/haruka.json

@@ -0,0 +1,99 @@
+{
+  "train": {
+    "log_interval": 200,
+    "eval_interval": 1000,
+    "seed": 42,
+    "epochs": 1000,
+    "learning_rate": 0.0002,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 4,
+    "fp16_run": false,
+    "lr_decay": 0.99995,
+    "segment_size": 16384,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "skip_optimizer": true
+  },
+  "data": {
+    "training_files": "Data/filelists/train.list",
+    "validation_files": "Data/filelists/val.list",
+    "max_wav_value": 32768.0,
+    "sampling_rate": 44100,
+    "filter_length": 2048,
+    "hop_length": 512,
+    "win_length": 2048,
+    "n_mel_channels": 128,
+    "mel_fmin": 0.0,
+    "mel_fmax": null,
+    "add_blank": true,
+    "n_speakers": 4,
+    "cleaned_text": true,
+    "spk2id": {
+      "希亜": 0,
+      "天": 1,
+      "春風": 2,
+      "都": 3
+    }
+  },
+  "model": {
+    "use_spk_conditioned_encoder": true,
+    "use_noise_scaled_mas": true,
+    "use_mel_posterior_encoder": false,
+    "use_duration_discriminator": true,
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.1,
+    "resblock": "1",
+    "resblock_kernel_sizes": [
+      3,
+      7,
+      11
+    ],
+    "resblock_dilation_sizes": [
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ]
+    ],
+    "upsample_rates": [
+      8,
+      8,
+      2,
+      2,
+      2
+    ],
+    "upsample_initial_channel": 512,
+    "upsample_kernel_sizes": [
+      16,
+      16,
+      8,
+      2,
+      2
+    ],
+    "n_layers_q": 3,
+    "use_spectral_norm": false,
+    "gin_channels": 256
+  },
+  "version": "2.1"
+}

Data/models/G_82000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cff7a84b5bc7fbb44df429e1737132eebe0a01588acb0e7c2a30cb638ac08f5
+size 718770997

app.py

@@ -0,0 +1,485 @@
+# flake8: noqa: E402
+import os
+import logging
+import re_matching
+from tools.sentence import split_by_language
+
+logging.getLogger("numba").setLevel(logging.WARNING)
+logging.getLogger("markdown_it").setLevel(logging.WARNING)
+logging.getLogger("urllib3").setLevel(logging.WARNING)
+logging.getLogger("matplotlib").setLevel(logging.WARNING)
+
+logging.basicConfig(
+    level=logging.INFO, format="| %(name)s | %(levelname)s | %(message)s"
+)
+
+logger = logging.getLogger(__name__)
+
+import torch
+import utils
+from infer import infer, latest_version, get_net_g, infer_multilang
+import gradio as gr
+import webbrowser
+import numpy as np
+from config import config
+from tools.translate import translate
+import librosa
+
+net_g = None
+
+device = config.webui_config.device
+if device == "mps":
+    os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"
+
+
+def generate_audio(
+    slices,
+    sdp_ratio,
+    noise_scale,
+    noise_scale_w,
+    length_scale,
+    speaker,
+    language,
+    reference_audio,
+    emotion,
+    skip_start=False,
+    skip_end=False,
+):
+    audio_list = []
+    # silence = np.zeros(hps.data.sampling_rate // 2, dtype=np.int16)
+    with torch.no_grad():
+        for idx, piece in enumerate(slices):
+            skip_start = (idx != 0) and skip_start
+            skip_end = (idx != len(slices) - 1) and skip_end
+            audio = infer(
+                piece,
+                reference_audio=reference_audio,
+                emotion=emotion,
+                sdp_ratio=sdp_ratio,
+                noise_scale=noise_scale,
+                noise_scale_w=noise_scale_w,
+                length_scale=length_scale,
+                sid=speaker,
+                language=language,
+                hps=hps,
+                net_g=net_g,
+                device=device,
+                skip_start=skip_start,
+                skip_end=skip_end,
+            )
+            audio16bit = gr.processing_utils.convert_to_16_bit_wav(audio)
+            audio_list.append(audio16bit)
+            # audio_list.append(silence)  # 将静音添加到列表中
+    return audio_list
+
+
+def generate_audio_multilang(
+    slices,
+    sdp_ratio,
+    noise_scale,
+    noise_scale_w,
+    length_scale,
+    speaker,
+    language,
+    reference_audio,
+    emotion,
+    skip_start=False,
+    skip_end=False,
+):
+    audio_list = []
+    # silence = np.zeros(hps.data.sampling_rate // 2, dtype=np.int16)
+    with torch.no_grad():
+        for idx, piece in enumerate(slices):
+            skip_start = (idx != 0) and skip_start
+            skip_end = (idx != len(slices) - 1) and skip_end
+            audio = infer_multilang(
+                piece,
+                reference_audio=reference_audio,
+                emotion=emotion,
+                sdp_ratio=sdp_ratio,
+                noise_scale=noise_scale,
+                noise_scale_w=noise_scale_w,
+                length_scale=length_scale,
+                sid=speaker,
+                language=language[idx],
+                hps=hps,
+                net_g=net_g,
+                device=device,
+                skip_start=skip_start,
+                skip_end=skip_end,
+            )
+            audio16bit = gr.processing_utils.convert_to_16_bit_wav(audio)
+            audio_list.append(audio16bit)
+            # audio_list.append(silence)  # 将静音添加到列表中
+    return audio_list
+
+
+def tts_split(
+    text: str,
+    speaker,
+    sdp_ratio,
+    noise_scale,
+    noise_scale_w,
+    length_scale,
+    language,
+    cut_by_sent,
+    interval_between_para,
+    interval_between_sent,
+    reference_audio,
+    emotion,
+):
+    if language == "mix":
+        return ("invalid", None)
+    while text.find("\n\n") != -1:
+        text = text.replace("\n\n", "\n")
+    para_list = re_matching.cut_para(text)
+    audio_list = []
+    if not cut_by_sent:
+        for idx, p in enumerate(para_list):
+            skip_start = idx != 0
+            skip_end = idx != len(para_list) - 1
+            audio = infer(
+                p,
+                reference_audio=reference_audio,
+                emotion=emotion,
+                sdp_ratio=sdp_ratio,
+                noise_scale=noise_scale,
+                noise_scale_w=noise_scale_w,
+                length_scale=length_scale,
+                sid=speaker,
+                language=language,
+                hps=hps,
+                net_g=net_g,
+                device=device,
+                skip_start=skip_start,
+                skip_end=skip_end,
+            )
+            audio16bit = gr.processing_utils.convert_to_16_bit_wav(audio)
+            audio_list.append(audio16bit)
+            silence = np.zeros((int)(44100 * interval_between_para), dtype=np.int16)
+            audio_list.append(silence)
+    else:
+        for idx, p in enumerate(para_list):
+            skip_start = idx != 0
+            skip_end = idx != len(para_list) - 1
+            audio_list_sent = []
+            sent_list = re_matching.cut_sent(p)
+            for idx, s in enumerate(sent_list):
+                skip_start = (idx != 0) and skip_start
+                skip_end = (idx != len(sent_list) - 1) and skip_end
+                audio = infer(
+                    s,
+                    reference_audio=reference_audio,
+                    emotion=emotion,
+                    sdp_ratio=sdp_ratio,
+                    noise_scale=noise_scale,
+                    noise_scale_w=noise_scale_w,
+                    length_scale=length_scale,
+                    sid=speaker,
+                    language=language,
+                    hps=hps,
+                    net_g=net_g,
+                    device=device,
+                    skip_start=skip_start,
+                    skip_end=skip_end,
+                )
+                audio_list_sent.append(audio)
+                silence = np.zeros((int)(44100 * interval_between_sent))
+                audio_list_sent.append(silence)
+            if (interval_between_para - interval_between_sent) > 0:
+                silence = np.zeros(
+                    (int)(44100 * (interval_between_para - interval_between_sent))
+                )
+                audio_list_sent.append(silence)
+            audio16bit = gr.processing_utils.convert_to_16_bit_wav(
+                np.concatenate(audio_list_sent)
+            )  # 对完整句子做音量归一
+            audio_list.append(audio16bit)
+    audio_concat = np.concatenate(audio_list)
+    return ("Success", (44100, audio_concat))
+
+
+def tts_fn(
+    text: str,
+    speaker,
+    sdp_ratio,
+    noise_scale,
+    noise_scale_w,
+    length_scale,
+    language,
+    reference_audio,
+    emotion,
+):
+    audio_list = []
+    if language == "mix":
+        bool_valid, str_valid = re_matching.validate_text(text)
+        if not bool_valid:
+            return str_valid, (
+                hps.data.sampling_rate,
+                np.concatenate([np.zeros(hps.data.sampling_rate // 2)]),
+            )
+        result = []
+        for slice in re_matching.text_matching(text):
+            _speaker = slice.pop()
+            temp_contant = []
+            temp_lang = []
+            for lang, content in slice:
+                if "|" in content:
+                    temp = []
+                    temp_ = []
+                    for i in content.split("|"):
+                        if i != "":
+                            temp.append([i])
+                            temp_.append([lang])
+                        else:
+                            temp.append([])
+                            temp_.append([])
+                    temp_contant += temp
+                    temp_lang += temp_
+                else:
+                    if len(temp_contant) == 0:
+                        temp_contant.append([])
+                        temp_lang.append([])
+                    temp_contant[-1].append(content)
+                    temp_lang[-1].append(lang)
+            for i, j in zip(temp_lang, temp_contant):
+                result.append([*zip(i, j), _speaker])
+        for i, one in enumerate(result):
+            skip_start = i != 0
+            skip_end = i != len(result) - 1
+            _speaker = one.pop()
+            idx = 0
+            while idx < len(one):
+                text_to_generate = []
+                lang_to_generate = []
+                while True:
+                    lang, content = one[idx]
+                    temp_text = [content]
+                    if len(text_to_generate) > 0:
+                        text_to_generate[-1] += [temp_text.pop(0)]
+                        lang_to_generate[-1] += [lang]
+                    if len(temp_text) > 0:
+                        text_to_generate += [[i] for i in temp_text]
+                        lang_to_generate += [[lang]] * len(temp_text)
+                    if idx + 1 < len(one):
+                        idx += 1
+                    else:
+                        break
+                skip_start = (idx != 0) and skip_start
+                skip_end = (idx != len(one) - 1) and skip_end
+                print(text_to_generate, lang_to_generate)
+                audio_list.extend(
+                    generate_audio_multilang(
+                        text_to_generate,
+                        sdp_ratio,
+                        noise_scale,
+                        noise_scale_w,
+                        length_scale,
+                        speaker,
+                        lang_to_generate,
+                        reference_audio,
+                        emotion,
+                        skip_start,
+                        skip_end,
+                    )
+                )
+                idx += 1
+    elif language.lower() == "auto":
+        for idx, slice in enumerate(text.split("|")):
+            if slice == "":
+                continue
+            skip_start = idx != 0
+            skip_end = idx != len(text.split("|")) - 1
+            sentences_list = split_by_language(
+                slice, target_languages=["zh", "ja", "en"]
+            )
+            idx = 0
+            while idx < len(sentences_list):
+                text_to_generate = []
+                lang_to_generate = []
+                while True:
+                    content, lang = sentences_list[idx]
+                    temp_text = [content]
+                    lang = lang.upper()
+                    if lang == "JA":
+                        lang = "JP"
+                    if len(text_to_generate) > 0:
+                        text_to_generate[-1] += [temp_text.pop(0)]
+                        lang_to_generate[-1] += [lang]
+                    if len(temp_text) > 0:
+                        text_to_generate += [[i] for i in temp_text]
+                        lang_to_generate += [[lang]] * len(temp_text)
+                    if idx + 1 < len(sentences_list):
+                        idx += 1
+                    else:
+                        break
+                skip_start = (idx != 0) and skip_start
+                skip_end = (idx != len(sentences_list) - 1) and skip_end
+                print(text_to_generate, lang_to_generate)
+                audio_list.extend(
+                    generate_audio_multilang(
+                        text_to_generate,
+                        sdp_ratio,
+                        noise_scale,
+                        noise_scale_w,
+                        length_scale,
+                        speaker,
+                        lang_to_generate,
+                        reference_audio,
+                        emotion,
+                        skip_start,
+                        skip_end,
+                    )
+                )
+                idx += 1
+    else:
+        audio_list.extend(
+            generate_audio(
+                text.split("|"),
+                sdp_ratio,
+                noise_scale,
+                noise_scale_w,
+                length_scale,
+                speaker,
+                language,
+                reference_audio,
+                emotion,
+            )
+        )
+
+    audio_concat = np.concatenate(audio_list)
+    return "Success", (hps.data.sampling_rate, audio_concat)
+
+
+if __name__ == "__main__":
+    if config.webui_config.debug:
+        logger.info("Enable DEBUG-LEVEL log")
+        logging.basicConfig(level=logging.DEBUG)
+    hps = utils.get_hparams_from_file(config.webui_config.config_path)
+    # 若config.json中未指定版本则默认为最新版本
+    version = hps.version if hasattr(hps, "version") else latest_version
+    net_g = get_net_g(
+        model_path=config.webui_config.model, version=version, device=device, hps=hps
+    )
+    speaker_ids = hps.data.spk2id
+    speakers = list(speaker_ids.keys())
+    languages = ["ZH", "JP", "EN", "mix", "auto"]
+    with gr.Blocks() as app:
+        with gr.Row():
+            with gr.Column():
+                text = gr.TextArea(
+                    label="输入文本内容",
+                    placeholder="""
+                    如果你选择语言为\'mix\'，必须按照格式输入，否则报错:
+                        格式举例(zh是中文，jp是日语，不区分大小写；说话人举例:gongzi):
+                         [说话人1]<zh>你好，こんにちは！ <jp>こんにちは，世界。
+                         [说话人2]<zh>你好吗？<jp>元気ですか？
+                         [说话人3]<zh>谢谢。<jp>どういたしまして。
+                         ...
+                    另外，所有的语言选项都可以用'|'分割长段实现分句生成。
+                    """,
+                )
+                trans = gr.Button("中翻日", variant="primary")
+                slicer = gr.Button("快速切分", variant="primary")
+                speaker = gr.Dropdown(
+                    choices=speakers, value=speakers[0], label="Speaker"
+                )
+                emotion = gr.Slider(
+                    minimum=0, maximum=9, value=0, step=1, label="Emotion"
+                )
+                sdp_ratio = gr.Slider(
+                    minimum=0, maximum=1, value=0.2, step=0.1, label="SDP Ratio"
+                )
+                noise_scale = gr.Slider(
+                    minimum=0.1, maximum=2, value=0.6, step=0.1, label="Noise"
+                )
+                noise_scale_w = gr.Slider(
+                    minimum=0.1, maximum=2, value=0.8, step=0.1, label="Noise_W"
+                )
+                length_scale = gr.Slider(
+                    minimum=0.1, maximum=2, value=1.0, step=0.1, label="Length"
+                )
+                language = gr.Dropdown(
+                    choices=languages, value=languages[0], label="Language"
+                )
+                btn = gr.Button("生成音频！", variant="primary")
+            with gr.Column():
+                with gr.Row():
+                    with gr.Column():
+                        interval_between_sent = gr.Slider(
+                            minimum=0,
+                            maximum=5,
+                            value=0.2,
+                            step=0.1,
+                            label="句间停顿(秒)，勾选按句切分才生效",
+                        )
+                        interval_between_para = gr.Slider(
+                            minimum=0,
+                            maximum=10,
+                            value=1,
+                            step=0.1,
+                            label="段间停顿(秒)，需要大于句间停顿才有效",
+                        )
+                        opt_cut_by_sent = gr.Checkbox(
+                            label="按句切分    在按段落切分的基础上再按句子切分文本"
+                        )
+                        slicer = gr.Button("切分生成", variant="primary")
+                text_output = gr.Textbox(label="状态信息")
+                audio_output = gr.Audio(label="输出音频")
+                # explain_image = gr.Image(
+                #     label="参数解释信息",
+                #     show_label=True,
+                #     show_share_button=False,
+                #     show_download_button=False,
+                #     value=os.path.abspath("./img/参数说明.png"),
+                # )
+                reference_text = gr.Markdown(value="## 情感参考音频（WAV 格式）：用于生成语音的情感参考。")
+                reference_audio = gr.Audio(label="情感参考音频（WAV 格式）", type="filepath")
+        btn.click(
+            tts_fn,
+            inputs=[
+                text,
+                speaker,
+                sdp_ratio,
+                noise_scale,
+                noise_scale_w,
+                length_scale,
+                language,
+                reference_audio,
+                emotion,
+            ],
+            outputs=[text_output, audio_output],
+        )
+
+        trans.click(
+            translate,
+            inputs=[text],
+            outputs=[text],
+        )
+        slicer.click(
+            tts_split,
+            inputs=[
+                text,
+                speaker,
+                sdp_ratio,
+                noise_scale,
+                noise_scale_w,
+                length_scale,
+                language,
+                opt_cut_by_sent,
+                interval_between_para,
+                interval_between_sent,
+                reference_audio,
+                emotion,
+            ],
+            outputs=[text_output, audio_output],
+        )
+
+        reference_audio.upload(
+            lambda x: librosa.load(x, 16000)[::-1],
+            inputs=[reference_audio],
+            outputs=[reference_audio],
+        )
+    print("推理页面已开启!")
+    webbrowser.open(f"http://127.0.0.1:{config.webui_config.port}")
+    app.launch(share=config.webui_config.share, server_port=config.webui_config.port)

attentions.py

@@ -0,0 +1,464 @@
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+import commons
+import logging
+
+logger = logging.getLogger(__name__)
+
+
+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)
+
+
+@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
+
+
+class Encoder(nn.Module):
+    def __init__(
+        self,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size=1,
+        p_dropout=0.0,
+        window_size=4,
+        isflow=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.window_size = window_size
+        # if isflow:
+        #  cond_layer = torch.nn.Conv1d(256, 2*hidden_channels*n_layers, 1)
+        #  self.cond_pre = torch.nn.Conv1d(hidden_channels, 2*hidden_channels, 1)
+        #  self.cond_layer = weight_norm(cond_layer, name='weight')
+        #  self.gin_channels = 256
+        self.cond_layer_idx = self.n_layers
+        if "gin_channels" in kwargs:
+            self.gin_channels = kwargs["gin_channels"]
+            if self.gin_channels != 0:
+                self.spk_emb_linear = nn.Linear(self.gin_channels, self.hidden_channels)
+                # vits2 says 3rd block, so idx is 2 by default
+                self.cond_layer_idx = (
+                    kwargs["cond_layer_idx"] if "cond_layer_idx" in kwargs else 2
+                )
+                logging.debug(self.gin_channels, self.cond_layer_idx)
+                assert (
+                    self.cond_layer_idx < self.n_layers
+                ), "cond_layer_idx should be less than n_layers"
+        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, g=None):
+        attn_mask = x_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+        x = x * x_mask
+        for i in range(self.n_layers):
+            if i == self.cond_layer_idx and g is not None:
+                g = self.spk_emb_linear(g.transpose(1, 2))
+                g = g.transpose(1, 2)
+                x = x + g
+                x = x * x_mask
+            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=0.0,
+        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=0.0,
+        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):
+        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()
+        # pad 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=0.0,
+        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

bert/bert_models.json

@@ -0,0 +1,14 @@
+{
+    "deberta-v2-large-japanese-char-wwm": {
+        "repo_id": "ku-nlp/deberta-v2-large-japanese-char-wwm",
+        "files": ["pytorch_model.bin"]
+    },
+    "chinese-roberta-wwm-ext-large": {
+        "repo_id": "hfl/chinese-roberta-wwm-ext-large",
+        "files": ["pytorch_model.bin"]
+    },
+    "deberta-v3-large": {
+        "repo_id": "microsoft/deberta-v3-large",
+        "files": ["spm.model", "pytorch_model.bin"]
+    }
+}

commons.py

@@ -0,0 +1,166 @@
+import math
+import torch
+from torch.nn import functional as F
+
+
+def init_weights(m, mean=0.0, std=0.01):
+    classname = m.__class__.__name__
+    if classname.find("Conv") != -1:
+        m.weight.data.normal_(mean, std)
+
+
+def get_padding(kernel_size, dilation=1):
+    return int((kernel_size * dilation - dilation) / 2)
+
+
+def convert_pad_shape(pad_shape):
+    layer = pad_shape[::-1]
+    pad_shape = [item for sublist in layer for item in sublist]
+    return pad_shape
+
+
+def intersperse(lst, item):
+    result = [item] * (len(lst) * 2 + 1)
+    result[1::2] = lst
+    return result
+
+
+def kl_divergence(m_p, logs_p, m_q, logs_q):
+    """KL(P||Q)"""
+    kl = (logs_q - logs_p) - 0.5
+    kl += (
+        0.5 * (torch.exp(2.0 * logs_p) + ((m_p - m_q) ** 2)) * torch.exp(-2.0 * logs_q)
+    )
+    return kl
+
+
+def rand_gumbel(shape):
+    """Sample from the Gumbel distribution, protect from overflows."""
+    uniform_samples = torch.rand(shape) * 0.99998 + 0.00001
+    return -torch.log(-torch.log(uniform_samples))
+
+
+def rand_gumbel_like(x):
+    g = rand_gumbel(x.size()).to(dtype=x.dtype, device=x.device)
+    return g
+
+
+def slice_segments(x, ids_str, segment_size=4):
+    ret = torch.zeros_like(x[:, :, :segment_size])
+    for i in range(x.size(0)):
+        idx_str = ids_str[i]
+        idx_end = idx_str + segment_size
+        if idx_str < 0:
+            i1 = x.size(2) + idx_str
+            r1 = x[i, :, i1:]
+            r2 = x[i, :, :idx_end]
+            ret[i] = torch.cat([r1, r2], dim=1)
+        else:
+            ret[i] = x[i, :, idx_str:idx_end]
+    return ret
+
+
+def rand_slice_segments(x, x_lengths=None, segment_size=4):
+    b, d, t = x.size()
+    if x_lengths is None:
+        x_lengths = t
+    ids_str_max = x_lengths - segment_size + 1
+    ids_str = (torch.rand([b]).to(device=x.device) * ids_str_max).to(dtype=torch.long)
+    ret = slice_segments(x, ids_str, segment_size)
+    return ret, ids_str
+
+
+def get_timing_signal_1d(length, channels, min_timescale=1.0, max_timescale=1.0e4):
+    position = torch.arange(length, dtype=torch.float)
+    num_timescales = channels // 2
+    log_timescale_increment = math.log(float(max_timescale) / float(min_timescale)) / (
+        num_timescales - 1
+    )
+    inv_timescales = min_timescale * torch.exp(
+        torch.arange(num_timescales, dtype=torch.float) * -log_timescale_increment
+    )
+    scaled_time = position.unsqueeze(0) * inv_timescales.unsqueeze(1)
+    signal = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], 0)
+    signal = F.pad(signal, [0, 0, 0, channels % 2])
+    signal = signal.view(1, channels, length)
+    return signal
+
+
+def add_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4):
+    b, channels, length = x.size()
+    signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+    return x + signal.to(dtype=x.dtype, device=x.device)
+
+
+def cat_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4, axis=1):
+    b, channels, length = x.size()
+    signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+    return torch.cat([x, signal.to(dtype=x.dtype, device=x.device)], axis)
+
+
+def subsequent_mask(length):
+    mask = torch.tril(torch.ones(length, length)).unsqueeze(0).unsqueeze(0)
+    return mask
+
+
+@torch.jit.script
+def fused_add_tanh_sigmoid_multiply(input_a, input_b, n_channels):
+    n_channels_int = n_channels[0]
+    in_act = input_a + input_b
+    t_act = torch.tanh(in_act[:, :n_channels_int, :])
+    s_act = torch.sigmoid(in_act[:, n_channels_int:, :])
+    acts = t_act * s_act
+    return acts
+
+
+def convert_pad_shape(pad_shape):
+    layer = pad_shape[::-1]
+    pad_shape = [item for sublist in layer for item in sublist]
+    return pad_shape
+
+
+def shift_1d(x):
+    x = F.pad(x, convert_pad_shape([[0, 0], [0, 0], [1, 0]]))[:, :, :-1]
+    return x
+
+
+def sequence_mask(length, max_length=None):
+    if max_length is None:
+        max_length = length.max()
+    x = torch.arange(max_length, dtype=length.dtype, device=length.device)
+    return x.unsqueeze(0) < length.unsqueeze(1)
+
+
+def generate_path(duration, mask):
+    """
+    duration: [b, 1, t_x]
+    mask: [b, 1, t_y, t_x]
+    """
+
+    b, _, t_y, t_x = mask.shape
+    cum_duration = torch.cumsum(duration, -1)
+
+    cum_duration_flat = cum_duration.view(b * t_x)
+    path = sequence_mask(cum_duration_flat, t_y).to(mask.dtype)
+    path = path.view(b, t_x, t_y)
+    path = path - F.pad(path, convert_pad_shape([[0, 0], [1, 0], [0, 0]]))[:, :-1]
+    path = path.unsqueeze(1).transpose(2, 3) * mask
+    return path
+
+
+def clip_grad_value_(parameters, clip_value, norm_type=2):
+    if isinstance(parameters, torch.Tensor):
+        parameters = [parameters]
+    parameters = list(filter(lambda p: p.grad is not None, parameters))
+    norm_type = float(norm_type)
+    if clip_value is not None:
+        clip_value = float(clip_value)
+
+    total_norm = 0
+    for p in parameters:
+        param_norm = p.grad.data.norm(norm_type)
+        total_norm += param_norm.item() ** norm_type
+        if clip_value is not None:
+            p.grad.data.clamp_(min=-clip_value, max=clip_value)
+    total_norm = total_norm ** (1.0 / norm_type)
+    return total_norm

config.py

@@ -0,0 +1,243 @@
+"""
+@Desc: 全局配置文件读取
+"""
+import argparse
+import yaml
+from typing import Dict, List
+import os
+import shutil
+import sys
+
+
+class Resample_config:
+    """重采样配置"""
+
+    def __init__(self, in_dir: str, out_dir: str, sampling_rate: int = 44100):
+        self.sampling_rate: int = sampling_rate  # 目标采样率
+        self.in_dir: str = in_dir  # 待处理音频目录路径
+        self.out_dir: str = out_dir  # 重采样输出路径
+
+    @classmethod
+    def from_dict(cls, dataset_path: str, data: Dict[str, any]):
+        """从字典中生成实例"""
+
+        # 不检查路径是否有效，此逻辑在resample.py中处理
+        data["in_dir"] = os.path.join(dataset_path, data["in_dir"])
+        data["out_dir"] = os.path.join(dataset_path, data["out_dir"])
+
+        return cls(**data)
+
+
+class Preprocess_text_config:
+    """数据预处理配置"""
+
+    def __init__(
+        self,
+        transcription_path: str,
+        cleaned_path: str,
+        train_path: str,
+        val_path: str,
+        config_path: str,
+        val_per_spk: int = 5,
+        max_val_total: int = 10000,
+        clean: bool = True,
+    ):
+        self.transcription_path: str = transcription_path  # 原始文本文件路径，文本格式应为{wav_path}|{speaker_name}|{language}|{text}。
+        self.cleaned_path: str = cleaned_path  # 数据清洗后文本路径，可以不填。不填则将在原始文本目录生成
+        self.train_path: str = train_path  # 训练集路径，可以不填。不填则将在原始文本目录生成
+        self.val_path: str = val_path  # 验证集路径，可以不填。不填则将在原始文本目录生成
+        self.config_path: str = config_path  # 配置文件路径
+        self.val_per_spk: int = val_per_spk  # 每个speaker的验证集条数
+        self.max_val_total: int = max_val_total  # 验证集最大条数，多于的会被截断并放到训练集中
+        self.clean: bool = clean  # 是否进行数据清洗
+
+    @classmethod
+    def from_dict(cls, dataset_path: str, data: Dict[str, any]):
+        """从字典中生成实例"""
+
+        data["transcription_path"] = os.path.join(
+            dataset_path, data["transcription_path"]
+        )
+        if data["cleaned_path"] == "" or data["cleaned_path"] is None:
+            data["cleaned_path"] = None
+        else:
+            data["cleaned_path"] = os.path.join(dataset_path, data["cleaned_path"])
+        data["train_path"] = os.path.join(dataset_path, data["train_path"])
+        data["val_path"] = os.path.join(dataset_path, data["val_path"])
+        data["config_path"] = os.path.join(dataset_path, data["config_path"])
+
+        return cls(**data)
+
+
+class Bert_gen_config:
+    """bert_gen 配置"""
+
+    def __init__(
+        self,
+        config_path: str,
+        num_processes: int = 2,
+        device: str = "cuda",
+        use_multi_device: bool = False,
+    ):
+        self.config_path = config_path
+        self.num_processes = num_processes
+        self.device = device
+        self.use_multi_device = use_multi_device
+
+    @classmethod
+    def from_dict(cls, dataset_path: str, data: Dict[str, any]):
+        data["config_path"] = os.path.join(dataset_path, data["config_path"])
+
+        return cls(**data)
+
+
+class Emo_gen_config:
+    """emo_gen 配置"""
+
+    def __init__(
+        self,
+        config_path: str,
+        num_processes: int = 2,
+        device: str = "cuda",
+    ):
+        self.config_path = config_path
+        self.num_processes = num_processes
+        self.device = device
+
+    @classmethod
+    def from_dict(cls, dataset_path: str, data: Dict[str, any]):
+        data["config_path"] = os.path.join(dataset_path, data["config_path"])
+
+        return cls(**data)
+
+
+class Train_ms_config:
+    """训练配置"""
+
+    def __init__(
+        self,
+        config_path: str,
+        env: Dict[str, any],
+        base: Dict[str, any],
+        model: str,
+        num_workers: int,
+        spec_cache: bool,
+        keep_ckpts: int,
+    ):
+        self.env = env  # 需要加载的环境变量
+        self.base = base  # 底模配置
+        self.model = model  # 训练模型存储目录，该路径为相对于dataset_path的路径，而非项目根目录
+        self.config_path = config_path  # 配置文件路径
+        self.num_workers = num_workers  # worker数量
+        self.spec_cache = spec_cache  # 是否启用spec缓存
+        self.keep_ckpts = keep_ckpts  # ckpt数量
+
+    @classmethod
+    def from_dict(cls, dataset_path: str, data: Dict[str, any]):
+        # data["model"] = os.path.join(dataset_path, data["model"])
+        data["config_path"] = os.path.join(dataset_path, data["config_path"])
+
+        return cls(**data)
+
+
+class Webui_config:
+    """webui 配置"""
+
+    def __init__(
+        self,
+        device: str,
+        model: str,
+        config_path: str,
+        language_identification_library: str,
+        port: int = 7860,
+        share: bool = False,
+        debug: bool = False,
+    ):
+        self.device: str = device
+        self.model: str = model  # 端口号
+        self.config_path: str = config_path  # 是否公开部署，对外网开放
+        self.port: int = port  # 是否开启debug模式
+        self.share: bool = share  # 模型路径
+        self.debug: bool = debug  # 配置文件路径
+        self.language_identification_library: str = (
+            language_identification_library  # 语种识别库
+        )
+
+    @classmethod
+    def from_dict(cls, dataset_path: str, data: Dict[str, any]):
+        data["config_path"] = os.path.join(dataset_path, data["config_path"])
+        data["model"] = os.path.join(dataset_path, data["model"])
+        return cls(**data)
+
+
+class Server_config:
+    def __init__(
+        self, models: List[Dict[str, any]], port: int = 5000, device: str = "cuda"
+    ):
+        self.models: List[Dict[str, any]] = models  # 需要加载的所有模型的配置
+        self.port: int = port  # 端口号
+        self.device: str = device  # 模型默认使用设备
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, any]):
+        return cls(**data)
+
+
+class Translate_config:
+    """翻译api配置"""
+
+    def __init__(self, app_key: str, secret_key: str):
+        self.app_key = app_key
+        self.secret_key = secret_key
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, any]):
+        return cls(**data)
+
+
+class Config:
+    def __init__(self, config_path: str):
+        if not os.path.isfile(config_path) and os.path.isfile("default_config.yml"):
+            shutil.copy(src="default_config.yml", dst=config_path)
+            print(
+                f"已根据默认配置文件default_config.yml生成配置文件{config_path}。请按该配置文件的说明进行配置后重新运行。"
+            )
+            print("如无特殊需求，请勿修改default_config.yml或备份该文件。")
+            sys.exit(0)
+        with open(file=config_path, mode="r", encoding="utf-8") as file:
+            yaml_config: Dict[str, any] = yaml.safe_load(file.read())
+            dataset_path: str = yaml_config["dataset_path"]
+            openi_token: str = yaml_config["openi_token"]
+            self.dataset_path: str = dataset_path
+            self.mirror: str = yaml_config["mirror"]
+            self.openi_token: str = openi_token
+            self.resample_config: Resample_config = Resample_config.from_dict(
+                dataset_path, yaml_config["resample"]
+            )
+            self.preprocess_text_config: Preprocess_text_config = (
+                Preprocess_text_config.from_dict(
+                    dataset_path, yaml_config["preprocess_text"]
+                )
+            )
+            self.bert_gen_config: Bert_gen_config = Bert_gen_config.from_dict(
+                dataset_path, yaml_config["bert_gen"]
+            )
+            self.train_ms_config: Train_ms_config = Train_ms_config.from_dict(
+                dataset_path, yaml_config["train_ms"]
+            )
+            self.webui_config: Webui_config = Webui_config.from_dict(
+                dataset_path, yaml_config["webui"]
+            )
+            self.server_config: Server_config = Server_config.from_dict(
+                yaml_config["server"]
+            )
+            self.translate_config: Translate_config = Translate_config.from_dict(
+                yaml_config["translate"]
+            )
+
+
+parser = argparse.ArgumentParser()
+# 为避免与以前的config.json起冲突，将其更名如下
+parser.add_argument("-y", "--yml_config", type=str, default="config.yml")
+args, _ = parser.parse_known_args()
+config = Config(args.yml_config)

config.yml

@@ -0,0 +1,174 @@
+# 全局配置
+# 对于希望在同一时间使用多个配置文件的情况，例如两个GPU同时跑两个训练集：通过环境变量指定配置文件，不指定则默认为./config.yml
+
+# 拟提供通用路径配置，统一存放数据，避免数据放得很乱
+# 每个数据集与其对应的模型存放至统一路径下，后续所有的路径配置均为相对于datasetPath的路径
+# 不填或者填空则路径为相对于项目根目录的路径
+dataset_path: "Data/"
+
+# 模型镜像源，默认huggingface，使用openi镜像源需指定openi_token
+mirror: ""
+openi_token: ""  # openi token
+
+# resample 音频重采样配置
+# 注意， “:” 后需要加空格
+resample:
+  # 目标重采样率
+  sampling_rate: 44100
+  # 音频文件输入路径，重采样会将该路径下所有.wav音频文件重采样
+  # 请填入相对于datasetPath的相对路径
+  in_dir: "audios/raw" # 相对于根目录的路径为 /datasetPath/in_dir
+  # 音频文件重采样后输出路径
+  out_dir: "audios/wavs"
+
+
+# preprocess_text 数据集预处理相关配置
+# 注意， “:” 后需要加空格
+preprocess_text:
+  # 原始文本文件路径，文本格式应为{wav_path}|{speaker_name}|{language}|{text}。
+  transcription_path: "filelists/haruka.list"
+  # 数据清洗后文本路径，可以不填。不填则将在原始文本目录生成
+  cleaned_path: ""
+  # 训练集路径
+  train_path: "filelists/train.list"
+  # 验证集路径
+  val_path: "filelists/val.list"
+  # 配置文件路径
+  config_path: "haruka.json"
+  # 每个speaker的验证集条数
+  val_per_spk:  40
+  # 验证集最大条数，多于的会被截断并放到训练集中
+  max_val_total: 200
+  # 是否进行数据清洗
+  clean: true
+
+
+# bert_gen 相关配置
+# 注意， “:” 后需要加空格
+bert_gen:
+  # 训练数据集配置文件路径
+  config_path: "haruka.json"
+  # 并行数
+  num_processes: 2
+  # 使用设备：可选项 "cuda" 显卡推理，"cpu" cpu推理
+  # 该选项同时决定了get_bert_feature的默认设备
+  device: "cuda"
+  # 使用多卡推理
+  use_multi_device: false
+
+# emo_gen 相关配置
+# 注意， “:” 后需要加空格
+emo_gen:
+  # 训练数据集配置文件路径
+  config_path: "haruka.json"
+  # 并行数
+  num_processes: 2
+  # 使用设备：可选项 "cuda" 显卡推理，"cpu" cpu推理
+  device: "cuda"
+
+# train 训练配置
+# 注意， “:” 后需要加空格
+train_ms:
+  env:
+    MASTER_ADDR: "localhost"
+    MASTER_PORT: 10086
+    WORLD_SIZE: 1
+    LOCAL_RANK: 0
+    RANK: 0
+    # 可以填写任意名的环境变量
+    # THE_ENV_VAR_YOU_NEED_TO_USE: "1234567"
+  # 底模设置
+  base:
+    use_base_model: false
+    repo_id: "Stardust_minus/Bert-VITS2"
+    model_image: "Bert-VITS2_2.1-Emo底模" # openi网页的模型名
+  # 训练模型存储目录：与旧版本的区别，原先数据集是存放在logs/model_name下的，现在改为统一存放在Data/你的数据集/models下
+  model: "models"
+  # 配置文件路径
+  config_path: "haruka.json"
+  # 训练使用的worker，不建议超过CPU核心数
+  num_workers: 16
+  # 关闭此项可以节约接近50%的磁盘空间，但是可能导致实际训练速度变慢和更高的CPU使用率。
+  spec_cache: True
+  # 保存的检查点数量，多于此数目的权重会被删除来节省空间。
+  keep_ckpts: 8
+
+
+# webui webui配置
+# 注意， “:” 后需要加空格
+webui:
+  # 推理设备
+  device: "cuda"
+  # 模型路径
+  model: "models/G_82000.pth"
+  # 配置文件路径
+  config_path: "configs/haruka.json"
+  # 端口号
+  port: 7860
+  # 是否公开部署，对外网开放
+  share: false
+  # 是否开启debug模式
+  debug: false
+  # 语种识别库，可选langid, fastlid
+  language_identification_library: "langid"
+
+
+# server api配置
+# 注意， “:” 后需要加空格
+# 注意，本配置下的所有配置均为相对于根目录的路径
+server:
+  # 端口号
+  port: 5000
+  # 模型默认使用设备：但是当前并没有实现这个配置。
+  device: "cuda"
+  # 需要加载的所有模型的配置
+  # 注意，所有模型都必须正确配置model与config的路径，空路径会导致加载错误。
+  models:
+    - # 模型的路径
+      model: ""
+      # 模型config.json的路径
+      config: ""
+      # 模型使用设备，若填写则会覆盖默认配置
+      device: "cuda"
+      # 模型默认使用的语言
+      language: "ZH"
+      # 模型人物默认参数
+      # 不必填写所有人物，不填的使用默认值
+      # 暂时不用填写，当前尚未实现按人区分配置
+      speakers:
+        - speaker: "科比"
+          sdp_ratio: 0.2
+          noise_scale: 0.6
+          noise_scale_w: 0.8
+          length_scale: 1
+        - speaker: "五条悟"
+          sdp_ratio: 0.3
+          noise_scale: 0.7
+          noise_scale_w: 0.8
+          length_scale: 0.5
+        - speaker: "安倍晋三"
+          sdp_ratio: 0.2
+          noise_scale: 0.6
+          noise_scale_w: 0.8
+          length_scale: 1.2
+    - # 模型的路径
+      model: ""
+      # 模型config.json的路径
+      config: ""
+      # 模型使用设备，若填写则会覆盖默认配置
+      device: "cpu"
+      # 模型默认使用的语言
+      language: "JP"
+      # 模型人物默认参数
+      # 不必填写所有人物，不填的使用默认值
+      speakers: [ ] # 也可以不填
+
+
+# 百度翻译开放平台 api配置
+# api接入文档 https://api.fanyi.baidu.com/doc/21
+# 请不要在github等网站公开分享你的app id 与 key
+translate:
+  # 你的APPID
+  "app_key": ""
+  # 你的密钥
+  "secret_key": ""

emo_gen.py

@@ -0,0 +1,162 @@
+import argparse
+import os
+from pathlib import Path
+
+import librosa
+import numpy as np
+import torch
+import torch.nn as nn
+from torch.utils.data import DataLoader, Dataset
+from tqdm import tqdm
+from transformers import Wav2Vec2Processor
+from transformers.models.wav2vec2.modeling_wav2vec2 import (
+    Wav2Vec2Model,
+    Wav2Vec2PreTrainedModel,
+)
+
+import utils
+from config import config
+
+
+class RegressionHead(nn.Module):
+    r"""Classification head."""
+
+    def __init__(self, config):
+        super().__init__()
+
+        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
+        self.dropout = nn.Dropout(config.final_dropout)
+        self.out_proj = nn.Linear(config.hidden_size, config.num_labels)
+
+    def forward(self, features, **kwargs):
+        x = features
+        x = self.dropout(x)
+        x = self.dense(x)
+        x = torch.tanh(x)
+        x = self.dropout(x)
+        x = self.out_proj(x)
+
+        return x
+
+
+class EmotionModel(Wav2Vec2PreTrainedModel):
+    r"""Speech emotion classifier."""
+
+    def __init__(self, config):
+        super().__init__(config)
+
+        self.config = config
+        self.wav2vec2 = Wav2Vec2Model(config)
+        self.classifier = RegressionHead(config)
+        self.init_weights()
+
+    def forward(
+        self,
+        input_values,
+    ):
+        outputs = self.wav2vec2(input_values)
+        hidden_states = outputs[0]
+        hidden_states = torch.mean(hidden_states, dim=1)
+        logits = self.classifier(hidden_states)
+
+        return hidden_states, logits
+
+
+class AudioDataset(Dataset):
+    def __init__(self, list_of_wav_files, sr, processor):
+        self.list_of_wav_files = list_of_wav_files
+        self.processor = processor
+        self.sr = sr
+
+    def __len__(self):
+        return len(self.list_of_wav_files)
+
+    def __getitem__(self, idx):
+        wav_file = self.list_of_wav_files[idx]
+        audio_data, _ = librosa.load(wav_file, sr=self.sr)
+        processed_data = self.processor(audio_data, sampling_rate=self.sr)[
+            "input_values"
+        ][0]
+        return torch.from_numpy(processed_data)
+
+
+def process_func(
+    x: np.ndarray,
+    sampling_rate: int,
+    model: EmotionModel,
+    processor: Wav2Vec2Processor,
+    device: str,
+    embeddings: bool = False,
+) -> np.ndarray:
+    r"""Predict emotions or extract embeddings from raw audio signal."""
+    model = model.to(device)
+    y = processor(x, sampling_rate=sampling_rate)
+    y = y["input_values"][0]
+    y = torch.from_numpy(y).unsqueeze(0).to(device)
+
+    # run through model
+    with torch.no_grad():
+        y = model(y)[0 if embeddings else 1]
+
+    # convert to numpy
+    y = y.detach().cpu().numpy()
+
+    return y
+
+
+def get_emo(path):
+    wav, sr = librosa.load(path, 16000)
+    device = config.bert_gen_config.device
+    return process_func(
+        np.expand_dims(wav, 0).astype(np.float64),
+        sr,
+        model,
+        processor,
+        device,
+        embeddings=True,
+    ).squeeze(0)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "-c", "--config", type=str, default=config.bert_gen_config.config_path
+    )
+    parser.add_argument(
+        "--num_processes", type=int, default=config.bert_gen_config.num_processes
+    )
+    args, _ = parser.parse_known_args()
+    config_path = args.config
+    hps = utils.get_hparams_from_file(config_path)
+
+    device = config.bert_gen_config.device
+
+    model_name = "./emotional/wav2vec2-large-robust-12-ft-emotion-msp-dim"
+    REPO_ID = "audeering/wav2vec2-large-robust-12-ft-emotion-msp-dim"
+    if not Path(model_name).joinpath("pytorch_model.bin").exists():
+        utils.download_emo_models(config.mirror, model_name, REPO_ID)
+
+    processor = Wav2Vec2Processor.from_pretrained(model_name)
+    model = EmotionModel.from_pretrained(model_name).to(device)
+
+    lines = []
+    with open(hps.data.training_files, encoding="utf-8") as f:
+        lines.extend(f.readlines())
+
+    with open(hps.data.validation_files, encoding="utf-8") as f:
+        lines.extend(f.readlines())
+
+    wavnames = [line.split("|")[0] for line in lines]
+    dataset = AudioDataset(wavnames, 16000, processor)
+    data_loader = DataLoader(dataset, batch_size=1, shuffle=False, num_workers=16)
+
+    with torch.no_grad():
+        for i, data in tqdm(enumerate(data_loader), total=len(data_loader)):
+            wavname = wavnames[i]
+            emo_path = wavname.replace(".wav", ".emo.npy")
+            if os.path.exists(emo_path):
+                continue
+            emb = model(data.to(device))[0].detach().cpu().numpy()
+            np.save(emo_path, emb)
+
+    print("Emo vec 生成完毕!")

infer.py

@@ -0,0 +1,261 @@
+"""
+版本管理、兼容推理及模型加载实现。
+版本说明：
+    1. 版本号与github的release版本号对应，使用哪个release版本训练的模型即对应其版本号
+    2. 请在模型的config.json中显示声明版本号，添加一个字段"version" : "你的版本号"
+特殊版本说明：
+    1.1.1-fix： 1.1.1版本训练的模型，但是在推理时使用dev的日语修复
+    1.1.1-dev： dev开发
+    2.1：当前版本
+"""
+import torch
+import commons
+from text import cleaned_text_to_sequence, get_bert
+from emo_gen import get_emo
+from text.cleaner import clean_text
+import utils
+
+from models import SynthesizerTrn
+from text.symbols import symbols
+
+# 当前版本信息
+latest_version = "2.1"
+
+
+
+def get_net_g(model_path: str, version: str, device: str, hps):
+    if version != latest_version:
+        pass
+    else:
+        # 当前版本模型 net_g
+        net_g = SynthesizerTrn(
+            len(symbols),
+            hps.data.filter_length // 2 + 1,
+            hps.train.segment_size // hps.data.hop_length,
+            n_speakers=hps.data.n_speakers,
+            **hps.model,
+        ).to(device)
+    _ = net_g.eval()
+    _ = utils.load_checkpoint(model_path, net_g, None, skip_optimizer=True)
+    return net_g
+
+
+def get_text(text, reference_audio, emotion, language_str, hps, device):
+    # 在此处实现当前版本的get_text
+    norm_text, phone, tone, word2ph = clean_text(text, language_str)
+    phone, tone, language = cleaned_text_to_sequence(phone, tone, language_str)
+
+    if hps.data.add_blank:
+        phone = commons.intersperse(phone, 0)
+        tone = commons.intersperse(tone, 0)
+        language = commons.intersperse(language, 0)
+        for i in range(len(word2ph)):
+            word2ph[i] = word2ph[i] * 2
+        word2ph[0] += 1
+    bert_ori = get_bert(norm_text, word2ph, language_str, device)
+    del word2ph
+    assert bert_ori.shape[-1] == len(phone), phone
+
+    if language_str == "ZH":
+        bert = bert_ori
+        ja_bert = torch.zeros(1024, len(phone))
+        en_bert = torch.zeros(1024, len(phone))
+    elif language_str == "JP":
+        bert = torch.zeros(1024, len(phone))
+        ja_bert = bert_ori
+        en_bert = torch.zeros(1024, len(phone))
+    elif language_str == "EN":
+        bert = torch.zeros(1024, len(phone))
+        ja_bert = torch.zeros(1024, len(phone))
+        en_bert = bert_ori
+    else:
+        raise ValueError("language_str should be ZH, JP or EN")
+
+    emo = (
+        torch.from_numpy(get_emo(reference_audio))
+        if reference_audio
+        else torch.Tensor([emotion])
+    )
+
+    assert bert.shape[-1] == len(
+        phone
+    ), f"Bert seq len {bert.shape[-1]} != {len(phone)}"
+
+    phone = torch.LongTensor(phone)
+    tone = torch.LongTensor(tone)
+    language = torch.LongTensor(language)
+    return bert, ja_bert, en_bert, emo, phone, tone, language
+
+
+def infer(
+    text,
+    sdp_ratio,
+    noise_scale,
+    noise_scale_w,
+    length_scale,
+    sid,
+    language,
+    hps,
+    net_g,
+    device,
+    reference_audio=None,
+    emotion=None,
+    skip_start=False,
+    skip_end=False,
+):
+    version = hps.version if hasattr(hps, "version") else latest_version
+    # 非当前版本，根据版本号选择合适的infer
+    if version != latest_version:
+        pass
+    # 在此处实现当前版本的推理
+    bert, ja_bert, en_bert, emo, phones, tones, lang_ids = get_text(
+        text, reference_audio, emotion, language, hps, device
+    )
+    if skip_start:
+        phones = phones[1:]
+        tones = tones[1:]
+        lang_ids = lang_ids[1:]
+        bert = bert[:, 1:]
+        ja_bert = ja_bert[:, 1:]
+        en_bert = en_bert[:, 1:]
+    if skip_end:
+        phones = phones[:-1]
+        tones = tones[:-1]
+        lang_ids = lang_ids[:-1]
+        bert = bert[:, :-1]
+        ja_bert = ja_bert[:, :-1]
+        en_bert = en_bert[:, :-1]
+    with torch.no_grad():
+        x_tst = phones.to(device).unsqueeze(0)
+        tones = tones.to(device).unsqueeze(0)
+        lang_ids = lang_ids.to(device).unsqueeze(0)
+        bert = bert.to(device).unsqueeze(0)
+        ja_bert = ja_bert.to(device).unsqueeze(0)
+        en_bert = en_bert.to(device).unsqueeze(0)
+        x_tst_lengths = torch.LongTensor([phones.size(0)]).to(device)
+        emo = emo.to(device).unsqueeze(0)
+        del phones
+        speakers = torch.LongTensor([hps.data.spk2id[sid]]).to(device)
+        audio = (
+            net_g.infer(
+                x_tst,
+                x_tst_lengths,
+                speakers,
+                tones,
+                lang_ids,
+                bert,
+                ja_bert,
+                en_bert,
+                emo,
+                sdp_ratio=sdp_ratio,
+                noise_scale=noise_scale,
+                noise_scale_w=noise_scale_w,
+                length_scale=length_scale,
+            )[0][0, 0]
+            .data.cpu()
+            .float()
+            .numpy()
+        )
+        del x_tst, tones, lang_ids, bert, x_tst_lengths, speakers, ja_bert, en_bert, emo
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+        return audio
+
+
+def infer_multilang(
+    text,
+    sdp_ratio,
+    noise_scale,
+    noise_scale_w,
+    length_scale,
+    sid,
+    language,
+    hps,
+    net_g,
+    device,
+    reference_audio=None,
+    emotion=None,
+    skip_start=False,
+    skip_end=False,
+):
+    bert, ja_bert, en_bert, emo, phones, tones, lang_ids = [], [], [], [], [], [], []
+    # bert, ja_bert, en_bert, phones, tones, lang_ids = get_text(
+    #     text, language, hps, device
+    # )
+    for idx, (txt, lang) in enumerate(zip(text, language)):
+        skip_start = (idx != 0) or (skip_start and idx == 0)
+        skip_end = (idx != len(text) - 1) or (skip_end and idx == len(text) - 1)
+        (
+            temp_bert,
+            temp_ja_bert,
+            temp_en_bert,
+            temp_emo,
+            temp_phones,
+            temp_tones,
+            temp_lang_ids,
+        ) = get_text(txt, reference_audio, emotion, language, hps, device)
+        if skip_start:
+            temp_bert = temp_bert[:, 1:]
+            temp_ja_bert = temp_ja_bert[:, 1:]
+            temp_en_bert = temp_en_bert[:, 1:]
+            temp_emo = temp_emo[:, 1:]
+            temp_phones = temp_phones[1:]
+            temp_tones = temp_tones[1:]
+            temp_lang_ids = temp_lang_ids[1:]
+        if skip_end:
+            temp_bert = temp_bert[:, :-1]
+            temp_ja_bert = temp_ja_bert[:, :-1]
+            temp_en_bert = temp_en_bert[:, :-1]
+            temp_emo = temp_emo[:, :-1]
+            temp_phones = temp_phones[:-1]
+            temp_tones = temp_tones[:-1]
+            temp_lang_ids = temp_lang_ids[:-1]
+        bert.append(temp_bert)
+        ja_bert.append(temp_ja_bert)
+        en_bert.append(temp_en_bert)
+        emo.append(temp_emo)
+        phones.append(temp_phones)
+        tones.append(temp_tones)
+        lang_ids.append(temp_lang_ids)
+    bert = torch.concatenate(bert, dim=1)
+    ja_bert = torch.concatenate(ja_bert, dim=1)
+    en_bert = torch.concatenate(en_bert, dim=1)
+    emo = torch.concatenate(emo, dim=1)
+    phones = torch.concatenate(phones, dim=0)
+    tones = torch.concatenate(tones, dim=0)
+    lang_ids = torch.concatenate(lang_ids, dim=0)
+    with torch.no_grad():
+        x_tst = phones.to(device).unsqueeze(0)
+        tones = tones.to(device).unsqueeze(0)
+        lang_ids = lang_ids.to(device).unsqueeze(0)
+        bert = bert.to(device).unsqueeze(0)
+        ja_bert = ja_bert.to(device).unsqueeze(0)
+        en_bert = en_bert.to(device).unsqueeze(0)
+        emo = emo.to(device).unsqueeze(0)
+        x_tst_lengths = torch.LongTensor([phones.size(0)]).to(device)
+        del phones
+        speakers = torch.LongTensor([hps.data.spk2id[sid]]).to(device)
+        audio = (
+            net_g.infer(
+                x_tst,
+                x_tst_lengths,
+                speakers,
+                tones,
+                lang_ids,
+                bert,
+                ja_bert,
+                en_bert,
+                emo,
+                sdp_ratio=sdp_ratio,
+                noise_scale=noise_scale,
+                noise_scale_w=noise_scale_w,
+                length_scale=length_scale,
+            )[0][0, 0]
+            .data.cpu()
+            .float()
+            .numpy()
+        )
+        del x_tst, tones, lang_ids, bert, x_tst_lengths, speakers, ja_bert, en_bert, emo
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+        return audio

models.py

@@ -0,0 +1,1044 @@
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+import commons
+import modules
+import attentions
+import monotonic_align
+
+from torch.nn import Conv1d, ConvTranspose1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm, spectral_norm
+from vector_quantize_pytorch import VectorQuantize
+
+from commons import init_weights, get_padding
+from text import symbols, num_tones, num_languages
+
+
+class DurationDiscriminator(nn.Module):  # vits2
+    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.dur_proj = nn.Conv1d(1, filter_channels, 1)
+
+        self.pre_out_conv_1 = nn.Conv1d(
+            2 * filter_channels, filter_channels, kernel_size, padding=kernel_size // 2
+        )
+        self.pre_out_norm_1 = modules.LayerNorm(filter_channels)
+        self.pre_out_conv_2 = nn.Conv1d(
+            filter_channels, filter_channels, kernel_size, padding=kernel_size // 2
+        )
+        self.pre_out_norm_2 = modules.LayerNorm(filter_channels)
+
+        if gin_channels != 0:
+            self.cond = nn.Conv1d(gin_channels, in_channels, 1)
+
+        self.output_layer = nn.Sequential(nn.Linear(filter_channels, 1), nn.Sigmoid())
+
+    def forward_probability(self, x, x_mask, dur, g=None):
+        dur = self.dur_proj(dur)
+        x = torch.cat([x, dur], dim=1)
+        x = self.pre_out_conv_1(x * x_mask)
+        x = torch.relu(x)
+        x = self.pre_out_norm_1(x)
+        x = self.drop(x)
+        x = self.pre_out_conv_2(x * x_mask)
+        x = torch.relu(x)
+        x = self.pre_out_norm_2(x)
+        x = self.drop(x)
+        x = x * x_mask
+        x = x.transpose(1, 2)
+        output_prob = self.output_layer(x)
+        return output_prob
+
+    def forward(self, x, x_mask, dur_r, dur_hat, 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)
+
+        output_probs = []
+        for dur in [dur_r, dur_hat]:
+            output_prob = self.forward_probability(x, x_mask, dur, g)
+            output_probs.append(output_prob)
+
+        return output_probs
+
+
+class TransformerCouplingBlock(nn.Module):
+    def __init__(
+        self,
+        channels,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size,
+        p_dropout,
+        n_flows=4,
+        gin_channels=0,
+        share_parameter=False,
+    ):
+        super().__init__()
+        self.channels = channels
+        self.hidden_channels = hidden_channels
+        self.kernel_size = kernel_size
+        self.n_layers = n_layers
+        self.n_flows = n_flows
+        self.gin_channels = gin_channels
+
+        self.flows = nn.ModuleList()
+
+        self.wn = (
+            attentions.FFT(
+                hidden_channels,
+                filter_channels,
+                n_heads,
+                n_layers,
+                kernel_size,
+                p_dropout,
+                isflow=True,
+                gin_channels=self.gin_channels,
+            )
+            if share_parameter
+            else None
+        )
+
+        for i in range(n_flows):
+            self.flows.append(
+                modules.TransformerCouplingLayer(
+                    channels,
+                    hidden_channels,
+                    kernel_size,
+                    n_layers,
+                    n_heads,
+                    p_dropout,
+                    filter_channels,
+                    mean_only=True,
+                    wn_sharing_parameter=self.wn,
+                    gin_channels=self.gin_channels,
+                )
+            )
+            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 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,
+        n_speakers,
+        gin_channels=0,
+    ):
+        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.gin_channels = gin_channels
+        self.emb = nn.Embedding(len(symbols), hidden_channels)
+        nn.init.normal_(self.emb.weight, 0.0, hidden_channels**-0.5)
+        self.tone_emb = nn.Embedding(num_tones, hidden_channels)
+        nn.init.normal_(self.tone_emb.weight, 0.0, hidden_channels**-0.5)
+        self.language_emb = nn.Embedding(num_languages, hidden_channels)
+        nn.init.normal_(self.language_emb.weight, 0.0, hidden_channels**-0.5)
+        self.bert_proj = nn.Conv1d(1024, hidden_channels, 1)
+        self.ja_bert_proj = nn.Conv1d(1024, hidden_channels, 1)
+        self.en_bert_proj = nn.Conv1d(1024, hidden_channels, 1)
+        self.emo_proj = nn.Linear(1024, 1024)
+        self.emo_quantizer = [
+            VectorQuantize(
+                dim=1024,
+                codebook_size=10,
+                decay=0.8,
+                commitment_weight=1.0,
+                learnable_codebook=True,
+                ema_update=False,
+            )
+        ] * n_speakers
+        self.emo_q_proj = nn.Linear(1024, hidden_channels)
+
+        self.encoder = attentions.Encoder(
+            hidden_channels,
+            filter_channels,
+            n_heads,
+            n_layers,
+            kernel_size,
+            p_dropout,
+            gin_channels=self.gin_channels,
+        )
+        self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+    def forward(
+        self, x, x_lengths, tone, language, bert, ja_bert, en_bert, emo, sid, g=None
+    ):
+        sid = sid.cpu()
+        bert_emb = self.bert_proj(bert).transpose(1, 2)
+        ja_bert_emb = self.ja_bert_proj(ja_bert).transpose(1, 2)
+        en_bert_emb = self.en_bert_proj(en_bert).transpose(1, 2)
+        if emo.size(-1) == 1024:
+            emo_emb = self.emo_proj(emo.unsqueeze(1))
+            emo_commit_loss = torch.zeros(1)
+            emo_emb_ = []
+            for i in range(emo_emb.size(0)):
+                temp_emo_emb, _, temp_emo_commit_loss = self.emo_quantizer[sid[i]](
+                    emo_emb[i].unsqueeze(0).cpu()
+                )
+                emo_commit_loss += temp_emo_commit_loss
+                emo_emb_.append(temp_emo_emb)
+            emo_emb = torch.cat(emo_emb_, dim=0).to(emo_emb.device)
+            emo_commit_loss = emo_commit_loss.to(emo_emb.device)
+        else:
+            emo_emb = (
+                self.emo_quantizer[sid[0]]
+                .get_output_from_indices(emo.to(torch.int).cpu())
+                .unsqueeze(0)
+                .to(emo.device)
+            )
+            emo_commit_loss = torch.zeros(1)
+        x = (
+            self.emb(x)
+            + self.tone_emb(tone)
+            + self.language_emb(language)
+            + bert_emb
+            + ja_bert_emb
+            + en_bert_emb
+            + self.emo_q_proj(emo_emb)
+        ) * math.sqrt(
+            self.hidden_channels
+        )  # [b, t, h]
+        x = torch.transpose(x, 1, -1)  # [b, h, t]
+        x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(
+            x.dtype
+        )
+
+        x = self.encoder(x * x_mask, x_mask, g=g)
+        stats = self.proj(x) * x_mask
+
+        m, logs = torch.split(stats, self.out_channels, dim=1)
+        return x, m, logs, x_mask, emo_commit_loss
+
+
+class ResidualCouplingBlock(nn.Module):
+    def __init__(
+        self,
+        channels,
+        hidden_channels,
+        kernel_size,
+        dilation_rate,
+        n_layers,
+        n_flows=4,
+        gin_channels=0,
+    ):
+        super().__init__()
+        self.channels = channels
+        self.hidden_channels = hidden_channels
+        self.kernel_size = kernel_size
+        self.dilation_rate = dilation_rate
+        self.n_layers = n_layers
+        self.n_flows = n_flows
+        self.gin_channels = gin_channels
+
+        self.flows = nn.ModuleList()
+        for i in range(n_flows):
+            self.flows.append(
+                modules.ResidualCouplingLayer(
+                    channels,
+                    hidden_channels,
+                    kernel_size,
+                    dilation_rate,
+                    n_layers,
+                    gin_channels=gin_channels,
+                    mean_only=True,
+                )
+            )
+            self.flows.append(modules.Flip())
+
+    def forward(self, x, x_mask, g=None, reverse=False):
+        if not reverse:
+            for flow in self.flows:
+                x, _ = flow(x, x_mask, g=g, reverse=reverse)
+        else:
+            for flow in reversed(self.flows):
+                x = flow(x, x_mask, g=g, reverse=reverse)
+        return x
+
+
+class PosteriorEncoder(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        hidden_channels,
+        kernel_size,
+        dilation_rate,
+        n_layers,
+        gin_channels=0,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.hidden_channels = hidden_channels
+        self.kernel_size = kernel_size
+        self.dilation_rate = dilation_rate
+        self.n_layers = n_layers
+        self.gin_channels = gin_channels
+
+        self.pre = nn.Conv1d(in_channels, hidden_channels, 1)
+        self.enc = modules.WN(
+            hidden_channels,
+            kernel_size,
+            dilation_rate,
+            n_layers,
+            gin_channels=gin_channels,
+        )
+        self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+    def forward(self, x, x_lengths, g=None):
+        x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(
+            x.dtype
+        )
+        x = self.pre(x) * x_mask
+        x = self.enc(x, x_mask, g=g)
+        stats = self.proj(x) * x_mask
+        m, logs = torch.split(stats, self.out_channels, dim=1)
+        z = (m + torch.randn_like(m) * torch.exp(logs)) * x_mask
+        return z, m, logs, x_mask
+
+
+class Generator(torch.nn.Module):
+    def __init__(
+        self,
+        initial_channel,
+        resblock,
+        resblock_kernel_sizes,
+        resblock_dilation_sizes,
+        upsample_rates,
+        upsample_initial_channel,
+        upsample_kernel_sizes,
+        gin_channels=0,
+    ):
+        super(Generator, self).__init__()
+        self.num_kernels = len(resblock_kernel_sizes)
+        self.num_upsamples = len(upsample_rates)
+        self.conv_pre = Conv1d(
+            initial_channel, upsample_initial_channel, 7, 1, padding=3
+        )
+        resblock = modules.ResBlock1 if resblock == "1" else modules.ResBlock2
+
+        self.ups = nn.ModuleList()
+        for i, (u, k) in enumerate(zip(upsample_rates, upsample_kernel_sizes)):
+            self.ups.append(
+                weight_norm(
+                    ConvTranspose1d(
+                        upsample_initial_channel // (2**i),
+                        upsample_initial_channel // (2 ** (i + 1)),
+                        k,
+                        u,
+                        padding=(k - u) // 2,
+                    )
+                )
+            )
+
+        self.resblocks = nn.ModuleList()
+        for i in range(len(self.ups)):
+            ch = upsample_initial_channel // (2 ** (i + 1))
+            for j, (k, d) in enumerate(
+                zip(resblock_kernel_sizes, resblock_dilation_sizes)
+            ):
+                self.resblocks.append(resblock(ch, k, d))
+
+        self.conv_post = Conv1d(ch, 1, 7, 1, padding=3, bias=False)
+        self.ups.apply(init_weights)
+
+        if gin_channels != 0:
+            self.cond = nn.Conv1d(gin_channels, upsample_initial_channel, 1)
+
+    def forward(self, x, g=None):
+        x = self.conv_pre(x)
+        if g is not None:
+            x = x + self.cond(g)
+
+        for i in range(self.num_upsamples):
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            x = self.ups[i](x)
+            xs = None
+            for j in range(self.num_kernels):
+                if xs is None:
+                    xs = self.resblocks[i * self.num_kernels + j](x)
+                else:
+                    xs += self.resblocks[i * self.num_kernels + j](x)
+            x = xs / self.num_kernels
+        x = F.leaky_relu(x)
+        x = self.conv_post(x)
+        x = torch.tanh(x)
+
+        return x
+
+    def remove_weight_norm(self):
+        print("Removing weight norm...")
+        for layer in self.ups:
+            remove_weight_norm(layer)
+        for layer in self.resblocks:
+            layer.remove_weight_norm()
+
+
+class DiscriminatorP(torch.nn.Module):
+    def __init__(self, period, kernel_size=5, stride=3, use_spectral_norm=False):
+        super(DiscriminatorP, self).__init__()
+        self.period = period
+        self.use_spectral_norm = use_spectral_norm
+        norm_f = weight_norm if use_spectral_norm is False else spectral_norm
+        self.convs = nn.ModuleList(
+            [
+                norm_f(
+                    Conv2d(
+                        1,
+                        32,
+                        (kernel_size, 1),
+                        (stride, 1),
+                        padding=(get_padding(kernel_size, 1), 0),
+                    )
+                ),
+                norm_f(
+                    Conv2d(
+                        32,
+                        128,
+                        (kernel_size, 1),
+                        (stride, 1),
+                        padding=(get_padding(kernel_size, 1), 0),
+                    )
+                ),
+                norm_f(
+                    Conv2d(
+                        128,
+                        512,
+                        (kernel_size, 1),
+                        (stride, 1),
+                        padding=(get_padding(kernel_size, 1), 0),
+                    )
+                ),
+                norm_f(
+                    Conv2d(
+                        512,
+                        1024,
+                        (kernel_size, 1),
+                        (stride, 1),
+                        padding=(get_padding(kernel_size, 1), 0),
+                    )
+                ),
+                norm_f(
+                    Conv2d(
+                        1024,
+                        1024,
+                        (kernel_size, 1),
+                        1,
+                        padding=(get_padding(kernel_size, 1), 0),
+                    )
+                ),
+            ]
+        )
+        self.conv_post = norm_f(Conv2d(1024, 1, (3, 1), 1, padding=(1, 0)))
+
+    def forward(self, x):
+        fmap = []
+
+        # 1d to 2d
+        b, c, t = x.shape
+        if t % self.period != 0:  # pad first
+            n_pad = self.period - (t % self.period)
+            x = F.pad(x, (0, n_pad), "reflect")
+            t = t + n_pad
+        x = x.view(b, c, t // self.period, self.period)
+
+        for layer in self.convs:
+            x = layer(x)
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            fmap.append(x)
+        x = self.conv_post(x)
+        fmap.append(x)
+        x = torch.flatten(x, 1, -1)
+
+        return x, fmap
+
+
+class DiscriminatorS(torch.nn.Module):
+    def __init__(self, use_spectral_norm=False):
+        super(DiscriminatorS, self).__init__()
+        norm_f = weight_norm if use_spectral_norm is False else spectral_norm
+        self.convs = nn.ModuleList(
+            [
+                norm_f(Conv1d(1, 16, 15, 1, padding=7)),
+                norm_f(Conv1d(16, 64, 41, 4, groups=4, padding=20)),
+                norm_f(Conv1d(64, 256, 41, 4, groups=16, padding=20)),
+                norm_f(Conv1d(256, 1024, 41, 4, groups=64, padding=20)),
+                norm_f(Conv1d(1024, 1024, 41, 4, groups=256, padding=20)),
+                norm_f(Conv1d(1024, 1024, 5, 1, padding=2)),
+            ]
+        )
+        self.conv_post = norm_f(Conv1d(1024, 1, 3, 1, padding=1))
+
+    def forward(self, x):
+        fmap = []
+
+        for layer in self.convs:
+            x = layer(x)
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            fmap.append(x)
+        x = self.conv_post(x)
+        fmap.append(x)
+        x = torch.flatten(x, 1, -1)
+
+        return x, fmap
+
+
+class MultiPeriodDiscriminator(torch.nn.Module):
+    def __init__(self, use_spectral_norm=False):
+        super(MultiPeriodDiscriminator, self).__init__()
+        periods = [2, 3, 5, 7, 11]
+
+        discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
+        discs = discs + [
+            DiscriminatorP(i, use_spectral_norm=use_spectral_norm) for i in periods
+        ]
+        self.discriminators = nn.ModuleList(discs)
+
+    def forward(self, y, y_hat):
+        y_d_rs = []
+        y_d_gs = []
+        fmap_rs = []
+        fmap_gs = []
+        for i, d in enumerate(self.discriminators):
+            y_d_r, fmap_r = d(y)
+            y_d_g, fmap_g = d(y_hat)
+            y_d_rs.append(y_d_r)
+            y_d_gs.append(y_d_g)
+            fmap_rs.append(fmap_r)
+            fmap_gs.append(fmap_g)
+
+        return y_d_rs, y_d_gs, fmap_rs, fmap_gs
+
+
+class ReferenceEncoder(nn.Module):
+    """
+    inputs --- [N, Ty/r, n_mels*r]  mels
+    outputs --- [N, ref_enc_gru_size]
+    """
+
+    def __init__(self, spec_channels, gin_channels=0):
+        super().__init__()
+        self.spec_channels = spec_channels
+        ref_enc_filters = [32, 32, 64, 64, 128, 128]
+        K = len(ref_enc_filters)
+        filters = [1] + ref_enc_filters
+        convs = [
+            weight_norm(
+                nn.Conv2d(
+                    in_channels=filters[i],
+                    out_channels=filters[i + 1],
+                    kernel_size=(3, 3),
+                    stride=(2, 2),
+                    padding=(1, 1),
+                )
+            )
+            for i in range(K)
+        ]
+        self.convs = nn.ModuleList(convs)
+        # self.wns = nn.ModuleList([weight_norm(num_features=ref_enc_filters[i]) for i in range(K)]) # noqa: E501
+
+        out_channels = self.calculate_channels(spec_channels, 3, 2, 1, K)
+        self.gru = nn.GRU(
+            input_size=ref_enc_filters[-1] * out_channels,
+            hidden_size=256 // 2,
+            batch_first=True,
+        )
+        self.proj = nn.Linear(128, gin_channels)
+
+    def forward(self, inputs, mask=None):
+        N = inputs.size(0)
+        out = inputs.view(N, 1, -1, self.spec_channels)  # [N, 1, Ty, n_freqs]
+        for conv in self.convs:
+            out = conv(out)
+            # out = wn(out)
+            out = F.relu(out)  # [N, 128, Ty//2^K, n_mels//2^K]
+
+        out = out.transpose(1, 2)  # [N, Ty//2^K, 128, n_mels//2^K]
+        T = out.size(1)
+        N = out.size(0)
+        out = out.contiguous().view(N, T, -1)  # [N, Ty//2^K, 128*n_mels//2^K]
+
+        self.gru.flatten_parameters()
+        memory, out = self.gru(out)  # out --- [1, N, 128]
+
+        return self.proj(out.squeeze(0))
+
+    def calculate_channels(self, L, kernel_size, stride, pad, n_convs):
+        for i in range(n_convs):
+            L = (L - kernel_size + 2 * pad) // stride + 1
+        return L
+
+
+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=256,
+        gin_channels=256,
+        use_sdp=True,
+        n_flow_layer=4,
+        n_layers_trans_flow=4,
+        flow_share_parameter=False,
+        use_transformer_flow=True,
+        **kwargs
+    ):
+        super().__init__()
+        self.n_vocab = n_vocab
+        self.spec_channels = spec_channels
+        self.inter_channels = inter_channels
+        self.hidden_channels = hidden_channels
+        self.filter_channels = filter_channels
+        self.n_heads = n_heads
+        self.n_layers = n_layers
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        self.resblock = resblock
+        self.resblock_kernel_sizes = resblock_kernel_sizes
+        self.resblock_dilation_sizes = resblock_dilation_sizes
+        self.upsample_rates = upsample_rates
+        self.upsample_initial_channel = upsample_initial_channel
+        self.upsample_kernel_sizes = upsample_kernel_sizes
+        self.segment_size = segment_size
+        self.n_speakers = n_speakers
+        self.gin_channels = gin_channels
+        self.n_layers_trans_flow = n_layers_trans_flow
+        self.use_spk_conditioned_encoder = kwargs.get(
+            "use_spk_conditioned_encoder", True
+        )
+        self.use_sdp = use_sdp
+        self.use_noise_scaled_mas = kwargs.get("use_noise_scaled_mas", False)
+        self.mas_noise_scale_initial = kwargs.get("mas_noise_scale_initial", 0.01)
+        self.noise_scale_delta = kwargs.get("noise_scale_delta", 2e-6)
+        self.current_mas_noise_scale = self.mas_noise_scale_initial
+        if self.use_spk_conditioned_encoder and gin_channels > 0:
+            self.enc_gin_channels = gin_channels
+        self.enc_p = TextEncoder(
+            n_vocab,
+            inter_channels,
+            hidden_channels,
+            filter_channels,
+            n_heads,
+            n_layers,
+            kernel_size,
+            p_dropout,
+            self.n_speakers,
+            gin_channels=self.enc_gin_channels,
+        )
+        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,
+        )
+        if use_transformer_flow:
+            self.flow = TransformerCouplingBlock(
+                inter_channels,
+                hidden_channels,
+                filter_channels,
+                n_heads,
+                n_layers_trans_flow,
+                5,
+                p_dropout,
+                n_flow_layer,
+                gin_channels=gin_channels,
+                share_parameter=flow_share_parameter,
+            )
+        else:
+            self.flow = ResidualCouplingBlock(
+                inter_channels,
+                hidden_channels,
+                5,
+                1,
+                n_flow_layer,
+                gin_channels=gin_channels,
+            )
+        self.sdp = StochasticDurationPredictor(
+            hidden_channels, 192, 3, 0.5, 4, gin_channels=gin_channels
+        )
+        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)
+        else:
+            self.ref_enc = ReferenceEncoder(spec_channels, gin_channels)
+
+    def forward(
+        self,
+        x,
+        x_lengths,
+        y,
+        y_lengths,
+        sid,
+        tone,
+        language,
+        bert,
+        ja_bert,
+        en_bert,
+        emo=None,
+    ):
+        if self.n_speakers > 0:
+            g = self.emb_g(sid).unsqueeze(-1)  # [b, h, 1]
+        else:
+            g = self.ref_enc(y.transpose(1, 2)).unsqueeze(-1)
+        x, m_p, logs_p, x_mask, loss_commit = self.enc_p(
+            x, x_lengths, tone, language, bert, ja_bert, en_bert, emo, sid, g=g
+        )
+        z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
+        z_p = self.flow(z, y_mask, g=g)
+
+        with torch.no_grad():
+            # negative cross-entropy
+            s_p_sq_r = torch.exp(-2 * logs_p)  # [b, d, t]
+            neg_cent1 = torch.sum(
+                -0.5 * math.log(2 * math.pi) - logs_p, [1], keepdim=True
+            )  # [b, 1, t_s]
+            neg_cent2 = torch.matmul(
+                -0.5 * (z_p**2).transpose(1, 2), s_p_sq_r
+            )  # [b, t_t, d] x [b, d, t_s] = [b, t_t, t_s]
+            neg_cent3 = torch.matmul(
+                z_p.transpose(1, 2), (m_p * s_p_sq_r)
+            )  # [b, t_t, d] x [b, d, t_s] = [b, t_t, t_s]
+            neg_cent4 = torch.sum(
+                -0.5 * (m_p**2) * s_p_sq_r, [1], keepdim=True
+            )  # [b, 1, t_s]
+            neg_cent = neg_cent1 + neg_cent2 + neg_cent3 + neg_cent4
+            if self.use_noise_scaled_mas:
+                epsilon = (
+                    torch.std(neg_cent)
+                    * torch.randn_like(neg_cent)
+                    * self.current_mas_noise_scale
+                )
+                neg_cent = neg_cent + epsilon
+
+            attn_mask = torch.unsqueeze(x_mask, 2) * torch.unsqueeze(y_mask, -1)
+            attn = (
+                monotonic_align.maximum_path(neg_cent, attn_mask.squeeze(1))
+                .unsqueeze(1)
+                .detach()
+            )
+
+        w = attn.sum(2)
+
+        l_length_sdp = self.sdp(x, x_mask, w, g=g)
+        l_length_sdp = l_length_sdp / torch.sum(x_mask)
+
+        logw_ = torch.log(w + 1e-6) * x_mask
+        logw = self.dp(x, x_mask, g=g)
+        l_length_dp = torch.sum((logw - logw_) ** 2, [1, 2]) / torch.sum(
+            x_mask
+        )  # for averaging
+
+        l_length = l_length_dp + l_length_sdp
+
+        # expand prior
+        m_p = torch.matmul(attn.squeeze(1), m_p.transpose(1, 2)).transpose(1, 2)
+        logs_p = torch.matmul(attn.squeeze(1), logs_p.transpose(1, 2)).transpose(1, 2)
+
+        z_slice, ids_slice = commons.rand_slice_segments(
+            z, y_lengths, self.segment_size
+        )
+        o = self.dec(z_slice, g=g)
+        return (
+            o,
+            l_length,
+            attn,
+            ids_slice,
+            x_mask,
+            y_mask,
+            (z, z_p, m_p, logs_p, m_q, logs_q),
+            (x, logw, logw_),
+            loss_commit,
+        )
+
+    def infer(
+        self,
+        x,
+        x_lengths,
+        sid,
+        tone,
+        language,
+        bert,
+        ja_bert,
+        en_bert,
+        emo=None,
+        noise_scale=0.667,
+        length_scale=1,
+        noise_scale_w=0.8,
+        max_len=None,
+        sdp_ratio=0,
+        y=None,
+    ):
+        # x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths, tone, language, bert)
+        # g = self.gst(y)
+        if self.n_speakers > 0:
+            g = self.emb_g(sid).unsqueeze(-1)  # [b, h, 1]
+        else:
+            g = self.ref_enc(y.transpose(1, 2)).unsqueeze(-1)
+        x, m_p, logs_p, x_mask, _ = self.enc_p(
+            x, x_lengths, tone, language, bert, ja_bert, en_bert, emo, sid, g=g
+        )
+        logw = self.sdp(x, x_mask, g=g, reverse=True, noise_scale=noise_scale_w) * (
+            sdp_ratio
+        ) + self.dp(x, x_mask, g=g) * (1 - sdp_ratio)
+        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)

modules.py

@@ -0,0 +1,597 @@
+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
+from attentions import Encoder
+
+LRELU_SLOPE = 0.1
+
+
+class LayerNorm(nn.Module):
+    def __init__(self, channels, eps=1e-5):
+        super().__init__()
+        self.channels = channels
+        self.eps = eps
+
+        self.gamma = nn.Parameter(torch.ones(channels))
+        self.beta = nn.Parameter(torch.zeros(channels))
+
+    def forward(self, x):
+        x = x.transpose(1, -1)
+        x = F.layer_norm(x, (self.channels,), self.gamma, self.beta, self.eps)
+        return x.transpose(1, -1)
+
+
+class ConvReluNorm(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        hidden_channels,
+        out_channels,
+        kernel_size,
+        n_layers,
+        p_dropout,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.hidden_channels = hidden_channels
+        self.out_channels = out_channels
+        self.kernel_size = kernel_size
+        self.n_layers = n_layers
+        self.p_dropout = p_dropout
+        assert n_layers > 1, "Number of layers should be larger than 0."
+
+        self.conv_layers = nn.ModuleList()
+        self.norm_layers = nn.ModuleList()
+        self.conv_layers.append(
+            nn.Conv1d(
+                in_channels, hidden_channels, kernel_size, padding=kernel_size // 2
+            )
+        )
+        self.norm_layers.append(LayerNorm(hidden_channels))
+        self.relu_drop = nn.Sequential(nn.ReLU(), nn.Dropout(p_dropout))
+        for _ in range(n_layers - 1):
+            self.conv_layers.append(
+                nn.Conv1d(
+                    hidden_channels,
+                    hidden_channels,
+                    kernel_size,
+                    padding=kernel_size // 2,
+                )
+            )
+            self.norm_layers.append(LayerNorm(hidden_channels))
+        self.proj = nn.Conv1d(hidden_channels, out_channels, 1)
+        self.proj.weight.data.zero_()
+        self.proj.bias.data.zero_()
+
+    def forward(self, x, x_mask):
+        x_org = x
+        for i in range(self.n_layers):
+            x = self.conv_layers[i](x * x_mask)
+            x = self.norm_layers[i](x)
+            x = self.relu_drop(x)
+        x = x_org + self.proj(x)
+        return x * x_mask
+
+
+class DDSConv(nn.Module):
+    """
+    Dialted and Depth-Separable Convolution
+    """
+
+    def __init__(self, channels, kernel_size, n_layers, p_dropout=0.0):
+        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=0.0)
+        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
+
+
+class TransformerCouplingLayer(nn.Module):
+    def __init__(
+        self,
+        channels,
+        hidden_channels,
+        kernel_size,
+        n_layers,
+        n_heads,
+        p_dropout=0,
+        filter_channels=0,
+        mean_only=False,
+        wn_sharing_parameter=None,
+        gin_channels=0,
+    ):
+        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.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 = (
+            Encoder(
+                hidden_channels,
+                filter_channels,
+                n_heads,
+                n_layers,
+                kernel_size,
+                p_dropout,
+                isflow=True,
+                gin_channels=gin_channels,
+            )
+            if wn_sharing_parameter is None
+            else wn_sharing_parameter
+        )
+        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
+
+        x1, logabsdet = piecewise_rational_quadratic_transform(
+            x1,
+            unnormalized_widths,
+            unnormalized_heights,
+            unnormalized_derivatives,
+            inverse=reverse,
+            tails="linear",
+            tail_bound=self.tail_bound,
+        )
+
+        x = torch.cat([x0, x1], 1) * x_mask
+        logdet = torch.sum(logabsdet * x_mask, [1, 2])
+        if not reverse:
+            return x, logdet
+        else:
+            return x

monotonic_align/__init__.py

@@ -0,0 +1,16 @@
+from numpy import zeros, int32, float32
+from torch import from_numpy
+
+from .core import maximum_path_jit
+
+
+def maximum_path(neg_cent, mask):
+    device = neg_cent.device
+    dtype = neg_cent.dtype
+    neg_cent = neg_cent.data.cpu().numpy().astype(float32)
+    path = zeros(neg_cent.shape, dtype=int32)
+
+    t_t_max = mask.sum(1)[:, 0].data.cpu().numpy().astype(int32)
+    t_s_max = mask.sum(2)[:, 0].data.cpu().numpy().astype(int32)
+    maximum_path_jit(path, neg_cent, t_t_max, t_s_max)
+    return from_numpy(path).to(device=device, dtype=dtype)

monotonic_align/core.py

@@ -0,0 +1,46 @@
+import numba
+
+
+@numba.jit(
+    numba.void(
+        numba.int32[:, :, ::1],
+        numba.float32[:, :, ::1],
+        numba.int32[::1],
+        numba.int32[::1],
+    ),
+    nopython=True,
+    nogil=True,
+)
+def maximum_path_jit(paths, values, t_ys, t_xs):
+    b = paths.shape[0]
+    max_neg_val = -1e9
+    for i in range(int(b)):
+        path = paths[i]
+        value = values[i]
+        t_y = t_ys[i]
+        t_x = t_xs[i]
+
+        v_prev = v_cur = 0.0
+        index = t_x - 1
+
+        for y in range(t_y):
+            for x in range(max(0, t_x + y - t_y), min(t_x, y + 1)):
+                if x == y:
+                    v_cur = max_neg_val
+                else:
+                    v_cur = value[y - 1, x]
+                if x == 0:
+                    if y == 0:
+                        v_prev = 0.0
+                    else:
+                        v_prev = max_neg_val
+                else:
+                    v_prev = value[y - 1, x - 1]
+                value[y, x] += max(v_prev, v_cur)
+
+        for y in range(t_y - 1, -1, -1):
+            path[y, index] = 1
+            if index != 0 and (
+                index == y or value[y - 1, index] < value[y - 1, index - 1]
+            ):
+                index = index - 1

re_matching.py

@@ -0,0 +1,82 @@
+import re
+
+
+def extract_language_and_text_updated(speaker, dialogue):
+    # 使用正则表达式匹配<语言>标签和其后的文本
+    pattern_language_text = r"<(\S+?)>([^<]+)"
+    matches = re.findall(pattern_language_text, dialogue, re.DOTALL)
+    speaker = speaker[1:-1]
+    # 清理文本：去除两边的空白字符
+    matches_cleaned = [(lang.upper(), text.strip()) for lang, text in matches]
+    matches_cleaned.append(speaker)
+    return matches_cleaned
+
+
+def validate_text(input_text):
+    # 验证说话人的正则表达式
+    pattern_speaker = r"(\[\S+?\])((?:\s*<\S+?>[^<\[\]]+?)+)"
+
+    # 使用re.DOTALL标志使.匹配包括换行符在内的所有字符
+    matches = re.findall(pattern_speaker, input_text, re.DOTALL)
+
+    # 对每个匹配到的说话人内容进行进一步验证
+    for _, dialogue in matches:
+        language_text_matches = extract_language_and_text_updated(_, dialogue)
+        if not language_text_matches:
+            return (
+                False,
+                "Error: Invalid format detected in dialogue content. Please check your input.",
+            )
+
+    # 如果输入的文本中没有找到任何匹配项
+    if not matches:
+        return (
+            False,
+            "Error: No valid speaker format detected. Please check your input.",
+        )
+
+    return True, "Input is valid."
+
+
+def text_matching(text: str) -> list:
+    speaker_pattern = r"(\[\S+?\])(.+?)(?=\[\S+?\]|$)"
+    matches = re.findall(speaker_pattern, text, re.DOTALL)
+    result = []
+    for speaker, dialogue in matches:
+        result.append(extract_language_and_text_updated(speaker, dialogue))
+    print(result)
+    return result
+
+
+def cut_para(text):
+    splitted_para = re.split("[\n]", text)  # 按段分
+    splitted_para = [
+        sentence.strip() for sentence in splitted_para if sentence.strip()
+    ]  # 删除空字符串
+    return splitted_para
+
+
+def cut_sent(para):
+    para = re.sub("([。！;？\?])([^”’])", r"\1\n\2", para)  # 单字符断句符
+    para = re.sub("(\.{6})([^”’])", r"\1\n\2", para)  # 英文省略号
+    para = re.sub("(\…{2})([^”’])", r"\1\n\2", para)  # 中文省略号
+    para = re.sub("([。！？\?][”’])([^，。！？\?])", r"\1\n\2", para)
+    para = para.rstrip()  # 段尾如果有多余的\n就去掉它
+    return para.split("\n")
+
+
+if __name__ == "__main__":
+    text = """
+    [说话人1]
+    [说话人2]<zh>你好吗？<jp>元気ですか？<jp>こんにちは，世界。<zh>你好吗？
+    [说话人3]<zh>谢谢。<jp>どういたしまして。
+    """
+    text_matching(text)
+    # 测试函数
+    test_text = """
+    [说话人1]<zh>你好，こんにちは！<jp>こんにちは，世界。
+    [说话人2]<zh>你好吗？
+    """
+    text_matching(test_text)
+    res = validate_text(test_text)
+    print(res)

requirements.txt

@@ -0,0 +1,33 @@
+librosa==0.9.1
+matplotlib
+numpy
+numba
+phonemizer
+scipy
+tensorboard
+Unidecode
+amfm_decompy
+jieba
+transformers
+pypinyin
+cn2an
+gradio
+av
+mecab-python3
+loguru
+unidic-lite
+cmudict
+fugashi
+num2words
+PyYAML
+requests
+pyopenjtalk; sys_platform == 'linux'
+openjtalk; sys_platform != 'linux'
+jaconv
+psutil
+GPUtil
+vector_quantize_pytorch
+g2p_en
+sentencepiece
+pykakasi
+langid

text/__init__.py

@@ -0,0 +1,51 @@
+from text.symbols import *
+
+_symbol_to_id = {s: i for i, s in enumerate(symbols)}
+
+
+def cleaned_text_to_sequence(cleaned_text, tones, language):
+    """Converts a string of text to a sequence of IDs corresponding to the symbols in the text.
+    Args:
+      text: string to convert to a sequence
+    Returns:
+      List of integers corresponding to the symbols in the text
+    """
+    phones = [_symbol_to_id[symbol] for symbol in cleaned_text]
+    tone_start = language_tone_start_map[language]
+    tones = [i + tone_start for i in tones]
+    lang_id = language_id_map[language]
+    lang_ids = [lang_id for i in phones]
+    return phones, tones, lang_ids
+
+
+def get_bert(norm_text, word2ph, language, device):
+    from .chinese_bert import get_bert_feature as zh_bert
+    from .english_bert_mock import get_bert_feature as en_bert
+    from .japanese_bert import get_bert_feature as jp_bert
+
+    lang_bert_func_map = {"ZH": zh_bert, "EN": en_bert, "JP": jp_bert}
+    bert = lang_bert_func_map[language](norm_text, word2ph, device)
+    return bert
+
+
+def check_bert_models():
+    import json
+    from pathlib import Path
+
+    from config import config
+    from .bert_utils import _check_bert
+
+    if config.mirror.lower() == "openi":
+        import openi
+
+        kwargs = {"token": config.openi_token} if config.openi_token else {}
+        openi.login(**kwargs)
+
+    with open("./bert/bert_models.json", "r") as fp:
+        models = json.load(fp)
+        for k, v in models.items():
+            local_path = Path("./bert").joinpath(k)
+            _check_bert(v["repo_id"], v["files"], local_path)
+
+
+#check_bert_models()

text/bert_utils.py

@@ -0,0 +1,23 @@
+from pathlib import Path
+
+from huggingface_hub import hf_hub_download
+
+from config import config
+
+
+MIRROR: str = config.mirror
+
+
+def _check_bert(repo_id, files, local_path):
+    for file in files:
+        if not Path(local_path).joinpath(file).exists():
+            if MIRROR.lower() == "openi":
+                import openi
+
+                openi.model.download_model(
+                    "Stardust_minus/Bert-VITS2", repo_id.split("/")[-1], "./bert"
+                )
+            else:
+                hf_hub_download(
+                    repo_id, file, local_dir=local_path, local_dir_use_symlinks=False
+                )

text/chinese.py

@@ -0,0 +1,199 @@
+import os
+import re
+
+import cn2an
+from pypinyin import lazy_pinyin, Style
+
+from text.symbols import punctuation
+from text.tone_sandhi import ToneSandhi
+
+current_file_path = os.path.dirname(__file__)
+pinyin_to_symbol_map = {
+    line.split("\t")[0]: line.strip().split("\t")[1]
+    for line in open(os.path.join(current_file_path, "opencpop-strict.txt")).readlines()
+}
+
+import jieba.posseg as psg
+
+
+rep_map = {
+    "：": ",",
+    "；": ",",
+    "，": ",",
+    "。": ".",
+    "！": "!",
+    "？": "?",
+    "\n": ".",
+    "·": ",",
+    "、": ",",
+    "...": "…",
+    "$": ".",
+    "“": "'",
+    "”": "'",
+    '"': "'",
+    "‘": "'",
+    "’": "'",
+    "（": "'",
+    "）": "'",
+    "(": "'",
+    ")": "'",
+    "《": "'",
+    "》": "'",
+    "【": "'",
+    "】": "'",
+    "[": "'",
+    "]": "'",
+    "—": "-",
+    "～": "-",
+    "~": "-",
+    "「": "'",
+    "」": "'",
+}
+
+tone_modifier = ToneSandhi()
+
+
+def replace_punctuation(text):
+    text = text.replace("嗯", "恩").replace("呣", "母")
+    pattern = re.compile("|".join(re.escape(p) for p in rep_map.keys()))
+
+    replaced_text = pattern.sub(lambda x: rep_map[x.group()], text)
+
+    replaced_text = re.sub(
+        r"[^\u4e00-\u9fa5" + "".join(punctuation) + r"]+", "", replaced_text
+    )
+
+    return replaced_text
+
+
+def g2p(text):
+    pattern = r"(?<=[{0}])\s*".format("".join(punctuation))
+    sentences = [i for i in re.split(pattern, text) if i.strip() != ""]
+    phones, tones, word2ph = _g2p(sentences)
+    assert sum(word2ph) == len(phones)
+    assert len(word2ph) == len(text)  # Sometimes it will crash,you can add a try-catch.
+    phones = ["_"] + phones + ["_"]
+    tones = [0] + tones + [0]
+    word2ph = [1] + word2ph + [1]
+    return phones, tones, word2ph
+
+
+def _get_initials_finals(word):
+    initials = []
+    finals = []
+    orig_initials = lazy_pinyin(word, neutral_tone_with_five=True, style=Style.INITIALS)
+    orig_finals = lazy_pinyin(
+        word, neutral_tone_with_five=True, style=Style.FINALS_TONE3
+    )
+    for c, v in zip(orig_initials, orig_finals):
+        initials.append(c)
+        finals.append(v)
+    return initials, finals
+
+
+def _g2p(segments):
+    phones_list = []
+    tones_list = []
+    word2ph = []
+    for seg in segments:
+        # Replace all English words in the sentence
+        seg = re.sub("[a-zA-Z]+", "", seg)
+        seg_cut = psg.lcut(seg)
+        initials = []
+        finals = []
+        seg_cut = tone_modifier.pre_merge_for_modify(seg_cut)
+        for word, pos in seg_cut:
+            if pos == "eng":
+                continue
+            sub_initials, sub_finals = _get_initials_finals(word)
+            sub_finals = tone_modifier.modified_tone(word, pos, sub_finals)
+            initials.append(sub_initials)
+            finals.append(sub_finals)
+
+            # assert len(sub_initials) == len(sub_finals) == len(word)
+        initials = sum(initials, [])
+        finals = sum(finals, [])
+        #
+        for c, v in zip(initials, finals):
+            raw_pinyin = c + v
+            # NOTE: post process for pypinyin outputs
+            # we discriminate i, ii and iii
+            if c == v:
+                assert c in punctuation
+                phone = [c]
+                tone = "0"
+                word2ph.append(1)
+            else:
+                v_without_tone = v[:-1]
+                tone = v[-1]
+
+                pinyin = c + v_without_tone
+                assert tone in "12345"
+
+                if c:
+                    # 多音节
+                    v_rep_map = {
+                        "uei": "ui",
+                        "iou": "iu",
+                        "uen": "un",
+                    }
+                    if v_without_tone in v_rep_map.keys():
+                        pinyin = c + v_rep_map[v_without_tone]
+                else:
+                    # 单音节
+                    pinyin_rep_map = {
+                        "ing": "ying",
+                        "i": "yi",
+                        "in": "yin",
+                        "u": "wu",
+                    }
+                    if pinyin in pinyin_rep_map.keys():
+                        pinyin = pinyin_rep_map[pinyin]
+                    else:
+                        single_rep_map = {
+                            "v": "yu",
+                            "e": "e",
+                            "i": "y",
+                            "u": "w",
+                        }
+                        if pinyin[0] in single_rep_map.keys():
+                            pinyin = single_rep_map[pinyin[0]] + pinyin[1:]
+
+                assert pinyin in pinyin_to_symbol_map.keys(), (pinyin, seg, raw_pinyin)
+                phone = pinyin_to_symbol_map[pinyin].split(" ")
+                word2ph.append(len(phone))
+
+            phones_list += phone
+            tones_list += [int(tone)] * len(phone)
+    return phones_list, tones_list, word2ph
+
+
+def text_normalize(text):
+    numbers = re.findall(r"\d+(?:\.?\d+)?", text)
+    for number in numbers:
+        text = text.replace(number, cn2an.an2cn(number), 1)
+    text = replace_punctuation(text)
+    return text
+
+
+def get_bert_feature(text, word2ph):
+    from text import chinese_bert
+
+    return chinese_bert.get_bert_feature(text, word2ph)
+
+
+if __name__ == "__main__":
+    from text.chinese_bert import get_bert_feature
+
+    text = "啊！但是《原神》是由,米哈\游自主，  [研发]的一款全.新开放世界.冒险游戏"
+    text = text_normalize(text)
+    print(text)
+    phones, tones, word2ph = g2p(text)
+    bert = get_bert_feature(text, word2ph)
+
+    print(phones, tones, word2ph, bert.shape)
+
+
+# # 示例用法
+# text = "这是一个示例文本：,你好！这是一个测试...."
+# print(g2p_paddle(text))  # 输出: 这是一个示例文本你好这是一个测试

text/chinese_bert.py

@@ -0,0 +1,102 @@
+import sys
+
+import torch
+from transformers import AutoModelForMaskedLM, AutoTokenizer
+
+from config import config
+
+#LOCAL_PATH = "./bert/chinese-roberta-wwm-ext-large"
+LOCAL_PATH = 'hfl/chinese-roberta-wwm-ext-large' 
+
+tokenizer = AutoTokenizer.from_pretrained(LOCAL_PATH)
+
+models = dict()
+
+
+def get_bert_feature(text, word2ph, device=config.bert_gen_config.device):
+    if (
+        sys.platform == "darwin"
+        and torch.backends.mps.is_available()
+        and device == "cpu"
+    ):
+        device = "mps"
+    if not device:
+        device = "cuda"
+    if device not in models.keys():
+        models[device] = AutoModelForMaskedLM.from_pretrained(LOCAL_PATH).to(device)
+    with torch.no_grad():
+        inputs = tokenizer(text, return_tensors="pt")
+        for i in inputs:
+            inputs[i] = inputs[i].to(device)
+        res = models[device](**inputs, output_hidden_states=True)
+        res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()
+
+    assert len(word2ph) == len(text) + 2
+    word2phone = word2ph
+    phone_level_feature = []
+    for i in range(len(word2phone)):
+        repeat_feature = res[i].repeat(word2phone[i], 1)
+        phone_level_feature.append(repeat_feature)
+
+    phone_level_feature = torch.cat(phone_level_feature, dim=0)
+
+    return phone_level_feature.T
+
+
+if __name__ == "__main__":
+    word_level_feature = torch.rand(38, 1024)  # 12个词,每个词1024维特征
+    word2phone = [
+        1,
+        2,
+        1,
+        2,
+        2,
+        1,
+        2,
+        2,
+        1,
+        2,
+        2,
+        1,
+        2,
+        2,
+        2,
+        2,
+        2,
+        1,
+        1,
+        2,
+        2,
+        1,
+        2,
+        2,
+        2,
+        2,
+        1,
+        2,
+        2,
+        2,
+        2,
+        2,
+        1,
+        2,
+        2,
+        2,
+        2,
+        1,
+    ]
+
+    # 计算总帧数
+    total_frames = sum(word2phone)
+    print(word_level_feature.shape)
+    print(word2phone)
+    phone_level_feature = []
+    for i in range(len(word2phone)):
+        print(word_level_feature[i].shape)
+
+        # 对每个词重复word2phone[i]次
+        repeat_feature = word_level_feature[i].repeat(word2phone[i], 1)
+        phone_level_feature.append(repeat_feature)
+
+    phone_level_feature = torch.cat(phone_level_feature, dim=0)
+    print(phone_level_feature.shape)  # torch.Size([36, 1024])

text/cleaner.py

@@ -0,0 +1,28 @@
+from text import chinese, japanese, english, cleaned_text_to_sequence
+
+
+language_module_map = {"ZH": chinese, "JP": japanese, "EN": english}
+
+
+def clean_text(text, language):
+    language_module = language_module_map[language]
+    norm_text = language_module.text_normalize(text)
+    phones, tones, word2ph = language_module.g2p(norm_text)
+    return norm_text, phones, tones, word2ph
+
+
+def clean_text_bert(text, language):
+    language_module = language_module_map[language]
+    norm_text = language_module.text_normalize(text)
+    phones, tones, word2ph = language_module.g2p(norm_text)
+    bert = language_module.get_bert_feature(norm_text, word2ph)
+    return phones, tones, bert
+
+
+def text_to_sequence(text, language):
+    norm_text, phones, tones, word2ph = clean_text(text, language)
+    return cleaned_text_to_sequence(phones, tones, language)
+
+
+if __name__ == "__main__":
+    pass

text/cmudict_cache.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9b21b20325471934ba92f2e4a5976989e7d920caa32e7a286eacb027d197949
+size 6212655

text/english.py

@@ -0,0 +1,454 @@
+import pickle
+import os
+import re
+from g2p_en import G2p
+from transformers import DebertaV2Tokenizer
+
+from text import symbols
+
+current_file_path = os.path.dirname(__file__)
+CMU_DICT_PATH = os.path.join(current_file_path, "cmudict.rep")
+CACHE_PATH = os.path.join(current_file_path, "cmudict_cache.pickle")
+_g2p = G2p()
+#LOCAL_PATH = "./bert/deberta-v3-large"
+LOCAL_PATH = 'microsoft/deberta-v3-large'
+tokenizer = DebertaV2Tokenizer.from_pretrained(LOCAL_PATH)
+
+arpa = {
+    "AH0",
+    "S",
+    "AH1",
+    "EY2",
+    "AE2",
+    "EH0",
+    "OW2",
+    "UH0",
+    "NG",
+    "B",
+    "G",
+    "AY0",
+    "M",
+    "AA0",
+    "F",
+    "AO0",
+    "ER2",
+    "UH1",
+    "IY1",
+    "AH2",
+    "DH",
+    "IY0",
+    "EY1",
+    "IH0",
+    "K",
+    "N",
+    "W",
+    "IY2",
+    "T",
+    "AA1",
+    "ER1",
+    "EH2",
+    "OY0",
+    "UH2",
+    "UW1",
+    "Z",
+    "AW2",
+    "AW1",
+    "V",
+    "UW2",
+    "AA2",
+    "ER",
+    "AW0",
+    "UW0",
+    "R",
+    "OW1",
+    "EH1",
+    "ZH",
+    "AE0",
+    "IH2",
+    "IH",
+    "Y",
+    "JH",
+    "P",
+    "AY1",
+    "EY0",
+    "OY2",
+    "TH",
+    "HH",
+    "D",
+    "ER0",
+    "CH",
+    "AO1",
+    "AE1",
+    "AO2",
+    "OY1",
+    "AY2",
+    "IH1",
+    "OW0",
+    "L",
+    "SH",
+}
+
+
+def post_replace_ph(ph):
+    rep_map = {
+        "：": ",",
+        "；": ",",
+        "，": ",",
+        "。": ".",
+        "！": "!",
+        "？": "?",
+        "\n": ".",
+        "·": ",",
+        "、": ",",
+        "…": "...",
+        "···": "...",
+        "・・・": "...",
+        "v": "V",
+    }
+    if ph in rep_map.keys():
+        ph = rep_map[ph]
+    if ph in symbols:
+        return ph
+    if ph not in symbols:
+        ph = "UNK"
+    return ph
+
+
+rep_map = {
+    "：": ",",
+    "；": ",",
+    "，": ",",
+    "。": ".",
+    "！": "!",
+    "？": "?",
+    "\n": ".",
+    "．": ".",
+    "…": "...",
+    "···": "...",
+    "・・・": "...",
+    "·": ",",
+    "・": ",",
+    "、": ",",
+    "$": ".",
+    "“": "'",
+    "”": "'",
+    '"': "'",
+    "‘": "'",
+    "’": "'",
+    "（": "'",
+    "）": "'",
+    "(": "'",
+    ")": "'",
+    "《": "'",
+    "》": "'",
+    "【": "'",
+    "】": "'",
+    "[": "'",
+    "]": "'",
+    "—": "-",
+    "−": "-",
+    "～": "-",
+    "~": "-",
+    "「": "'",
+    "」": "'",
+}
+
+
+def replace_punctuation(text):
+    pattern = re.compile("|".join(re.escape(p) for p in rep_map.keys()))
+
+    replaced_text = pattern.sub(lambda x: rep_map[x.group()], text)
+
+    # replaced_text = re.sub(
+    #     r"[^\u3040-\u309F\u30A0-\u30FF\u4E00-\u9FFF\u3400-\u4DBF\u3005"
+    #     + "".join(punctuation)
+    #     + r"]+",
+    #     "",
+    #     replaced_text,
+    # )
+
+    return replaced_text
+
+
+def read_dict():
+    g2p_dict = {}
+    start_line = 49
+    with open(CMU_DICT_PATH) as f:
+        line = f.readline()
+        line_index = 1
+        while line:
+            if line_index >= start_line:
+                line = line.strip()
+                word_split = line.split("  ")
+                word = word_split[0]
+
+                syllable_split = word_split[1].split(" - ")
+                g2p_dict[word] = []
+                for syllable in syllable_split:
+                    phone_split = syllable.split(" ")
+                    g2p_dict[word].append(phone_split)
+
+            line_index = line_index + 1
+            line = f.readline()
+
+    return g2p_dict
+
+
+def cache_dict(g2p_dict, file_path):
+    with open(file_path, "wb") as pickle_file:
+        pickle.dump(g2p_dict, pickle_file)
+
+
+def get_dict():
+    if os.path.exists(CACHE_PATH):
+        with open(CACHE_PATH, "rb") as pickle_file:
+            g2p_dict = pickle.load(pickle_file)
+    else:
+        g2p_dict = read_dict()
+        cache_dict(g2p_dict, CACHE_PATH)
+
+    return g2p_dict
+
+
+eng_dict = get_dict()
+
+
+def refine_ph(phn):
+    tone = 0
+    if re.search(r"\d$", phn):
+        tone = int(phn[-1]) + 1
+        phn = phn[:-1]
+    return phn.lower(), tone
+
+
+def refine_syllables(syllables):
+    tones = []
+    phonemes = []
+    for phn_list in syllables:
+        for i in range(len(phn_list)):
+            phn = phn_list[i]
+            phn, tone = refine_ph(phn)
+            phonemes.append(phn)
+            tones.append(tone)
+    return phonemes, tones
+
+
+import re
+import inflect
+
+_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_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 _remove_commas(m):
+    return m.group(1).replace(",", "")
+
+
+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 _expand_decimal_point(m):
+    return m.group(1).replace(".", " point ")
+
+
+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 text_normalize(text):
+    text = normalize_numbers(text)
+    text = replace_punctuation(text)
+    text = re.sub(r"([,;.\?\!])([\w])", r"\1 \2", text)
+    return text
+
+
+def distribute_phone(n_phone, n_word):
+    phones_per_word = [0] * n_word
+    for task in range(n_phone):
+        min_tasks = min(phones_per_word)
+        min_index = phones_per_word.index(min_tasks)
+        phones_per_word[min_index] += 1
+    return phones_per_word
+
+
+def sep_text(text):
+    words = re.split(r"([,;.\?\!\s+])", text)
+    words = [word for word in words if word.strip() != ""]
+    return words
+
+
+def g2p(text):
+    phones = []
+    tones = []
+    # word2ph = []
+    words = sep_text(text)
+    tokens = [tokenizer.tokenize(i) for i in words]
+    for word in words:
+        if word.upper() in eng_dict:
+            phns, tns = refine_syllables(eng_dict[word.upper()])
+            phones.append([post_replace_ph(i) for i in phns])
+            tones.append(tns)
+            # word2ph.append(len(phns))
+        else:
+            phone_list = list(filter(lambda p: p != " ", _g2p(word)))
+            phns = []
+            tns = []
+            for ph in phone_list:
+                if ph in arpa:
+                    ph, tn = refine_ph(ph)
+                    phns.append(ph)
+                    tns.append(tn)
+                else:
+                    phns.append(ph)
+                    tns.append(0)
+            phones.append([post_replace_ph(i) for i in phns])
+            tones.append(tns)
+            # word2ph.append(len(phns))
+    # phones = [post_replace_ph(i) for i in phones]
+
+    word2ph = []
+    for token, phoneme in zip(tokens, phones):
+        phone_len = len(phoneme)
+        word_len = len(token)
+
+        aaa = distribute_phone(phone_len, word_len)
+        word2ph += aaa
+
+    phones = ["_"] + [j for i in phones for j in i] + ["_"]
+    tones = [0] + [j for i in tones for j in i] + [0]
+    word2ph = [1] + word2ph + [1]
+    assert len(phones) == len(tones), text
+    assert len(phones) == sum(word2ph), text
+
+    return phones, tones, word2ph
+
+
+def get_bert_feature(text, word2ph):
+    from text import english_bert_mock
+
+    return english_bert_mock.get_bert_feature(text, word2ph)
+
+
+if __name__ == "__main__":
+    # print(get_dict())
+    # print(eng_word_to_phoneme("hello"))
+    print(g2p("In this paper, we propose 1 DSPGAN, a GAN-based universal vocoder."))
+    # all_phones = set()
+    # for k, syllables in eng_dict.items():
+    #     for group in syllables:
+    #         for ph in group:
+    #             all_phones.add(ph)
+    # print(all_phones)

text/english_bert_mock.py

@@ -0,0 +1,43 @@
+import sys
+
+import torch
+from transformers import DebertaV2Model, DebertaV2Tokenizer
+
+from config import config
+
+
+#LOCAL_PATH = "./bert/deberta-v3-large"
+LOCAL_PATH = 'microsoft/deberta-v2-xlarge'
+
+tokenizer = DebertaV2Tokenizer.from_pretrained(LOCAL_PATH)
+
+models = dict()
+
+
+def get_bert_feature(text, word2ph, device=config.bert_gen_config.device):
+    if (
+        sys.platform == "darwin"
+        and torch.backends.mps.is_available()
+        and device == "cpu"
+    ):
+        device = "mps"
+    if not device:
+        device = "cuda"
+    if device not in models.keys():
+        models[device] = DebertaV2Model.from_pretrained(LOCAL_PATH).to(device)
+    with torch.no_grad():
+        inputs = tokenizer(text, return_tensors="pt")
+        for i in inputs:
+            inputs[i] = inputs[i].to(device)
+        res = models[device](**inputs, output_hidden_states=True)
+        res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()
+    assert len(word2ph) == res.shape[0], (text, res.shape[0], len(word2ph))
+    word2phone = word2ph
+    phone_level_feature = []
+    for i in range(len(word2phone)):
+        repeat_feature = res[i].repeat(word2phone[i], 1)
+        phone_level_feature.append(repeat_feature)
+
+    phone_level_feature = torch.cat(phone_level_feature, dim=0)
+
+    return phone_level_feature.T

text/japanese.py

@@ -0,0 +1,432 @@
+# Convert Japanese text to phonemes which is
+# compatible with Julius https://github.com/julius-speech/segmentation-kit
+import re
+import unicodedata
+
+from transformers import AutoTokenizer
+
+from text import punctuation, symbols
+
+from num2words import num2words
+
+import pyopenjtalk
+import jaconv
+
+
+def kata2phoneme(text: str) -> str:
+    """Convert katakana text to phonemes."""
+    text = text.strip()
+    if text == "ー":
+        return ["ー"]
+    elif text.startswith("ー"):
+        return ["ー"] + kata2phoneme(text[1:])
+    res = []
+    prev = None
+    while text:
+        if re.match(_MARKS, text):
+            res.append(text)
+            text = text[1:]
+            continue
+        if text.startswith("ー"):
+            if prev:
+                res.append(prev[-1])
+            text = text[1:]
+            continue
+        res += pyopenjtalk.g2p(text).lower().replace("cl", "q").split(" ")
+        break
+    # res = _COLON_RX.sub(":", res)
+    return res
+
+
+def hira2kata(text: str) -> str:
+    return jaconv.hira2kata(text)
+
+
+_SYMBOL_TOKENS = set(list("・、。？！"))
+_NO_YOMI_TOKENS = set(list("「」『』―（）［］[]"))
+_MARKS = re.compile(
+    r"[^A-Za-z\d\u3005\u3040-\u30ff\u4e00-\u9fff\uff11-\uff19\uff21-\uff3a\uff41-\uff5a\uff66-\uff9d]"
+)
+
+
+def text2kata(text: str) -> str:
+    parsed = pyopenjtalk.run_frontend(text)
+
+    res = []
+    for parts in parsed:
+        word, yomi = replace_punctuation(parts["string"]), parts["pron"].replace(
+            "’", ""
+        )
+        if yomi:
+            if re.match(_MARKS, yomi):
+                if len(word) > 1:
+                    word = [replace_punctuation(i) for i in list(word)]
+                    yomi = word
+                    res += yomi
+                    sep += word
+                    continue
+                elif word not in rep_map.keys() and word not in rep_map.values():
+                    word = ","
+                yomi = word
+            res.append(yomi)
+        else:
+            if word in _SYMBOL_TOKENS:
+                res.append(word)
+            elif word in ("っ", "ッ"):
+                res.append("ッ")
+            elif word in _NO_YOMI_TOKENS:
+                pass
+            else:
+                res.append(word)
+    return hira2kata("".join(res))
+
+
+def text2sep_kata(text: str) -> (list, list):
+    parsed = pyopenjtalk.run_frontend(text)
+
+    res = []
+    sep = []
+    for parts in parsed:
+        word, yomi = replace_punctuation(parts["string"]), parts["pron"].replace(
+            "’", ""
+        )
+        if yomi:
+            if re.match(_MARKS, yomi):
+                if len(word) > 1:
+                    word = [replace_punctuation(i) for i in list(word)]
+                    yomi = word
+                    res += yomi
+                    sep += word
+                    continue
+                elif word not in rep_map.keys() and word not in rep_map.values():
+                    word = ","
+                yomi = word
+            res.append(yomi)
+        else:
+            if word in _SYMBOL_TOKENS:
+                res.append(word)
+            elif word in ("っ", "ッ"):
+                res.append("ッ")
+            elif word in _NO_YOMI_TOKENS:
+                pass
+            else:
+                res.append(word)
+        sep.append(word)
+    return sep, [hira2kata(i) for i in res], get_accent(parsed)
+
+
+def get_accent(parsed):
+    labels = pyopenjtalk.make_label(parsed)
+
+    phonemes = []
+    accents = []
+    for n, label in enumerate(labels):
+        phoneme = re.search(r"\-([^\+]*)\+", label).group(1)
+        if phoneme not in ["sil", "pau"]:
+            phonemes.append(phoneme.replace("cl", "q").lower())
+        else:
+            continue
+        a1 = int(re.search(r"/A:(\-?[0-9]+)\+", label).group(1))
+        a2 = 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))
+        # Falling
+        if a1 == 0 and a2_next == a2 + 1:
+            accents.append(-1)
+        # Rising
+        elif a2 == 1 and a2_next == 2:
+            accents.append(1)
+        else:
+            accents.append(0)
+    return list(zip(phonemes, accents))
+
+
+_ALPHASYMBOL_YOMI = {
+    "#": "シャープ",
+    "%": "パーセント",
+    "&": "アンド",
+    "+": "プラス",
+    "-": "マイナス",
+    ":": "コロン",
+    ";": "セミコロン",
+    "<": "小なり",
+    "=": "イコール",
+    ">": "大なり",
+    "@": "アット",
+    "a": "エー",
+    "b": "ビー",
+    "c": "シー",
+    "d": "ディー",
+    "e": "イー",
+    "f": "エフ",
+    "g": "ジー",
+    "h": "エイチ",
+    "i": "アイ",
+    "j": "ジェー",
+    "k": "ケー",
+    "l": "エル",
+    "m": "エム",
+    "n": "エヌ",
+    "o": "オー",
+    "p": "ピー",
+    "q": "キュー",
+    "r": "アール",
+    "s": "エス",
+    "t": "ティー",
+    "u": "ユー",
+    "v": "ブイ",
+    "w": "ダブリュー",
+    "x": "エックス",
+    "y": "ワイ",
+    "z": "ゼット",
+    "α": "アルファ",
+    "β": "ベータ",
+    "γ": "ガンマ",
+    "δ": "デルタ",
+    "ε": "イプシロン",
+    "ζ": "ゼータ",
+    "η": "イータ",
+    "θ": "シータ",
+    "ι": "イオタ",
+    "κ": "カッパ",
+    "λ": "ラムダ",
+    "μ": "ミュー",
+    "ν": "ニュー",
+    "ξ": "クサイ",
+    "ο": "オミクロン",
+    "π": "パイ",
+    "ρ": "ロー",
+    "σ": "シグマ",
+    "τ": "タウ",
+    "υ": "ウプシロン",
+    "φ": "ファイ",
+    "χ": "カイ",
+    "ψ": "プサイ",
+    "ω": "オメガ",
+}
+
+
+_NUMBER_WITH_SEPARATOR_RX = re.compile("[0-9]{1,3}(,[0-9]{3})+")
+_CURRENCY_MAP = {"$": "ドル", "¥": "円", "£": "ポンド", "€": "ユーロ"}
+_CURRENCY_RX = re.compile(r"([$¥£€])([0-9.]*[0-9])")
+_NUMBER_RX = re.compile(r"[0-9]+(\.[0-9]+)?")
+
+
+def japanese_convert_numbers_to_words(text: str) -> str:
+    res = _NUMBER_WITH_SEPARATOR_RX.sub(lambda m: m[0].replace(",", ""), text)
+    res = _CURRENCY_RX.sub(lambda m: m[2] + _CURRENCY_MAP.get(m[1], m[1]), res)
+    res = _NUMBER_RX.sub(lambda m: num2words(m[0], lang="ja"), res)
+    return res
+
+
+def japanese_convert_alpha_symbols_to_words(text: str) -> str:
+    return "".join([_ALPHASYMBOL_YOMI.get(ch, ch) for ch in text.lower()])
+
+
+def japanese_text_to_phonemes(text: str) -> str:
+    """Convert Japanese text to phonemes."""
+    res = unicodedata.normalize("NFKC", text)
+    res = japanese_convert_numbers_to_words(res)
+    # res = japanese_convert_alpha_symbols_to_words(res)
+    res = text2kata(res)
+    res = kata2phoneme(res)
+    return res
+
+
+def is_japanese_character(char):
+    # 定义日语文字系统的 Unicode 范围
+    japanese_ranges = [
+        (0x3040, 0x309F),  # 平假名
+        (0x30A0, 0x30FF),  # 片假名
+        (0x4E00, 0x9FFF),  # 汉字 (CJK Unified Ideographs)
+        (0x3400, 0x4DBF),  # 汉字扩展 A
+        (0x20000, 0x2A6DF),  # 汉字扩展 B
+        # 可以根据需要添加其他汉字扩展范围
+    ]
+
+    # 将字符的 Unicode 编码转换为整数
+    char_code = ord(char)
+
+    # 检查字符是否在任何一个日语范围内
+    for start, end in japanese_ranges:
+        if start <= char_code <= end:
+            return True
+
+    return False
+
+
+rep_map = {
+    "：": ",",
+    "；": ",",
+    "，": ",",
+    "。": ".",
+    "！": "!",
+    "？": "?",
+    "\n": ".",
+    "．": ".",
+    "…": "...",
+    "···": "...",
+    "・・・": "...",
+    "·": ",",
+    "・": ",",
+    "、": ",",
+    "$": ".",
+    "“": "'",
+    "”": "'",
+    '"': "'",
+    "‘": "'",
+    "’": "'",
+    "（": "'",
+    "）": "'",
+    "(": "'",
+    ")": "'",
+    "《": "'",
+    "》": "'",
+    "【": "'",
+    "】": "'",
+    "[": "'",
+    "]": "'",
+    "—": "-",
+    "−": "-",
+    "～": "-",
+    "~": "-",
+    "「": "'",
+    "」": "'",
+}
+
+
+def replace_punctuation(text):
+    pattern = re.compile("|".join(re.escape(p) for p in rep_map.keys()))
+
+    replaced_text = pattern.sub(lambda x: rep_map[x.group()], text)
+
+    replaced_text = re.sub(
+        r"[^\u3040-\u309F\u30A0-\u30FF\u4E00-\u9FFF\u3400-\u4DBF\u3005"
+        + "".join(punctuation)
+        + r"]+",
+        "",
+        replaced_text,
+    )
+
+    return replaced_text
+
+
+def text_normalize(text):
+    res = unicodedata.normalize("NFKC", text)
+    res = japanese_convert_numbers_to_words(res)
+    # res = "".join([i for i in res if is_japanese_character(i)])
+    res = replace_punctuation(res)
+    res = res.replace("゙", "")
+    return res
+
+
+def distribute_phone(n_phone, n_word):
+    phones_per_word = [0] * n_word
+    for task in range(n_phone):
+        min_tasks = min(phones_per_word)
+        min_index = phones_per_word.index(min_tasks)
+        phones_per_word[min_index] += 1
+    return phones_per_word
+
+
+def handle_long(sep_phonemes):
+    for i in range(len(sep_phonemes)):
+        if sep_phonemes[i][0] == "ー":
+            sep_phonemes[i][0] = sep_phonemes[i - 1][-1]
+        if "ー" in sep_phonemes[i]:
+            for j in range(len(sep_phonemes[i])):
+                if sep_phonemes[i][j] == "ー":
+                    sep_phonemes[i][j] = sep_phonemes[i][j - 1][-1]
+    return sep_phonemes
+
+
+#tokenizer = AutoTokenizer.from_pretrained("./bert/deberta-v2-large-japanese-char-wwm")
+tokenizer = AutoTokenizer.from_pretrained("ku-nlp/deberta-v2-large-japanese-char-wwm")
+
+def align_tones(phones, tones):
+    res = []
+    for pho in phones:
+        temp = [0] * len(pho)
+        for idx, p in enumerate(pho):
+            if len(tones) == 0:
+                break
+            if p == tones[0][0]:
+                temp[idx] = tones[0][1]
+                if idx > 0:
+                    temp[idx] += temp[idx - 1]
+                tones.pop(0)
+        temp = [0] + temp
+        temp = temp[:-1]
+        if -1 in temp:
+            temp = [i + 1 for i in temp]
+        res.append(temp)
+    res = [i for j in res for i in j]
+    assert not any([i < 0 for i in res]) and not any([i > 1 for i in res])
+    return res
+
+
+def rearrange_tones(tones, phones):
+    res = [0] * len(tones)
+    for i in range(len(tones)):
+        if i == 0:
+            if tones[i] not in punctuation:
+                res[i] = 1
+        elif tones[i] == prev:
+            if phones[i] in punctuation:
+                res[i] = 0
+            else:
+                res[i] = 1
+        elif tones[i] > prev:
+            res[i] = 2
+        elif tones[i] < prev:
+            res[i - 1] = 3
+            res[i] = 1
+        prev = tones[i]
+    return res
+
+
+def g2p(norm_text):
+    sep_text, sep_kata, acc = text2sep_kata(norm_text)
+    sep_tokenized = []
+    for i in sep_text:
+        if i not in punctuation:
+            sep_tokenized.append(tokenizer.tokenize(i))
+        else:
+            sep_tokenized.append([i])
+
+    sep_phonemes = handle_long([kata2phoneme(i) for i in sep_kata])
+    # 异常处理，MeCab不认识的词的话会一路传到这里来，然后炸掉。目前来看只有那些超级稀有的生僻词会出现这种情况
+    for i in sep_phonemes:
+        for j in i:
+            assert j in symbols, (sep_text, sep_kata, sep_phonemes)
+    tones = align_tones(sep_phonemes, acc)
+
+    word2ph = []
+    for token, phoneme in zip(sep_tokenized, sep_phonemes):
+        phone_len = len(phoneme)
+        word_len = len(token)
+
+        aaa = distribute_phone(phone_len, word_len)
+        word2ph += aaa
+    phones = ["_"] + [j for i in sep_phonemes for j in i] + ["_"]
+    # tones = [0] + rearrange_tones(tones, phones[1:-1]) + [0]
+    tones = [0] + tones + [0]
+    word2ph = [1] + word2ph + [1]
+    assert len(phones) == len(tones)
+    return phones, tones, word2ph
+
+
+if __name__ == "__main__":
+    tokenizer = AutoTokenizer.from_pretrained("./bert/deberta-v2-large-japanese")
+    text = "hello,こんにちは、世界ー！……"
+    from text.japanese_bert import get_bert_feature
+
+    text = text_normalize(text)
+    print(text)
+
+    phones, tones, word2ph = g2p(text)
+    bert = get_bert_feature(text, word2ph)
+
+    print(phones, tones, word2ph, bert.shape)