Upload 13 files

README.md

@@ -0,0 +1,14 @@
+---
+title: Vits Models
+emoji: 🏃
+colorFrom: pink
+colorTo: indigo
+sdk: gradio
+sdk_version: 3.17.0
+app_file: app.py
+pinned: false
+license: apache-2.0
+duplicated_from: sayashi/vits-models
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,264 @@
+# coding=utf-8
+import os
+import re
+import argparse
+import utils
+import commons
+import json
+import torch
+import gradio as gr
+from models import SynthesizerTrn
+from text import text_to_sequence, _clean_text
+from torch import no_grad, LongTensor
+import gradio.processing_utils as gr_processing_utils
+import logging
+logging.getLogger('numba').setLevel(logging.WARNING)
+limitation = os.getenv("SYSTEM") == "spaces"  # limit text and audio length in huggingface spaces
+
+hps_ms = utils.get_hparams_from_file(r'config/config.json')
+
+audio_postprocess_ori = gr.Audio.postprocess
+
+def audio_postprocess(self, y):
+    data = audio_postprocess_ori(self, y)
+    if data is None:
+        return None
+    return gr_processing_utils.encode_url_or_file_to_base64(data["name"])
+
+
+gr.Audio.postprocess = audio_postprocess
+
+def get_text(text, hps, is_symbol):
+    text_norm, clean_text = text_to_sequence(text, hps.symbols, [] if is_symbol else hps.data.text_cleaners)
+    if hps.data.add_blank:
+        text_norm = commons.intersperse(text_norm, 0)
+    text_norm = LongTensor(text_norm)
+    return text_norm, clean_text
+
+def create_tts_fn(net_g_ms, speaker_id):
+    def tts_fn(text, language, noise_scale, noise_scale_w, length_scale, is_symbol):
+        text = text.replace('\n', ' ').replace('\r', '').replace(" ", "")
+        if limitation:
+            text_len = len(re.sub("\[([A-Z]{2})\]", "", text))
+            max_len = 100
+            if is_symbol:
+                max_len *= 3
+            if text_len > max_len:
+                return "Error: Text is too long", None
+        if not is_symbol:
+            if language == 0:
+                text = f"[ZH]{text}[ZH]"
+            elif language == 1:
+                text = f"[JA]{text}[JA]"
+            else:
+                text = f"{text}"
+        stn_tst, clean_text = get_text(text, hps_ms, is_symbol)
+        with no_grad():
+            x_tst = stn_tst.unsqueeze(0).to(device)
+            x_tst_lengths = LongTensor([stn_tst.size(0)]).to(device)
+            sid = LongTensor([speaker_id]).to(device)
+            audio = net_g_ms.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale, noise_scale_w=noise_scale_w,
+                                   length_scale=length_scale)[0][0, 0].data.cpu().float().numpy()
+
+        return "Success", (22050, audio)
+    return tts_fn
+
+def create_to_symbol_fn(hps):
+    def to_symbol_fn(is_symbol_input, input_text, temp_lang):
+        if temp_lang == 0:
+            clean_text = f'[ZH]{input_text}[ZH]'
+        elif temp_lang == 1:
+            clean_text = f'[JA]{input_text}[JA]'
+        else:
+            clean_text = input_text
+        return _clean_text(clean_text, hps.data.text_cleaners) if is_symbol_input else ''
+
+    return to_symbol_fn
+def change_lang(language):
+    if language == 0:
+        return 0.6, 0.668, 1.2
+    elif language == 1:
+        return 0.6, 0.668, 1
+    else:
+        return 0.6, 0.668, 1
+
+download_audio_js = """
+() =>{{
+    let root = document.querySelector("body > gradio-app");
+    if (root.shadowRoot != null)
+        root = root.shadowRoot;
+    let audio = root.querySelector("#tts-audio-{audio_id}").querySelector("audio");
+    let text = root.querySelector("#input-text-{audio_id}").querySelector("textarea");
+    if (audio == undefined)
+        return;
+    text = text.value;
+    if (text == undefined)
+        text = Math.floor(Math.random()*100000000);
+    audio = audio.src;
+    let oA = document.createElement("a");
+    oA.download = text.substr(0, 20)+'.wav';
+    oA.href = audio;
+    document.body.appendChild(oA);
+    oA.click();
+    oA.remove();
+}}
+"""
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--device', type=str, default='cpu')
+    parser.add_argument('--api', action="store_true", default=False)
+    parser.add_argument("--share", action="store_true", default=False, help="share gradio app")
+    args = parser.parse_args()
+    device = torch.device(args.device)
+    
+    models = []
+    with open("pretrained_models/info.json", "r", encoding="utf-8") as f:
+        models_info = json.load(f)
+    for i, info in models_info.items():
+        if not info['enable']:
+            continue
+        sid = info['sid']
+        name_en = info['name_en']
+        name_zh = info['name_zh']
+        title = info['title']
+        cover = f"pretrained_models/{i}/{info['cover']}"
+        example = info['example']
+        language = info['language']
+        net_g_ms = SynthesizerTrn(
+            len(hps_ms.symbols),
+            hps_ms.data.filter_length // 2 + 1,
+            hps_ms.train.segment_size // hps_ms.data.hop_length,
+            n_speakers=hps_ms.data.n_speakers if info['type'] == "multi" else 0,
+            **hps_ms.model)
+        utils.load_checkpoint(f'pretrained_models/{i}/{i}.pth', net_g_ms, None)
+        _ = net_g_ms.eval().to(device)
+        models.append((sid, name_en, name_zh, title, cover, example, language, net_g_ms, create_tts_fn(net_g_ms, sid), create_to_symbol_fn(hps_ms)))
+    with gr.Blocks() as app:
+        gr.Markdown(
+            "# <center> vits-models\n"
+            "## <center> Please do not generate content that could infringe upon the rights or cause harm to individuals or organizations.\n"
+            "## <center> ·请不要生成会对个人以及组织造成侵害的内容\n"
+            "![visitor badge](https://visitor-badge.glitch.me/badge?page_id=sayashi.vits-models)\n\n"
+            "[Open In Colab]"
+            "(https://colab.research.google.com/drive/10QOk9NPgoKZUXkIhhuVaZ7SYra1MPMKH?usp=share_link)"
+            " without queue and length limitation.(无需等待队列，并且没有长度限制)\n\n"
+            "[Finetune your own model](https://github.com/SayaSS/vits-finetuning)"
+        )
+
+        with gr.Tabs():
+            with gr.TabItem("EN"):
+                for (sid, name_en, name_zh, title, cover, example, language, net_g_ms, tts_fn, to_symbol_fn) in models:
+                    with gr.TabItem(name_en):
+                        with gr.Row():
+                            gr.Markdown(
+                                '<div align="center">'
+                                f'<a><strong>{title}</strong></a>'
+                                f'<img style="width:auto;height:300px;" src="file/{cover}">' if cover else ""
+                                '</div>'
+                            )
+                        with gr.Row():
+                            with gr.Column():
+                                input_text = gr.Textbox(label="Text (100 words limitation)" if limitation else "Text", lines=5, value=example, elem_id=f"input-text-en-{name_en.replace(' ','')}")
+                                lang = gr.Dropdown(label="Language", choices=["Chinese", "Japanese", "Mix（wrap the Chinese text with [ZH][ZH], wrap the Japanese text with [JA][JA]）"],
+                                            type="index", value=language)
+                                with gr.Accordion(label="Advanced Options", open=False):
+                                    symbol_input = gr.Checkbox(value=False, label="Symbol input")
+                                    symbol_list = gr.Dataset(label="Symbol list", components=[input_text],
+                                                             samples=[[x] for x in hps_ms.symbols])
+                                    symbol_list_json = gr.Json(value=hps_ms.symbols, visible=False)
+                                btn = gr.Button(value="Generate", variant="primary")
+                                with gr.Row():
+                                    ns = gr.Slider(label="noise_scale", minimum=0.1, maximum=1.0, step=0.1, value=0.6, interactive=True)
+                                    nsw = gr.Slider(label="noise_scale_w", minimum=0.1, maximum=1.0, step=0.1, value=0.668, interactive=True)
+                                    ls = gr.Slider(label="length_scale", minimum=0.1, maximum=2.0, step=0.1, value=1.2 if language=="Chinese" else 1, interactive=True)
+                            with gr.Column():
+                                o1 = gr.Textbox(label="Output Message")
+                                o2 = gr.Audio(label="Output Audio", elem_id=f"tts-audio-en-{name_en.replace(' ','')}")
+                                download = gr.Button("Download Audio")
+                            btn.click(tts_fn, inputs=[input_text, lang,  ns, nsw, ls, symbol_input], outputs=[o1, o2], api_name=f"tts-{name_en}")
+                            download.click(None, [], [], _js=download_audio_js.format(audio_id=f"en-{name_en.replace(' ', '')}"))
+                            lang.change(change_lang, inputs=[lang], outputs=[ns, nsw, ls])
+                            symbol_input.change(
+                                to_symbol_fn,
+                                [symbol_input, input_text, lang],
+                                [input_text]
+                            )
+                            symbol_list.click(None, [symbol_list, symbol_list_json], [input_text],
+                                              _js=f"""
+                            (i,symbols) => {{
+                                let root = document.querySelector("body > gradio-app");
+                                if (root.shadowRoot != null)
+                                    root = root.shadowRoot;
+                                let text_input = root.querySelector("#input-text-en-{name_en.replace(' ', '')}").querySelector("textarea");
+                                let startPos = text_input.selectionStart;
+                                let endPos = text_input.selectionEnd;
+                                let oldTxt = text_input.value;
+                                let result = oldTxt.substring(0, startPos) + symbols[i] + oldTxt.substring(endPos);
+                                text_input.value = result;
+                                let x = window.scrollX, y = window.scrollY;
+                                text_input.focus();
+                                text_input.selectionStart = startPos + symbols[i].length;
+                                text_input.selectionEnd = startPos + symbols[i].length;
+                                text_input.blur();
+                                window.scrollTo(x, y);
+                                return text_input.value;
+                            }}""")
+            with gr.TabItem("中文"):
+                for (sid, name_en, name_zh, title, cover, example, language,  net_g_ms, tts_fn, to_symbol_fn) in models:
+                    with gr.TabItem(name_zh):
+                        with gr.Row():
+                            gr.Markdown(
+                                '<div align="center">'
+                                f'<a><strong>{title}</strong></a>'
+                                f'<img style="width:auto;height:300px;" src="file/{cover}">' if cover else ""
+                                '</div>'
+                            )
+                        with gr.Row():
+                            with gr.Column():
+                                input_text = gr.Textbox(label="文本 (100字上限)" if limitation else "文本", lines=5, value=example, elem_id=f"input-text-zh-{name_zh}")
+                                lang = gr.Dropdown(label="语言", choices=["中文", "日语", "中日混合（中文用[ZH][ZH]包裹起来，日文用[JA][JA]包裹起来）"],
+                                            type="index", value="中文"if language == "Chinese" else "日语")
+                                with gr.Accordion(label="高级选项", open=False):
+                                    symbol_input = gr.Checkbox(value=False, label="符号输入")
+                                    symbol_list = gr.Dataset(label="符号列表", components=[input_text],
+                                                             samples=[[x] for x in hps_ms.symbols])
+                                    symbol_list_json = gr.Json(value=hps_ms.symbols, visible=False)
+                                btn = gr.Button(value="生成", variant="primary")
+                                with gr.Row():
+                                    ns = gr.Slider(label="控制感情变化程度", minimum=0.1, maximum=1.0, step=0.1, value=0.6, interactive=True)
+                                    nsw = gr.Slider(label="控制音素发音长度", minimum=0.1, maximum=1.0, step=0.1, value=0.668, interactive=True)
+                                    ls = gr.Slider(label="控制整体语速", minimum=0.1, maximum=2.0, step=0.1, value=1.2 if language=="Chinese" else 1, interactive=True)
+                            with gr.Column():
+                                o1 = gr.Textbox(label="输出信息")
+                                o2 = gr.Audio(label="输出音频", elem_id=f"tts-audio-zh-{name_zh}")
+                                download = gr.Button("下载音频")
+                            btn.click(tts_fn, inputs=[input_text, lang,  ns, nsw, ls, symbol_input], outputs=[o1, o2])
+                            download.click(None, [], [], _js=download_audio_js.format(audio_id=f"zh-{name_zh}"))
+                            lang.change(change_lang, inputs=[lang], outputs=[ns, nsw, ls])
+                            symbol_input.change(
+                                to_symbol_fn,
+                                [symbol_input, input_text, lang],
+                                [input_text]
+                            )
+                            symbol_list.click(None, [symbol_list, symbol_list_json], [input_text],
+                                              _js=f"""
+                            (i,symbols) => {{
+                                let root = document.querySelector("body > gradio-app");
+                                if (root.shadowRoot != null)
+                                    root = root.shadowRoot;
+                                let text_input = root.querySelector("#input-text-zh-{name_zh}").querySelector("textarea");
+                                let startPos = text_input.selectionStart;
+                                let endPos = text_input.selectionEnd;
+                                let oldTxt = text_input.value;
+                                let result = oldTxt.substring(0, startPos) + symbols[i] + oldTxt.substring(endPos);
+                                text_input.value = result;
+                                let x = window.scrollX, y = window.scrollY;
+                                text_input.focus();
+                                text_input.selectionStart = startPos + symbols[i].length;
+                                text_input.selectionEnd = startPos + symbols[i].length;
+                                text_input.blur();
+                                window.scrollTo(x, y);
+                                return text_input.value;
+                            }}""")
+    app.queue(concurrency_count=1, api_open=args.api).launch(share=args.share)

attentions.py

@@ -0,0 +1,300 @@
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+import commons
+from modules import LayerNorm
+   
+
+class Encoder(nn.Module):
+  def __init__(self, hidden_channels, filter_channels, n_heads, n_layers, kernel_size=1, p_dropout=0., window_size=4, **kwargs):
+    super().__init__()
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.window_size = window_size
+
+    self.drop = nn.Dropout(p_dropout)
+    self.attn_layers = nn.ModuleList()
+    self.norm_layers_1 = nn.ModuleList()
+    self.ffn_layers = nn.ModuleList()
+    self.norm_layers_2 = nn.ModuleList()
+    for i in range(self.n_layers):
+      self.attn_layers.append(MultiHeadAttention(hidden_channels, hidden_channels, n_heads, p_dropout=p_dropout, window_size=window_size))
+      self.norm_layers_1.append(LayerNorm(hidden_channels))
+      self.ffn_layers.append(FFN(hidden_channels, hidden_channels, filter_channels, kernel_size, p_dropout=p_dropout))
+      self.norm_layers_2.append(LayerNorm(hidden_channels))
+
+  def forward(self, x, x_mask):
+    attn_mask = x_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+    x = x * x_mask
+    for i in range(self.n_layers):
+      y = self.attn_layers[i](x, x, attn_mask)
+      y = self.drop(y)
+      x = self.norm_layers_1[i](x + y)
+
+      y = self.ffn_layers[i](x, x_mask)
+      y = self.drop(y)
+      x = self.norm_layers_2[i](x + y)
+    x = x * x_mask
+    return x
+
+
+class Decoder(nn.Module):
+  def __init__(self, hidden_channels, filter_channels, n_heads, n_layers, kernel_size=1, p_dropout=0., proximal_bias=False, proximal_init=True, **kwargs):
+    super().__init__()
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.proximal_bias = proximal_bias
+    self.proximal_init = proximal_init
+
+    self.drop = nn.Dropout(p_dropout)
+    self.self_attn_layers = nn.ModuleList()
+    self.norm_layers_0 = nn.ModuleList()
+    self.encdec_attn_layers = nn.ModuleList()
+    self.norm_layers_1 = nn.ModuleList()
+    self.ffn_layers = nn.ModuleList()
+    self.norm_layers_2 = nn.ModuleList()
+    for i in range(self.n_layers):
+      self.self_attn_layers.append(MultiHeadAttention(hidden_channels, hidden_channels, n_heads, p_dropout=p_dropout, proximal_bias=proximal_bias, proximal_init=proximal_init))
+      self.norm_layers_0.append(LayerNorm(hidden_channels))
+      self.encdec_attn_layers.append(MultiHeadAttention(hidden_channels, hidden_channels, n_heads, p_dropout=p_dropout))
+      self.norm_layers_1.append(LayerNorm(hidden_channels))
+      self.ffn_layers.append(FFN(hidden_channels, hidden_channels, filter_channels, kernel_size, p_dropout=p_dropout, causal=True))
+      self.norm_layers_2.append(LayerNorm(hidden_channels))
+
+  def forward(self, x, x_mask, h, h_mask):
+    """
+    x: decoder input
+    h: encoder output
+    """
+    self_attn_mask = commons.subsequent_mask(x_mask.size(2)).to(device=x.device, dtype=x.dtype)
+    encdec_attn_mask = h_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+    x = x * x_mask
+    for i in range(self.n_layers):
+      y = self.self_attn_layers[i](x, x, self_attn_mask)
+      y = self.drop(y)
+      x = self.norm_layers_0[i](x + y)
+
+      y = self.encdec_attn_layers[i](x, h, encdec_attn_mask)
+      y = self.drop(y)
+      x = self.norm_layers_1[i](x + y)
+      
+      y = self.ffn_layers[i](x, x_mask)
+      y = self.drop(y)
+      x = self.norm_layers_2[i](x + y)
+    x = x * x_mask
+    return x
+
+
+class MultiHeadAttention(nn.Module):
+  def __init__(self, channels, out_channels, n_heads, p_dropout=0., window_size=None, heads_share=True, block_length=None, proximal_bias=False, proximal_init=False):
+    super().__init__()
+    assert channels % n_heads == 0
+
+    self.channels = channels
+    self.out_channels = out_channels
+    self.n_heads = n_heads
+    self.p_dropout = p_dropout
+    self.window_size = window_size
+    self.heads_share = heads_share
+    self.block_length = block_length
+    self.proximal_bias = proximal_bias
+    self.proximal_init = proximal_init
+    self.attn = None
+
+    self.k_channels = channels // n_heads
+    self.conv_q = nn.Conv1d(channels, channels, 1)
+    self.conv_k = nn.Conv1d(channels, channels, 1)
+    self.conv_v = nn.Conv1d(channels, channels, 1)
+    self.conv_o = nn.Conv1d(channels, out_channels, 1)
+    self.drop = nn.Dropout(p_dropout)
+
+    if window_size is not None:
+      n_heads_rel = 1 if heads_share else n_heads
+      rel_stddev = self.k_channels**-0.5
+      self.emb_rel_k = nn.Parameter(torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels) * rel_stddev)
+      self.emb_rel_v = nn.Parameter(torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels) * rel_stddev)
+
+    nn.init.xavier_uniform_(self.conv_q.weight)
+    nn.init.xavier_uniform_(self.conv_k.weight)
+    nn.init.xavier_uniform_(self.conv_v.weight)
+    if proximal_init:
+      with torch.no_grad():
+        self.conv_k.weight.copy_(self.conv_q.weight)
+        self.conv_k.bias.copy_(self.conv_q.bias)
+      
+  def forward(self, x, c, attn_mask=None):
+    q = self.conv_q(x)
+    k = self.conv_k(c)
+    v = self.conv_v(c)
+    
+    x, self.attn = self.attention(q, k, v, mask=attn_mask)
+
+    x = self.conv_o(x)
+    return x
+
+  def attention(self, query, key, value, mask=None):
+    # reshape [b, d, t] -> [b, n_h, t, d_k]
+    b, d, t_s, t_t = (*key.size(), query.size(2))
+    query = query.view(b, self.n_heads, self.k_channels, t_t).transpose(2, 3)
+    key = key.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
+    value = value.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
+
+    scores = torch.matmul(query / math.sqrt(self.k_channels), key.transpose(-2, -1))
+    if self.window_size is not None:
+      assert t_s == t_t, "Relative attention is only available for self-attention."
+      key_relative_embeddings = self._get_relative_embeddings(self.emb_rel_k, t_s)
+      rel_logits = self._matmul_with_relative_keys(query /math.sqrt(self.k_channels), key_relative_embeddings)
+      scores_local = self._relative_position_to_absolute_position(rel_logits)
+      scores = scores + scores_local
+    if self.proximal_bias:
+      assert t_s == t_t, "Proximal bias is only available for self-attention."
+      scores = scores + self._attention_bias_proximal(t_s).to(device=scores.device, dtype=scores.dtype)
+    if mask is not None:
+      scores = scores.masked_fill(mask == 0, -1e4)
+      if self.block_length is not None:
+        assert t_s == t_t, "Local attention is only available for self-attention."
+        block_mask = torch.ones_like(scores).triu(-self.block_length).tril(self.block_length)
+        scores = scores.masked_fill(block_mask == 0, -1e4)
+    p_attn = F.softmax(scores, dim=-1) # [b, n_h, t_t, t_s]
+    p_attn = self.drop(p_attn)
+    output = torch.matmul(p_attn, value)
+    if self.window_size is not None:
+      relative_weights = self._absolute_position_to_relative_position(p_attn)
+      value_relative_embeddings = self._get_relative_embeddings(self.emb_rel_v, t_s)
+      output = output + self._matmul_with_relative_values(relative_weights, value_relative_embeddings)
+    output = output.transpose(2, 3).contiguous().view(b, d, t_t) # [b, n_h, t_t, d_k] -> [b, d, t_t]
+    return output, p_attn
+
+  def _matmul_with_relative_values(self, x, y):
+    """
+    x: [b, h, l, m]
+    y: [h or 1, m, d]
+    ret: [b, h, l, d]
+    """
+    ret = torch.matmul(x, y.unsqueeze(0))
+    return ret
+
+  def _matmul_with_relative_keys(self, x, y):
+    """
+    x: [b, h, l, d]
+    y: [h or 1, m, d]
+    ret: [b, h, l, m]
+    """
+    ret = torch.matmul(x, y.unsqueeze(0).transpose(-2, -1))
+    return ret
+
+  def _get_relative_embeddings(self, relative_embeddings, length):
+    max_relative_position = 2 * self.window_size + 1
+    # Pad first before slice to avoid using cond ops.
+    pad_length = max(length - (self.window_size + 1), 0)
+    slice_start_position = max((self.window_size + 1) - length, 0)
+    slice_end_position = slice_start_position + 2 * length - 1
+    if pad_length > 0:
+      padded_relative_embeddings = F.pad(
+          relative_embeddings,
+          commons.convert_pad_shape([[0, 0], [pad_length, pad_length], [0, 0]]))
+    else:
+      padded_relative_embeddings = relative_embeddings
+    used_relative_embeddings = padded_relative_embeddings[:,slice_start_position:slice_end_position]
+    return used_relative_embeddings
+
+  def _relative_position_to_absolute_position(self, x):
+    """
+    x: [b, h, l, 2*l-1]
+    ret: [b, h, l, l]
+    """
+    batch, heads, length, _ = x.size()
+    # Concat columns of pad to shift from relative to absolute indexing.
+    x = F.pad(x, commons.convert_pad_shape([[0,0],[0,0],[0,0],[0,1]]))
+
+    # Concat extra elements so to add up to shape (len+1, 2*len-1).
+    x_flat = x.view([batch, heads, length * 2 * length])
+    x_flat = F.pad(x_flat, commons.convert_pad_shape([[0,0],[0,0],[0,length-1]]))
+
+    # Reshape and slice out the padded elements.
+    x_final = x_flat.view([batch, heads, length+1, 2*length-1])[:, :, :length, length-1:]
+    return x_final
+
+  def _absolute_position_to_relative_position(self, x):
+    """
+    x: [b, h, l, l]
+    ret: [b, h, l, 2*l-1]
+    """
+    batch, heads, length, _ = x.size()
+    # padd along column
+    x = F.pad(x, commons.convert_pad_shape([[0, 0], [0, 0], [0, 0], [0, length-1]]))
+    x_flat = x.view([batch, heads, length**2 + length*(length -1)])
+    # add 0's in the beginning that will skew the elements after reshape
+    x_flat = F.pad(x_flat, commons.convert_pad_shape([[0, 0], [0, 0], [length, 0]]))
+    x_final = x_flat.view([batch, heads, length, 2*length])[:,:,:,1:]
+    return x_final
+
+  def _attention_bias_proximal(self, length):
+    """Bias for self-attention to encourage attention to close positions.
+    Args:
+      length: an integer scalar.
+    Returns:
+      a Tensor with shape [1, 1, length, length]
+    """
+    r = torch.arange(length, dtype=torch.float32)
+    diff = torch.unsqueeze(r, 0) - torch.unsqueeze(r, 1)
+    return torch.unsqueeze(torch.unsqueeze(-torch.log1p(torch.abs(diff)), 0), 0)
+
+
+class FFN(nn.Module):
+  def __init__(self, in_channels, out_channels, filter_channels, kernel_size, p_dropout=0., activation=None, causal=False):
+    super().__init__()
+    self.in_channels = in_channels
+    self.out_channels = out_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.activation = activation
+    self.causal = causal
+
+    if causal:
+      self.padding = self._causal_padding
+    else:
+      self.padding = self._same_padding
+
+    self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size)
+    self.conv_2 = nn.Conv1d(filter_channels, out_channels, kernel_size)
+    self.drop = nn.Dropout(p_dropout)
+
+  def forward(self, x, x_mask):
+    x = self.conv_1(self.padding(x * x_mask))
+    if self.activation == "gelu":
+      x = x * torch.sigmoid(1.702 * x)
+    else:
+      x = torch.relu(x)
+    x = self.drop(x)
+    x = self.conv_2(self.padding(x * x_mask))
+    return x * x_mask
+  
+  def _causal_padding(self, x):
+    if self.kernel_size == 1:
+      return x
+    pad_l = self.kernel_size - 1
+    pad_r = 0
+    padding = [[0, 0], [0, 0], [pad_l, pad_r]]
+    x = F.pad(x, commons.convert_pad_shape(padding))
+    return x
+
+  def _same_padding(self, x):
+    if self.kernel_size == 1:
+      return x
+    pad_l = (self.kernel_size - 1) // 2
+    pad_r = self.kernel_size // 2
+    padding = [[0, 0], [0, 0], [pad_l, pad_r]]
+    x = F.pad(x, commons.convert_pad_shape(padding))
+    return x

commons.py

@@ -0,0 +1,172 @@
+import math
+import torch
+from torch.nn import functional as F
+import torch.jit
+
+
+def script_method(fn, _rcb=None):
+  return fn
+
+
+def script(obj, optimize=True, _frames_up=0, _rcb=None):
+  return obj
+
+
+torch.jit.script_method = script_method
+torch.jit.script = script
+
+
+def init_weights(m, mean=0.0, std=0.01):
+  classname = m.__class__.__name__
+  if classname.find("Conv") != -1:
+    m.weight.data.normal_(mean, std)
+
+
+def get_padding(kernel_size, dilation=1):
+  return int((kernel_size*dilation - dilation)/2)
+
+
+def convert_pad_shape(pad_shape):
+  l = pad_shape[::-1]
+  pad_shape = [item for sublist in l for item in sublist]
+  return pad_shape
+
+
+def intersperse(lst, item):
+  result = [item] * (len(lst) * 2 + 1)
+  result[1::2] = lst
+  return result
+
+
+def kl_divergence(m_p, logs_p, m_q, logs_q):
+  """KL(P||Q)"""
+  kl = (logs_q - logs_p) - 0.5
+  kl += 0.5 * (torch.exp(2. * logs_p) + ((m_p - m_q)**2)) * torch.exp(-2. * logs_q)
+  return kl
+
+
+def rand_gumbel(shape):
+  """Sample from the Gumbel distribution, protect from overflows."""
+  uniform_samples = torch.rand(shape) * 0.99998 + 0.00001
+  return -torch.log(-torch.log(uniform_samples))
+
+
+def rand_gumbel_like(x):
+  g = rand_gumbel(x.size()).to(dtype=x.dtype, device=x.device)
+  return g
+
+
+def slice_segments(x, ids_str, segment_size=4):
+  ret = torch.zeros_like(x[:, :, :segment_size])
+  for i in range(x.size(0)):
+    idx_str = ids_str[i]
+    idx_end = idx_str + segment_size
+    ret[i] = x[i, :, idx_str:idx_end]
+  return ret
+
+
+def rand_slice_segments(x, x_lengths=None, segment_size=4):
+  b, d, t = x.size()
+  if x_lengths is None:
+    x_lengths = t
+  ids_str_max = x_lengths - segment_size + 1
+  ids_str = (torch.rand([b]).to(device=x.device) * ids_str_max).to(dtype=torch.long)
+  ret = slice_segments(x, ids_str, segment_size)
+  return ret, ids_str
+
+
+def get_timing_signal_1d(
+    length, channels, min_timescale=1.0, max_timescale=1.0e4):
+  position = torch.arange(length, dtype=torch.float)
+  num_timescales = channels // 2
+  log_timescale_increment = (
+      math.log(float(max_timescale) / float(min_timescale)) /
+      (num_timescales - 1))
+  inv_timescales = min_timescale * torch.exp(
+      torch.arange(num_timescales, dtype=torch.float) * -log_timescale_increment)
+  scaled_time = position.unsqueeze(0) * inv_timescales.unsqueeze(1)
+  signal = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], 0)
+  signal = F.pad(signal, [0, 0, 0, channels % 2])
+  signal = signal.view(1, channels, length)
+  return signal
+
+
+def add_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4):
+  b, channels, length = x.size()
+  signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+  return x + signal.to(dtype=x.dtype, device=x.device)
+
+
+def cat_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4, axis=1):
+  b, channels, length = x.size()
+  signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+  return torch.cat([x, signal.to(dtype=x.dtype, device=x.device)], axis)
+
+
+def subsequent_mask(length):
+  mask = torch.tril(torch.ones(length, length)).unsqueeze(0).unsqueeze(0)
+  return mask
+
+
+@torch.jit.script
+def fused_add_tanh_sigmoid_multiply(input_a, input_b, n_channels):
+  n_channels_int = n_channels[0]
+  in_act = input_a + input_b
+  t_act = torch.tanh(in_act[:, :n_channels_int, :])
+  s_act = torch.sigmoid(in_act[:, n_channels_int:, :])
+  acts = t_act * s_act
+  return acts
+
+
+def convert_pad_shape(pad_shape):
+  l = pad_shape[::-1]
+  pad_shape = [item for sublist in l for item in sublist]
+  return pad_shape
+
+
+def shift_1d(x):
+  x = F.pad(x, convert_pad_shape([[0, 0], [0, 0], [1, 0]]))[:, :, :-1]
+  return x
+
+
+def sequence_mask(length, max_length=None):
+  if max_length is None:
+    max_length = length.max()
+  x = torch.arange(max_length, dtype=length.dtype, device=length.device)
+  return x.unsqueeze(0) < length.unsqueeze(1)
+
+
+def generate_path(duration, mask):
+  """
+  duration: [b, 1, t_x]
+  mask: [b, 1, t_y, t_x]
+  """
+  device = duration.device
+  
+  b, _, t_y, t_x = mask.shape
+  cum_duration = torch.cumsum(duration, -1)
+  
+  cum_duration_flat = cum_duration.view(b * t_x)
+  path = sequence_mask(cum_duration_flat, t_y).to(mask.dtype)
+  path = path.view(b, t_x, t_y)
+  path = path - F.pad(path, convert_pad_shape([[0, 0], [1, 0], [0, 0]]))[:, :-1]
+  path = path.unsqueeze(1).transpose(2,3) * mask
+  return path
+
+
+def clip_grad_value_(parameters, clip_value, norm_type=2):
+  if isinstance(parameters, torch.Tensor):
+    parameters = [parameters]
+  parameters = list(filter(lambda p: p.grad is not None, parameters))
+  norm_type = float(norm_type)
+  if clip_value is not None:
+    clip_value = float(clip_value)
+
+  total_norm = 0
+  for p in parameters:
+    param_norm = p.grad.data.norm(norm_type)
+    total_norm += param_norm.item() ** norm_type
+    if clip_value is not None:
+      p.grad.data.clamp_(min=-clip_value, max=clip_value)
+  total_norm = total_norm ** (1. / norm_type)
+  return total_norm

config.json

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

gitattributes.txt

@@ -0,0 +1,38 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+doom/cover.png filter=lfs diff=lfs merge=lfs -text
+zenyatta/cover.png filter=lfs diff=lfs merge=lfs -text
+pretrained_models/doom/cover.png filter=lfs diff=lfs merge=lfs -text
+pretrained_models/zenyatta/cover.png filter=lfs diff=lfs merge=lfs -text

gitignore.txt

@@ -0,0 +1,382 @@
+## Ignore Visual Studio temporary files, build results, and
+## files generated by popular Visual Studio add-ons.
+##
+## Get latest from https://github.com/github/gitignore/blob/master/VisualStudio.gitignore
+
+# User-specific files
+*.rsuser
+*.suo
+*.user
+*.userosscache
+*.sln.docstates
+
+# User-specific files (MonoDevelop/Xamarin Studio)
+*.userprefs
+
+# Mono auto generated files
+mono_crash.*
+
+# Build results
+[Dd]ebug/
+[Dd]ebugPublic/
+[Rr]elease/
+[Rr]eleases/
+x64/
+x86/
+[Ww][Ii][Nn]32/
+[Aa][Rr][Mm]/
+[Aa][Rr][Mm]64/
+bld/
+[Bb]in/
+[Oo]bj/
+[Oo]ut/
+[Ll]og/
+[Ll]ogs/
+
+# Visual Studio 2015/2017 cache/options directory
+.vs/
+# Uncomment if you have tasks that create the project's static files in wwwroot
+#wwwroot/
+
+# Visual Studio 2017 auto generated files
+Generated\ Files/
+
+# MSTest test Results
+[Tt]est[Rr]esult*/
+[Bb]uild[Ll]og.*
+
+# NUnit
+*.VisualState.xml
+TestResult.xml
+nunit-*.xml
+
+# Build Results of an ATL Project
+[Dd]ebugPS/
+[Rr]eleasePS/
+dlldata.c
+
+# Benchmark Results
+BenchmarkDotNet.Artifacts/
+
+# .NET Core
+project.lock.json
+project.fragment.lock.json
+artifacts/
+
+# ASP.NET Scaffolding
+ScaffoldingReadMe.txt
+
+# StyleCop
+StyleCopReport.xml
+
+# Files built by Visual Studio
+*_i.c
+*_p.c
+*_h.h
+*.ilk
+*.meta
+*.obj
+*.iobj
+*.pch
+*.pdb
+*.ipdb
+*.pgc
+*.pgd
+*.rsp
+*.sbr
+*.tlb
+*.tli
+*.tlh
+*.tmp
+*.tmp_proj
+*_wpftmp.csproj
+*.log
+*.vspscc
+*.vssscc
+.builds
+*.pidb
+*.svclog
+*.scc
+
+# Chutzpah Test files
+_Chutzpah*
+
+# Visual C++ cache files
+ipch/
+*.aps
+*.ncb
+*.opendb
+*.opensdf
+*.sdf
+*.cachefile
+*.VC.db
+*.VC.VC.opendb
+
+# Visual Studio profiler
+*.psess
+*.vsp
+*.vspx
+*.sap
+
+# Visual Studio Trace Files
+*.e2e
+
+# TFS 2012 Local Workspace
+$tf/
+
+# Guidance Automation Toolkit
+*.gpState
+
+# ReSharper is a .NET coding add-in
+_ReSharper*/
+*.[Rr]e[Ss]harper
+*.DotSettings.user
+
+# TeamCity is a build add-in
+_TeamCity*
+
+# DotCover is a Code Coverage Tool
+*.dotCover
+
+# AxoCover is a Code Coverage Tool
+.axoCover/*
+!.axoCover/settings.json
+
+# Coverlet is a free, cross platform Code Coverage Tool
+coverage*.json
+coverage*.xml
+coverage*.info
+
+# Visual Studio code coverage results
+*.coverage
+*.coveragexml
+
+# NCrunch
+_NCrunch_*
+.*crunch*.local.xml
+nCrunchTemp_*
+
+# MightyMoose
+*.mm.*
+AutoTest.Net/
+
+# Web workbench (sass)
+.sass-cache/
+
+# Installshield output folder
+[Ee]xpress/
+
+# DocProject is a documentation generator add-in
+DocProject/buildhelp/
+DocProject/Help/*.HxT
+DocProject/Help/*.HxC
+DocProject/Help/*.hhc
+DocProject/Help/*.hhk
+DocProject/Help/*.hhp
+DocProject/Help/Html2
+DocProject/Help/html
+
+# Click-Once directory
+publish/
+
+# Publish Web Output
+*.[Pp]ublish.xml
+*.azurePubxml
+# Note: Comment the next line if you want to checkin your web deploy settings,
+# but database connection strings (with potential passwords) will be unencrypted
+*.pubxml
+*.publishproj
+
+# Microsoft Azure Web App publish settings. Comment the next line if you want to
+# checkin your Azure Web App publish settings, but sensitive information contained
+# in these scripts will be unencrypted
+PublishScripts/
+
+# NuGet Packages
+*.nupkg
+# NuGet Symbol Packages
+*.snupkg
+# The packages folder can be ignored because of Package Restore
+**/[Pp]ackages/*
+# except build/, which is used as an MSBuild target.
+!**/[Pp]ackages/build/
+# Uncomment if necessary however generally it will be regenerated when needed
+#!**/[Pp]ackages/repositories.config
+# NuGet v3's project.json files produces more ignorable files
+*.nuget.props
+*.nuget.targets
+
+# Microsoft Azure Build Output
+csx/
+*.build.csdef
+
+# Microsoft Azure Emulator
+ecf/
+rcf/
+
+# Windows Store app package directories and files
+AppPackages/
+BundleArtifacts/
+Package.StoreAssociation.xml
+_pkginfo.txt
+*.appx
+*.appxbundle
+*.appxupload
+
+# Visual Studio cache files
+# files ending in .cache can be ignored
+*.[Cc]ache
+# but keep track of directories ending in .cache
+!?*.[Cc]ache/
+
+# Others
+ClientBin/
+~$*
+*~
+*.dbmdl
+*.dbproj.schemaview
+*.jfm
+*.pfx
+*.publishsettings
+orleans.codegen.cs
+
+# Including strong name files can present a security risk
+# (https://github.com/github/gitignore/pull/2483#issue-259490424)
+#*.snk
+
+# Since there are multiple workflows, uncomment next line to ignore bower_components
+# (https://github.com/github/gitignore/pull/1529#issuecomment-104372622)
+#bower_components/
+
+# RIA/Silverlight projects
+Generated_Code/
+
+# Backup & report files from converting an old project file
+# to a newer Visual Studio version. Backup files are not needed,
+# because we have git ;-)
+_UpgradeReport_Files/
+Backup*/
+UpgradeLog*.XML
+UpgradeLog*.htm
+ServiceFabricBackup/
+*.rptproj.bak
+
+# SQL Server files
+*.mdf
+*.ldf
+*.ndf
+
+# Business Intelligence projects
+*.rdl.data
+*.bim.layout
+*.bim_*.settings
+*.rptproj.rsuser
+*- [Bb]ackup.rdl
+*- [Bb]ackup ([0-9]).rdl
+*- [Bb]ackup ([0-9][0-9]).rdl
+
+# Microsoft Fakes
+FakesAssemblies/
+
+# GhostDoc plugin setting file
+*.GhostDoc.xml
+
+# Node.js Tools for Visual Studio
+.ntvs_analysis.dat
+node_modules/
+
+# Visual Studio 6 build log
+*.plg
+
+# Visual Studio 6 workspace options file
+*.opt
+
+# Visual Studio 6 auto-generated workspace file (contains which files were open etc.)
+*.vbw
+
+# Visual Studio LightSwitch build output
+**/*.HTMLClient/GeneratedArtifacts
+**/*.DesktopClient/GeneratedArtifacts
+**/*.DesktopClient/ModelManifest.xml
+**/*.Server/GeneratedArtifacts
+**/*.Server/ModelManifest.xml
+_Pvt_Extensions
+
+# Paket dependency manager
+.paket/paket.exe
+paket-files/
+
+# FAKE - F# Make
+.fake/
+
+# CodeRush personal settings
+.cr/personal
+
+# Python Tools for Visual Studio (PTVS)
+__pycache__/
+
+
+# Cake - Uncomment if you are using it
+# tools/**
+# !tools/packages.config
+
+# Tabs Studio
+*.tss
+
+# Telerik's JustMock configuration file
+*.jmconfig
+
+# BizTalk build output
+*.btp.cs
+*.btm.cs
+*.odx.cs
+*.xsd.cs
+
+# OpenCover UI analysis results
+OpenCover/
+
+# Azure Stream Analytics local run output
+ASALocalRun/
+
+# MSBuild Binary and Structured Log
+*.binlog
+
+# NVidia Nsight GPU debugger configuration file
+*.nvuser
+
+# MFractors (Xamarin productivity tool) working folder
+.mfractor/
+
+# Local History for Visual Studio
+.localhistory/
+
+# BeatPulse healthcheck temp database
+healthchecksdb
+
+# Backup folder for Package Reference Convert tool in Visual Studio 2017
+MigrationBackup/
+
+# Ionide (cross platform F# VS Code tools) working folder
+.ionide/
+
+# Fody - auto-generated XML schema
+FodyWeavers.xsd
+
+# build
+build
+monotonic_align/core.c
+*.o
+*.so
+*.dll
+
+# data
+/config.json
+/*.pth
+*.wav
+/monotonic_align/monotonic_align
+/resources
+/MoeGoe.spec
+/dist/MoeGoe
+/dist
+
+.idea

mel_processing.py

@@ -0,0 +1,101 @@
+import torch
+import torch.utils.data
+from librosa.filters import mel as librosa_mel_fn
+
+MAX_WAV_VALUE = 32768.0
+
+
+def dynamic_range_compression_torch(x, C=1, clip_val=1e-5):
+    """
+    PARAMS
+    ------
+    C: compression factor
+    """
+    return torch.log(torch.clamp(x, min=clip_val) * C)
+
+
+def dynamic_range_decompression_torch(x, C=1):
+    """
+    PARAMS
+    ------
+    C: compression factor used to compress
+    """
+    return torch.exp(x) / C
+
+
+def spectral_normalize_torch(magnitudes):
+    output = dynamic_range_compression_torch(magnitudes)
+    return output
+
+
+def spectral_de_normalize_torch(magnitudes):
+    output = dynamic_range_decompression_torch(magnitudes)
+    return output
+
+
+mel_basis = {}
+hann_window = {}
+
+
+def spectrogram_torch(y, n_fft, sampling_rate, hop_size, win_size, center=False):
+    if torch.min(y) < -1.:
+        print('min value is ', torch.min(y))
+    if torch.max(y) > 1.:
+        print('max value is ', torch.max(y))
+
+    global hann_window
+    dtype_device = str(y.dtype) + '_' + str(y.device)
+    wnsize_dtype_device = str(win_size) + '_' + dtype_device
+    if wnsize_dtype_device not in hann_window:
+        hann_window[wnsize_dtype_device] = torch.hann_window(win_size).to(dtype=y.dtype, device=y.device)
+
+    y = torch.nn.functional.pad(y.unsqueeze(1), (int((n_fft-hop_size)/2), int((n_fft-hop_size)/2)), mode='reflect')
+    y = y.squeeze(1)
+
+    spec = torch.stft(y, n_fft, hop_length=hop_size, win_length=win_size, window=hann_window[wnsize_dtype_device],
+                      center=center, pad_mode='reflect', normalized=False, onesided=True, return_complex=False)
+
+    spec = torch.sqrt(spec.pow(2).sum(-1) + 1e-6)
+    return spec
+
+
+def spec_to_mel_torch(spec, n_fft, num_mels, sampling_rate, fmin, fmax):
+    global mel_basis
+    dtype_device = str(spec.dtype) + '_' + str(spec.device)
+    fmax_dtype_device = str(fmax) + '_' + dtype_device
+    if fmax_dtype_device not in mel_basis:
+        mel = librosa_mel_fn(sampling_rate, n_fft, num_mels, fmin, fmax)
+        mel_basis[fmax_dtype_device] = torch.from_numpy(mel).to(dtype=spec.dtype, device=spec.device)
+    spec = torch.matmul(mel_basis[fmax_dtype_device], spec)
+    spec = spectral_normalize_torch(spec)
+    return spec
+
+
+def mel_spectrogram_torch(y, n_fft, num_mels, sampling_rate, hop_size, win_size, fmin, fmax, center=False):
+    if torch.min(y) < -1.:
+        print('min value is ', torch.min(y))
+    if torch.max(y) > 1.:
+        print('max value is ', torch.max(y))
+
+    global mel_basis, hann_window
+    dtype_device = str(y.dtype) + '_' + str(y.device)
+    fmax_dtype_device = str(fmax) + '_' + dtype_device
+    wnsize_dtype_device = str(win_size) + '_' + dtype_device
+    if fmax_dtype_device not in mel_basis:
+        mel = librosa_mel_fn(sampling_rate, n_fft, num_mels, fmin, fmax)
+        mel_basis[fmax_dtype_device] = torch.from_numpy(mel).to(dtype=y.dtype, device=y.device)
+    if wnsize_dtype_device not in hann_window:
+        hann_window[wnsize_dtype_device] = torch.hann_window(win_size).to(dtype=y.dtype, device=y.device)
+
+    y = torch.nn.functional.pad(y.unsqueeze(1), (int((n_fft-hop_size)/2), int((n_fft-hop_size)/2)), mode='reflect')
+    y = y.squeeze(1)
+
+    spec = torch.stft(y, n_fft, hop_length=hop_size, win_length=win_size, window=hann_window[wnsize_dtype_device],
+                      center=center, pad_mode='reflect', normalized=False, onesided=True)
+
+    spec = torch.sqrt(spec.pow(2).sum(-1) + 1e-6)
+
+    spec = torch.matmul(mel_basis[fmax_dtype_device], spec)
+    spec = spectral_normalize_torch(spec)
+
+    return spec

models.py

@@ -0,0 +1,533 @@
+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 commons import init_weights, get_padding
+
+
+class StochasticDurationPredictor(nn.Module):
+  def __init__(self, in_channels, filter_channels, kernel_size, p_dropout, n_flows=4, gin_channels=0):
+    super().__init__()
+    filter_channels = in_channels # it needs to be removed from future version.
+    self.in_channels = in_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.n_flows = n_flows
+    self.gin_channels = gin_channels
+
+    self.log_flow = modules.Log()
+    self.flows = nn.ModuleList()
+    self.flows.append(modules.ElementwiseAffine(2))
+    for i in range(n_flows):
+      self.flows.append(modules.ConvFlow(2, filter_channels, kernel_size, n_layers=3))
+      self.flows.append(modules.Flip())
+
+    self.post_pre = nn.Conv1d(1, filter_channels, 1)
+    self.post_proj = nn.Conv1d(filter_channels, filter_channels, 1)
+    self.post_convs = modules.DDSConv(filter_channels, kernel_size, n_layers=3, p_dropout=p_dropout)
+    self.post_flows = nn.ModuleList()
+    self.post_flows.append(modules.ElementwiseAffine(2))
+    for i in range(4):
+      self.post_flows.append(modules.ConvFlow(2, filter_channels, kernel_size, n_layers=3))
+      self.post_flows.append(modules.Flip())
+
+    self.pre = nn.Conv1d(in_channels, filter_channels, 1)
+    self.proj = nn.Conv1d(filter_channels, filter_channels, 1)
+    self.convs = modules.DDSConv(filter_channels, kernel_size, n_layers=3, p_dropout=p_dropout)
+    if gin_channels != 0:
+      self.cond = nn.Conv1d(gin_channels, filter_channels, 1)
+
+  def forward(self, x, x_mask, w=None, g=None, reverse=False, noise_scale=1.0):
+    x = torch.detach(x)
+    x = self.pre(x)
+    if g is not None:
+      g = torch.detach(g)
+      x = x + self.cond(g)
+    x = self.convs(x, x_mask)
+    x = self.proj(x) * x_mask
+
+    if not reverse:
+      flows = self.flows
+      assert w is not None
+
+      logdet_tot_q = 0 
+      h_w = self.post_pre(w)
+      h_w = self.post_convs(h_w, x_mask)
+      h_w = self.post_proj(h_w) * x_mask
+      e_q = torch.randn(w.size(0), 2, w.size(2)).to(device=x.device, dtype=x.dtype) * x_mask
+      z_q = e_q
+      for flow in self.post_flows:
+        z_q, logdet_q = flow(z_q, x_mask, g=(x + h_w))
+        logdet_tot_q += logdet_q
+      z_u, z1 = torch.split(z_q, [1, 1], 1) 
+      u = torch.sigmoid(z_u) * x_mask
+      z0 = (w - u) * x_mask
+      logdet_tot_q += torch.sum((F.logsigmoid(z_u) + F.logsigmoid(-z_u)) * x_mask, [1,2])
+      logq = torch.sum(-0.5 * (math.log(2*math.pi) + (e_q**2)) * x_mask, [1,2]) - logdet_tot_q
+
+      logdet_tot = 0
+      z0, logdet = self.log_flow(z0, x_mask)
+      logdet_tot += logdet
+      z = torch.cat([z0, z1], 1)
+      for flow in flows:
+        z, logdet = flow(z, x_mask, g=x, reverse=reverse)
+        logdet_tot = logdet_tot + logdet
+      nll = torch.sum(0.5 * (math.log(2*math.pi) + (z**2)) * x_mask, [1,2]) - logdet_tot
+      return nll + logq # [b]
+    else:
+      flows = list(reversed(self.flows))
+      flows = flows[:-2] + [flows[-1]] # remove a useless vflow
+      z = torch.randn(x.size(0), 2, x.size(2)).to(device=x.device, dtype=x.dtype) * noise_scale
+      for flow in flows:
+        z = flow(z, x_mask, g=x, reverse=reverse)
+      z0, z1 = torch.split(z, [1, 1], 1)
+      logw = z0
+      return logw
+
+
+class DurationPredictor(nn.Module):
+  def __init__(self, in_channels, filter_channels, kernel_size, p_dropout, gin_channels=0):
+    super().__init__()
+
+    self.in_channels = in_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.gin_channels = gin_channels
+
+    self.drop = nn.Dropout(p_dropout)
+    self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size, padding=kernel_size//2)
+    self.norm_1 = modules.LayerNorm(filter_channels)
+    self.conv_2 = nn.Conv1d(filter_channels, filter_channels, kernel_size, padding=kernel_size//2)
+    self.norm_2 = modules.LayerNorm(filter_channels)
+    self.proj = nn.Conv1d(filter_channels, 1, 1)
+
+    if gin_channels != 0:
+      self.cond = nn.Conv1d(gin_channels, in_channels, 1)
+
+  def forward(self, x, x_mask, g=None):
+    x = torch.detach(x)
+    if g is not None:
+      g = torch.detach(g)
+      x = x + self.cond(g)
+    x = self.conv_1(x * x_mask)
+    x = torch.relu(x)
+    x = self.norm_1(x)
+    x = self.drop(x)
+    x = self.conv_2(x * x_mask)
+    x = torch.relu(x)
+    x = self.norm_2(x)
+    x = self.drop(x)
+    x = self.proj(x * x_mask)
+    return x * x_mask
+
+
+class TextEncoder(nn.Module):
+  def __init__(self,
+      n_vocab,
+      out_channels,
+      hidden_channels,
+      filter_channels,
+      n_heads,
+      n_layers,
+      kernel_size,
+      p_dropout):
+    super().__init__()
+    self.n_vocab = n_vocab
+    self.out_channels = out_channels
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+
+    self.emb = nn.Embedding(n_vocab, hidden_channels)
+    nn.init.normal_(self.emb.weight, 0.0, hidden_channels**-0.5)
+
+    self.encoder = attentions.Encoder(
+      hidden_channels,
+      filter_channels,
+      n_heads,
+      n_layers,
+      kernel_size,
+      p_dropout)
+    self.proj= nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+  def forward(self, x, x_lengths):
+    x = self.emb(x) * math.sqrt(self.hidden_channels) # [b, t, h]
+    x = torch.transpose(x, 1, -1) # [b, h, t]
+    x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(x.dtype)
+
+    x = self.encoder(x * x_mask, x_mask)
+    stats = self.proj(x) * x_mask
+
+    m, logs = torch.split(stats, self.out_channels, dim=1)
+    return x, m, logs, x_mask
+
+
+class ResidualCouplingBlock(nn.Module):
+  def __init__(self,
+      channels,
+      hidden_channels,
+      kernel_size,
+      dilation_rate,
+      n_layers,
+      n_flows=4,
+      gin_channels=0):
+    super().__init__()
+    self.channels = channels
+    self.hidden_channels = hidden_channels
+    self.kernel_size = kernel_size
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.n_flows = n_flows
+    self.gin_channels = gin_channels
+
+    self.flows = nn.ModuleList()
+    for i in range(n_flows):
+      self.flows.append(modules.ResidualCouplingLayer(channels, hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=gin_channels, mean_only=True))
+      self.flows.append(modules.Flip())
+
+  def forward(self, x, x_mask, g=None, reverse=False):
+    if not reverse:
+      for flow in self.flows:
+        x, _ = flow(x, x_mask, g=g, reverse=reverse)
+    else:
+      for flow in reversed(self.flows):
+        x = flow(x, x_mask, g=g, reverse=reverse)
+    return x
+
+
+class PosteriorEncoder(nn.Module):
+  def __init__(self,
+      in_channels,
+      out_channels,
+      hidden_channels,
+      kernel_size,
+      dilation_rate,
+      n_layers,
+      gin_channels=0):
+    super().__init__()
+    self.in_channels = in_channels
+    self.out_channels = out_channels
+    self.hidden_channels = hidden_channels
+    self.kernel_size = kernel_size
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.gin_channels = gin_channels
+
+    self.pre = nn.Conv1d(in_channels, hidden_channels, 1)
+    self.enc = modules.WN(hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=gin_channels)
+    self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+  def forward(self, x, x_lengths, g=None):
+    x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(x.dtype)
+    x = self.pre(x) * x_mask
+    x = self.enc(x, x_mask, g=g)
+    stats = self.proj(x) * x_mask
+    m, logs = torch.split(stats, self.out_channels, dim=1)
+    z = (m + torch.randn_like(m) * torch.exp(logs)) * x_mask
+    return z, m, logs, x_mask
+
+
+class Generator(torch.nn.Module):
+    def __init__(self, initial_channel, resblock, resblock_kernel_sizes, resblock_dilation_sizes, upsample_rates, upsample_initial_channel, upsample_kernel_sizes, gin_channels=0):
+        super(Generator, self).__init__()
+        self.num_kernels = len(resblock_kernel_sizes)
+        self.num_upsamples = len(upsample_rates)
+        self.conv_pre = Conv1d(initial_channel, upsample_initial_channel, 7, 1, padding=3)
+        resblock = modules.ResBlock1 if resblock == '1' else modules.ResBlock2
+
+        self.ups = nn.ModuleList()
+        for i, (u, k) in enumerate(zip(upsample_rates, upsample_kernel_sizes)):
+            self.ups.append(weight_norm(
+                ConvTranspose1d(upsample_initial_channel//(2**i), upsample_initial_channel//(2**(i+1)),
+                                k, u, padding=(k-u)//2)))
+
+        self.resblocks = nn.ModuleList()
+        for i in range(len(self.ups)):
+            ch = upsample_initial_channel//(2**(i+1))
+            for j, (k, d) in enumerate(zip(resblock_kernel_sizes, resblock_dilation_sizes)):
+                self.resblocks.append(resblock(ch, k, d))
+
+        self.conv_post = Conv1d(ch, 1, 7, 1, padding=3, bias=False)
+        self.ups.apply(init_weights)
+
+        if gin_channels != 0:
+            self.cond = nn.Conv1d(gin_channels, upsample_initial_channel, 1)
+
+    def forward(self, x, g=None):
+        x = self.conv_pre(x)
+        if g is not None:
+          x = x + self.cond(g)
+
+        for i in range(self.num_upsamples):
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            x = self.ups[i](x)
+            xs = None
+            for j in range(self.num_kernels):
+                if xs is None:
+                    xs = self.resblocks[i*self.num_kernels+j](x)
+                else:
+                    xs += self.resblocks[i*self.num_kernels+j](x)
+            x = xs / self.num_kernels
+        x = F.leaky_relu(x)
+        x = self.conv_post(x)
+        x = torch.tanh(x)
+
+        return x
+
+    def remove_weight_norm(self):
+        print('Removing weight norm...')
+        for l in self.ups:
+            remove_weight_norm(l)
+        for l in self.resblocks:
+            l.remove_weight_norm()
+
+
+class DiscriminatorP(torch.nn.Module):
+    def __init__(self, period, kernel_size=5, stride=3, use_spectral_norm=False):
+        super(DiscriminatorP, self).__init__()
+        self.period = period
+        self.use_spectral_norm = use_spectral_norm
+        norm_f = weight_norm if use_spectral_norm == False else spectral_norm
+        self.convs = nn.ModuleList([
+            norm_f(Conv2d(1, 32, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(32, 128, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(128, 512, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(512, 1024, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
+            norm_f(Conv2d(1024, 1024, (kernel_size, 1), 1, padding=(get_padding(kernel_size, 1), 0))),
+        ])
+        self.conv_post = norm_f(Conv2d(1024, 1, (3, 1), 1, padding=(1, 0)))
+
+    def forward(self, x):
+        fmap = []
+
+        # 1d to 2d
+        b, c, t = x.shape
+        if t % self.period != 0: # pad first
+            n_pad = self.period - (t % self.period)
+            x = F.pad(x, (0, n_pad), "reflect")
+            t = t + n_pad
+        x = x.view(b, c, t // self.period, self.period)
+
+        for l in self.convs:
+            x = l(x)
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            fmap.append(x)
+        x = self.conv_post(x)
+        fmap.append(x)
+        x = torch.flatten(x, 1, -1)
+
+        return x, fmap
+
+
+class DiscriminatorS(torch.nn.Module):
+    def __init__(self, use_spectral_norm=False):
+        super(DiscriminatorS, self).__init__()
+        norm_f = weight_norm if use_spectral_norm == False else spectral_norm
+        self.convs = nn.ModuleList([
+            norm_f(Conv1d(1, 16, 15, 1, padding=7)),
+            norm_f(Conv1d(16, 64, 41, 4, groups=4, padding=20)),
+            norm_f(Conv1d(64, 256, 41, 4, groups=16, padding=20)),
+            norm_f(Conv1d(256, 1024, 41, 4, groups=64, padding=20)),
+            norm_f(Conv1d(1024, 1024, 41, 4, groups=256, padding=20)),
+            norm_f(Conv1d(1024, 1024, 5, 1, padding=2)),
+        ])
+        self.conv_post = norm_f(Conv1d(1024, 1, 3, 1, padding=1))
+
+    def forward(self, x):
+        fmap = []
+
+        for l in self.convs:
+            x = l(x)
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            fmap.append(x)
+        x = self.conv_post(x)
+        fmap.append(x)
+        x = torch.flatten(x, 1, -1)
+
+        return x, fmap
+
+
+class MultiPeriodDiscriminator(torch.nn.Module):
+    def __init__(self, use_spectral_norm=False):
+        super(MultiPeriodDiscriminator, self).__init__()
+        periods = [2,3,5,7,11]
+
+        discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
+        discs = discs + [DiscriminatorP(i, use_spectral_norm=use_spectral_norm) for i in periods]
+        self.discriminators = nn.ModuleList(discs)
+
+    def forward(self, y, y_hat):
+        y_d_rs = []
+        y_d_gs = []
+        fmap_rs = []
+        fmap_gs = []
+        for i, d in enumerate(self.discriminators):
+            y_d_r, fmap_r = d(y)
+            y_d_g, fmap_g = d(y_hat)
+            y_d_rs.append(y_d_r)
+            y_d_gs.append(y_d_g)
+            fmap_rs.append(fmap_r)
+            fmap_gs.append(fmap_g)
+
+        return y_d_rs, y_d_gs, fmap_rs, fmap_gs
+
+
+
+class SynthesizerTrn(nn.Module):
+  """
+  Synthesizer for Training
+  """
+
+  def __init__(self, 
+    n_vocab,
+    spec_channels,
+    segment_size,
+    inter_channels,
+    hidden_channels,
+    filter_channels,
+    n_heads,
+    n_layers,
+    kernel_size,
+    p_dropout,
+    resblock, 
+    resblock_kernel_sizes, 
+    resblock_dilation_sizes, 
+    upsample_rates, 
+    upsample_initial_channel, 
+    upsample_kernel_sizes,
+    n_speakers=0,
+    gin_channels=0,
+    use_sdp=True,
+    **kwargs):
+
+    super().__init__()
+    self.n_vocab = n_vocab
+    self.spec_channels = spec_channels
+    self.inter_channels = inter_channels
+    self.hidden_channels = hidden_channels
+    self.filter_channels = filter_channels
+    self.n_heads = n_heads
+    self.n_layers = n_layers
+    self.kernel_size = kernel_size
+    self.p_dropout = p_dropout
+    self.resblock = resblock
+    self.resblock_kernel_sizes = resblock_kernel_sizes
+    self.resblock_dilation_sizes = resblock_dilation_sizes
+    self.upsample_rates = upsample_rates
+    self.upsample_initial_channel = upsample_initial_channel
+    self.upsample_kernel_sizes = upsample_kernel_sizes
+    self.segment_size = segment_size
+    self.n_speakers = n_speakers
+    self.gin_channels = gin_channels
+
+    self.use_sdp = use_sdp
+
+    self.enc_p = TextEncoder(n_vocab,
+        inter_channels,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size,
+        p_dropout)
+    self.dec = Generator(inter_channels, resblock, resblock_kernel_sizes, resblock_dilation_sizes, upsample_rates, upsample_initial_channel, upsample_kernel_sizes, gin_channels=gin_channels)
+    self.enc_q = PosteriorEncoder(spec_channels, inter_channels, hidden_channels, 5, 1, 16, gin_channels=gin_channels)
+    self.flow = ResidualCouplingBlock(inter_channels, hidden_channels, 5, 1, 4, gin_channels=gin_channels)
+
+    if use_sdp:
+      self.dp = StochasticDurationPredictor(hidden_channels, 192, 3, 0.5, 4, gin_channels=gin_channels)
+    else:
+      self.dp = DurationPredictor(hidden_channels, 256, 3, 0.5, gin_channels=gin_channels)
+
+    if n_speakers > 1:
+      self.emb_g = nn.Embedding(n_speakers, gin_channels)
+
+  def forward(self, x, x_lengths, y, y_lengths, sid=None):
+
+    x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths)
+    if self.n_speakers > 0:
+      g = self.emb_g(sid).unsqueeze(-1) # [b, h, 1]
+    else:
+      g = None
+
+    z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
+    z_p = self.flow(z, y_mask, g=g)
+
+    with torch.no_grad():
+      # negative cross-entropy
+      s_p_sq_r = torch.exp(-2 * logs_p) # [b, d, t]
+      neg_cent1 = torch.sum(-0.5 * math.log(2 * math.pi) - logs_p, [1], keepdim=True) # [b, 1, t_s]
+      neg_cent2 = torch.matmul(-0.5 * (z_p ** 2).transpose(1, 2), s_p_sq_r) # [b, t_t, d] x [b, d, t_s] = [b, t_t, t_s]
+      neg_cent3 = torch.matmul(z_p.transpose(1, 2), (m_p * s_p_sq_r)) # [b, t_t, d] x [b, d, t_s] = [b, t_t, t_s]
+      neg_cent4 = torch.sum(-0.5 * (m_p ** 2) * s_p_sq_r, [1], keepdim=True) # [b, 1, t_s]
+      neg_cent = neg_cent1 + neg_cent2 + neg_cent3 + neg_cent4
+
+      attn_mask = torch.unsqueeze(x_mask, 2) * torch.unsqueeze(y_mask, -1)
+      attn = monotonic_align.maximum_path(neg_cent, attn_mask.squeeze(1)).unsqueeze(1).detach()
+
+    w = attn.sum(2)
+    if self.use_sdp:
+      l_length = self.dp(x, x_mask, w, g=g)
+      l_length = l_length / torch.sum(x_mask)
+    else:
+      logw_ = torch.log(w + 1e-6) * x_mask
+      logw = self.dp(x, x_mask, g=g)
+      l_length = torch.sum((logw - logw_)**2, [1,2]) / torch.sum(x_mask) # for averaging 
+
+    # expand prior
+    m_p = torch.matmul(attn.squeeze(1), m_p.transpose(1, 2)).transpose(1, 2)
+    logs_p = torch.matmul(attn.squeeze(1), logs_p.transpose(1, 2)).transpose(1, 2)
+
+    z_slice, ids_slice = commons.rand_slice_segments(z, y_lengths, self.segment_size)
+    o = self.dec(z_slice, g=g)
+    return o, l_length, attn, ids_slice, x_mask, y_mask, (z, z_p, m_p, logs_p, m_q, logs_q)
+
+  def infer(self, x, x_lengths, sid=None, noise_scale=1, length_scale=1, noise_scale_w=1., max_len=None):
+    x, m_p, logs_p, x_mask = self.enc_p(x, x_lengths)
+    if self.n_speakers > 0:
+      g = self.emb_g(sid).unsqueeze(-1) # [b, h, 1]
+    else:
+      g = None
+
+    if self.use_sdp:
+      logw = self.dp(x, x_mask, g=g, reverse=True, noise_scale=noise_scale_w)
+    else:
+      logw = self.dp(x, x_mask, g=g)
+    w = torch.exp(logw) * x_mask * length_scale
+    w_ceil = torch.ceil(w)
+    y_lengths = torch.clamp_min(torch.sum(w_ceil, [1, 2]), 1).long()
+    y_mask = torch.unsqueeze(commons.sequence_mask(y_lengths, None), 1).to(x_mask.dtype)
+    attn_mask = torch.unsqueeze(x_mask, 2) * torch.unsqueeze(y_mask, -1)
+    attn = commons.generate_path(w_ceil, attn_mask)
+
+    m_p = torch.matmul(attn.squeeze(1), m_p.transpose(1, 2)).transpose(1, 2) # [b, t', t], [b, t, d] -> [b, d, t']
+    logs_p = torch.matmul(attn.squeeze(1), logs_p.transpose(1, 2)).transpose(1, 2) # [b, t', t], [b, t, d] -> [b, d, t']
+
+    z_p = m_p + torch.randn_like(m_p) * torch.exp(logs_p) * noise_scale
+    z = self.flow(z_p, y_mask, g=g, reverse=True)
+    o = self.dec((z * y_mask)[:,:,:max_len], g=g)
+    return o, attn, y_mask, (z, z_p, m_p, logs_p)
+
+  def voice_conversion(self, y, y_lengths, sid_src, sid_tgt):
+    assert self.n_speakers > 0, "n_speakers have to be larger than 0."
+    g_src = self.emb_g(sid_src).unsqueeze(-1)
+    g_tgt = self.emb_g(sid_tgt).unsqueeze(-1)
+    z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g_src)
+    z_p = self.flow(z, y_mask, g=g_src)
+    z_hat = self.flow(z_p, y_mask, g=g_tgt, reverse=True)
+    o_hat = self.dec(z_hat * y_mask, g=g_tgt)
+    return o_hat, y_mask, (z, z_p, z_hat)
+

modules.py

@@ -0,0 +1,388 @@
+import math
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+from torch.nn import Conv1d, ConvTranspose1d, AvgPool1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm
+
+import commons
+from commons import init_weights, get_padding
+from transforms import piecewise_rational_quadratic_transform
+
+
+LRELU_SLOPE = 0.1
+
+
+class LayerNorm(nn.Module):
+  def __init__(self, channels, eps=1e-5):
+    super().__init__()
+    self.channels = channels
+    self.eps = eps
+
+    self.gamma = nn.Parameter(torch.ones(channels))
+    self.beta = nn.Parameter(torch.zeros(channels))
+
+  def forward(self, x):
+    x = x.transpose(1, -1)
+    x = F.layer_norm(x, (self.channels,), self.gamma, self.beta, self.eps)
+    return x.transpose(1, -1)
+
+ 
+class ConvReluNorm(nn.Module):
+  def __init__(self, in_channels, hidden_channels, out_channels, kernel_size, n_layers, p_dropout):
+    super().__init__()
+    self.in_channels = in_channels
+    self.hidden_channels = hidden_channels
+    self.out_channels = out_channels
+    self.kernel_size = kernel_size
+    self.n_layers = n_layers
+    self.p_dropout = p_dropout
+    assert n_layers > 1, "Number of layers should be larger than 0."
+
+    self.conv_layers = nn.ModuleList()
+    self.norm_layers = nn.ModuleList()
+    self.conv_layers.append(nn.Conv1d(in_channels, hidden_channels, kernel_size, padding=kernel_size//2))
+    self.norm_layers.append(LayerNorm(hidden_channels))
+    self.relu_drop = nn.Sequential(
+        nn.ReLU(),
+        nn.Dropout(p_dropout))
+    for _ in range(n_layers-1):
+      self.conv_layers.append(nn.Conv1d(hidden_channels, hidden_channels, kernel_size, padding=kernel_size//2))
+      self.norm_layers.append(LayerNorm(hidden_channels))
+    self.proj = nn.Conv1d(hidden_channels, out_channels, 1)
+    self.proj.weight.data.zero_()
+    self.proj.bias.data.zero_()
+
+  def forward(self, x, x_mask):
+    x_org = x
+    for i in range(self.n_layers):
+      x = self.conv_layers[i](x * x_mask)
+      x = self.norm_layers[i](x)
+      x = self.relu_drop(x)
+    x = x_org + self.proj(x)
+    return x * x_mask
+
+
+class DDSConv(nn.Module):
+  """
+  Dialted and Depth-Separable Convolution
+  """
+  def __init__(self, channels, kernel_size, n_layers, p_dropout=0.):
+    super().__init__()
+    self.channels = channels
+    self.kernel_size = kernel_size
+    self.n_layers = n_layers
+    self.p_dropout = p_dropout
+
+    self.drop = nn.Dropout(p_dropout)
+    self.convs_sep = nn.ModuleList()
+    self.convs_1x1 = nn.ModuleList()
+    self.norms_1 = nn.ModuleList()
+    self.norms_2 = nn.ModuleList()
+    for i in range(n_layers):
+      dilation = kernel_size ** i
+      padding = (kernel_size * dilation - dilation) // 2
+      self.convs_sep.append(nn.Conv1d(channels, channels, kernel_size, 
+          groups=channels, dilation=dilation, padding=padding
+      ))
+      self.convs_1x1.append(nn.Conv1d(channels, channels, 1))
+      self.norms_1.append(LayerNorm(channels))
+      self.norms_2.append(LayerNorm(channels))
+
+  def forward(self, x, x_mask, g=None):
+    if g is not None:
+      x = x + g
+    for i in range(self.n_layers):
+      y = self.convs_sep[i](x * x_mask)
+      y = self.norms_1[i](y)
+      y = F.gelu(y)
+      y = self.convs_1x1[i](y)
+      y = self.norms_2[i](y)
+      y = F.gelu(y)
+      y = self.drop(y)
+      x = x + y
+    return x * x_mask
+
+
+class WN(torch.nn.Module):
+  def __init__(self, hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=0, p_dropout=0):
+    super(WN, self).__init__()
+    assert(kernel_size % 2 == 1)
+    self.hidden_channels =hidden_channels
+    self.kernel_size = kernel_size,
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.gin_channels = gin_channels
+    self.p_dropout = p_dropout
+
+    self.in_layers = torch.nn.ModuleList()
+    self.res_skip_layers = torch.nn.ModuleList()
+    self.drop = nn.Dropout(p_dropout)
+
+    if gin_channels != 0:
+      cond_layer = torch.nn.Conv1d(gin_channels, 2*hidden_channels*n_layers, 1)
+      self.cond_layer = torch.nn.utils.weight_norm(cond_layer, name='weight')
+
+    for i in range(n_layers):
+      dilation = dilation_rate ** i
+      padding = int((kernel_size * dilation - dilation) / 2)
+      in_layer = torch.nn.Conv1d(hidden_channels, 2*hidden_channels, kernel_size,
+                                 dilation=dilation, padding=padding)
+      in_layer = torch.nn.utils.weight_norm(in_layer, name='weight')
+      self.in_layers.append(in_layer)
+
+      # last one is not necessary
+      if i < n_layers - 1:
+        res_skip_channels = 2 * hidden_channels
+      else:
+        res_skip_channels = hidden_channels
+
+      res_skip_layer = torch.nn.Conv1d(hidden_channels, res_skip_channels, 1)
+      res_skip_layer = torch.nn.utils.weight_norm(res_skip_layer, name='weight')
+      self.res_skip_layers.append(res_skip_layer)
+
+  def forward(self, x, x_mask, g=None, **kwargs):
+    output = torch.zeros_like(x)
+    n_channels_tensor = torch.IntTensor([self.hidden_channels])
+
+    if g is not None:
+      g = self.cond_layer(g)
+
+    for i in range(self.n_layers):
+      x_in = self.in_layers[i](x)
+      if g is not None:
+        cond_offset = i * 2 * self.hidden_channels
+        g_l = g[:,cond_offset:cond_offset+2*self.hidden_channels,:]
+      else:
+        g_l = torch.zeros_like(x_in)
+
+      acts = commons.fused_add_tanh_sigmoid_multiply(
+          x_in,
+          g_l,
+          n_channels_tensor)
+      acts = self.drop(acts)
+
+      res_skip_acts = self.res_skip_layers[i](acts)
+      if i < self.n_layers - 1:
+        res_acts = res_skip_acts[:,:self.hidden_channels,:]
+        x = (x + res_acts) * x_mask
+        output = output + res_skip_acts[:,self.hidden_channels:,:]
+      else:
+        output = output + res_skip_acts
+    return output * x_mask
+
+  def remove_weight_norm(self):
+    if self.gin_channels != 0:
+      torch.nn.utils.remove_weight_norm(self.cond_layer)
+    for l in self.in_layers:
+      torch.nn.utils.remove_weight_norm(l)
+    for l in self.res_skip_layers:
+     torch.nn.utils.remove_weight_norm(l)
+
+
+class ResBlock1(torch.nn.Module):
+    def __init__(self, channels, kernel_size=3, dilation=(1, 3, 5)):
+        super(ResBlock1, self).__init__()
+        self.convs1 = nn.ModuleList([
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[0],
+                               padding=get_padding(kernel_size, dilation[0]))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[1],
+                               padding=get_padding(kernel_size, dilation[1]))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[2],
+                               padding=get_padding(kernel_size, dilation[2])))
+        ])
+        self.convs1.apply(init_weights)
+
+        self.convs2 = nn.ModuleList([
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
+                               padding=get_padding(kernel_size, 1))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
+                               padding=get_padding(kernel_size, 1))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
+                               padding=get_padding(kernel_size, 1)))
+        ])
+        self.convs2.apply(init_weights)
+
+    def forward(self, x, x_mask=None):
+        for c1, c2 in zip(self.convs1, self.convs2):
+            xt = F.leaky_relu(x, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c1(xt)
+            xt = F.leaky_relu(xt, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c2(xt)
+            x = xt + x
+        if x_mask is not None:
+            x = x * x_mask
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.convs1:
+            remove_weight_norm(l)
+        for l in self.convs2:
+            remove_weight_norm(l)
+
+
+class ResBlock2(torch.nn.Module):
+    def __init__(self, channels, kernel_size=3, dilation=(1, 3)):
+        super(ResBlock2, self).__init__()
+        self.convs = nn.ModuleList([
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[0],
+                               padding=get_padding(kernel_size, dilation[0]))),
+            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[1],
+                               padding=get_padding(kernel_size, dilation[1])))
+        ])
+        self.convs.apply(init_weights)
+
+    def forward(self, x, x_mask=None):
+        for c in self.convs:
+            xt = F.leaky_relu(x, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c(xt)
+            x = xt + x
+        if x_mask is not None:
+            x = x * x_mask
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.convs:
+            remove_weight_norm(l)
+
+
+class Log(nn.Module):
+  def forward(self, x, x_mask, reverse=False, **kwargs):
+    if not reverse:
+      y = torch.log(torch.clamp_min(x, 1e-5)) * x_mask
+      logdet = torch.sum(-y, [1, 2])
+      return y, logdet
+    else:
+      x = torch.exp(x) * x_mask
+      return x
+    
+
+class Flip(nn.Module):
+  def forward(self, x, *args, reverse=False, **kwargs):
+    x = torch.flip(x, [1])
+    if not reverse:
+      logdet = torch.zeros(x.size(0)).to(dtype=x.dtype, device=x.device)
+      return x, logdet
+    else:
+      return x
+
+
+class ElementwiseAffine(nn.Module):
+  def __init__(self, channels):
+    super().__init__()
+    self.channels = channels
+    self.m = nn.Parameter(torch.zeros(channels,1))
+    self.logs = nn.Parameter(torch.zeros(channels,1))
+
+  def forward(self, x, x_mask, reverse=False, **kwargs):
+    if not reverse:
+      y = self.m + torch.exp(self.logs) * x
+      y = y * x_mask
+      logdet = torch.sum(self.logs * x_mask, [1,2])
+      return y, logdet
+    else:
+      x = (x - self.m) * torch.exp(-self.logs) * x_mask
+      return x
+
+
+class ResidualCouplingLayer(nn.Module):
+  def __init__(self,
+      channels,
+      hidden_channels,
+      kernel_size,
+      dilation_rate,
+      n_layers,
+      p_dropout=0,
+      gin_channels=0,
+      mean_only=False):
+    assert channels % 2 == 0, "channels should be divisible by 2"
+    super().__init__()
+    self.channels = channels
+    self.hidden_channels = hidden_channels
+    self.kernel_size = kernel_size
+    self.dilation_rate = dilation_rate
+    self.n_layers = n_layers
+    self.half_channels = channels // 2
+    self.mean_only = mean_only
+
+    self.pre = nn.Conv1d(self.half_channels, hidden_channels, 1)
+    self.enc = WN(hidden_channels, kernel_size, dilation_rate, n_layers, p_dropout=p_dropout, gin_channels=gin_channels)
+    self.post = nn.Conv1d(hidden_channels, self.half_channels * (2 - mean_only), 1)
+    self.post.weight.data.zero_()
+    self.post.bias.data.zero_()
+
+  def forward(self, x, x_mask, g=None, reverse=False):
+    x0, x1 = torch.split(x, [self.half_channels]*2, 1)
+    h = self.pre(x0) * x_mask
+    h = self.enc(h, x_mask, g=g)
+    stats = self.post(h) * x_mask
+    if not self.mean_only:
+      m, logs = torch.split(stats, [self.half_channels]*2, 1)
+    else:
+      m = stats
+      logs = torch.zeros_like(m)
+
+    if not reverse:
+      x1 = m + x1 * torch.exp(logs) * x_mask
+      x = torch.cat([x0, x1], 1)
+      logdet = torch.sum(logs, [1,2])
+      return x, logdet
+    else:
+      x1 = (x1 - m) * torch.exp(-logs) * x_mask
+      x = torch.cat([x0, x1], 1)
+      return x
+
+
+class ConvFlow(nn.Module):
+  def __init__(self, in_channels, filter_channels, kernel_size, n_layers, num_bins=10, tail_bound=5.0):
+    super().__init__()
+    self.in_channels = in_channels
+    self.filter_channels = filter_channels
+    self.kernel_size = kernel_size
+    self.n_layers = n_layers
+    self.num_bins = num_bins
+    self.tail_bound = tail_bound
+    self.half_channels = in_channels // 2
+
+    self.pre = nn.Conv1d(self.half_channels, filter_channels, 1)
+    self.convs = DDSConv(filter_channels, kernel_size, n_layers, p_dropout=0.)
+    self.proj = nn.Conv1d(filter_channels, self.half_channels * (num_bins * 3 - 1), 1)
+    self.proj.weight.data.zero_()
+    self.proj.bias.data.zero_()
+
+  def forward(self, x, x_mask, g=None, reverse=False):
+    x0, x1 = torch.split(x, [self.half_channels]*2, 1)
+    h = self.pre(x0)
+    h = self.convs(h, x_mask, g=g)
+    h = self.proj(h) * x_mask
+
+    b, c, t = x0.shape
+    h = h.reshape(b, c, -1, t).permute(0, 1, 3, 2) # [b, cx?, t] -> [b, c, t, ?]
+
+    unnormalized_widths = h[..., :self.num_bins] / math.sqrt(self.filter_channels)
+    unnormalized_heights = h[..., self.num_bins:2*self.num_bins] / math.sqrt(self.filter_channels)
+    unnormalized_derivatives = h[..., 2 * self.num_bins:]
+
+    x1, logabsdet = piecewise_rational_quadratic_transform(x1,
+        unnormalized_widths,
+        unnormalized_heights,
+        unnormalized_derivatives,
+        inverse=reverse,
+        tails='linear',
+        tail_bound=self.tail_bound
+    )
+
+    x = torch.cat([x0, x1], 1) * x_mask
+    logdet = torch.sum(logabsdet * x_mask, [1,2])
+    if not reverse:
+        return x, logdet
+    else:
+        return x

requirements.txt

@@ -0,0 +1,17 @@
+Cython
+librosa
+matplotlib
+numpy
+phonemizer
+scipy
+tensorboard
+torch
+torchvision
+Unidecode
+pyopenjtalk
+ffmpeg
+jamo
+cn2an
+gradio
+pypinyin
+jieba

transforms.py

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

utils.py

@@ -0,0 +1,225 @@
+import os
+import sys
+import argparse
+import logging
+import json
+import subprocess
+import numpy as np
+import librosa
+import torch
+
+MATPLOTLIB_FLAG = False
+
+logging.basicConfig(stream=sys.stdout, level=logging.DEBUG)
+logger = logging
+
+
+def load_checkpoint(checkpoint_path, model, optimizer=None):
+  assert os.path.isfile(checkpoint_path)
+  checkpoint_dict = torch.load(checkpoint_path, map_location='cpu')
+  iteration = checkpoint_dict['iteration']
+  learning_rate = checkpoint_dict['learning_rate']
+  if optimizer is not None:
+    optimizer.load_state_dict(checkpoint_dict['optimizer'])
+  saved_state_dict = checkpoint_dict['model']
+  if hasattr(model, 'module'):
+    state_dict = model.module.state_dict()
+  else:
+    state_dict = model.state_dict()
+  new_state_dict= {}
+  for k, v in state_dict.items():
+    try:
+      new_state_dict[k] = saved_state_dict[k]
+    except:
+      logger.info("%s is not in the checkpoint" % k)
+      new_state_dict[k] = v
+  if hasattr(model, 'module'):
+    model.module.load_state_dict(new_state_dict)
+  else:
+    model.load_state_dict(new_state_dict)
+  logger.info("Loaded checkpoint '{}' (iteration {})" .format(
+    checkpoint_path, iteration))
+  return model, optimizer, learning_rate, iteration
+
+
+def plot_spectrogram_to_numpy(spectrogram):
+  global MATPLOTLIB_FLAG
+  if not MATPLOTLIB_FLAG:
+    import matplotlib
+    matplotlib.use("Agg")
+    MATPLOTLIB_FLAG = True
+    mpl_logger = logging.getLogger('matplotlib')
+    mpl_logger.setLevel(logging.WARNING)
+  import matplotlib.pylab as plt
+  import numpy as np
+
+  fig, ax = plt.subplots(figsize=(10,2))
+  im = ax.imshow(spectrogram, aspect="auto", origin="lower",
+                  interpolation='none')
+  plt.colorbar(im, ax=ax)
+  plt.xlabel("Frames")
+  plt.ylabel("Channels")
+  plt.tight_layout()
+
+  fig.canvas.draw()
+  data = np.fromstring(fig.canvas.tostring_rgb(), dtype=np.uint8, sep='')
+  data = data.reshape(fig.canvas.get_width_height()[::-1] + (3,))
+  plt.close()
+  return data
+
+
+def plot_alignment_to_numpy(alignment, info=None):
+  global MATPLOTLIB_FLAG
+  if not MATPLOTLIB_FLAG:
+    import matplotlib
+    matplotlib.use("Agg")
+    MATPLOTLIB_FLAG = True
+    mpl_logger = logging.getLogger('matplotlib')
+    mpl_logger.setLevel(logging.WARNING)
+  import matplotlib.pylab as plt
+  import numpy as np
+
+  fig, ax = plt.subplots(figsize=(6, 4))
+  im = ax.imshow(alignment.transpose(), aspect='auto', origin='lower',
+                  interpolation='none')
+  fig.colorbar(im, ax=ax)
+  xlabel = 'Decoder timestep'
+  if info is not None:
+      xlabel += '\n\n' + info
+  plt.xlabel(xlabel)
+  plt.ylabel('Encoder timestep')
+  plt.tight_layout()
+
+  fig.canvas.draw()
+  data = np.fromstring(fig.canvas.tostring_rgb(), dtype=np.uint8, sep='')
+  data = data.reshape(fig.canvas.get_width_height()[::-1] + (3,))
+  plt.close()
+  return data
+
+
+def load_audio_to_torch(full_path, target_sampling_rate):
+  audio, sampling_rate = librosa.load(full_path, sr=target_sampling_rate, mono=True)
+  return torch.FloatTensor(audio.astype(np.float32))
+
+
+def load_filepaths_and_text(filename, split="|"):
+  with open(filename, encoding='utf-8') as f:
+    filepaths_and_text = [line.strip().split(split) for line in f]
+  return filepaths_and_text
+
+
+def get_hparams(init=True):
+  parser = argparse.ArgumentParser()
+  parser.add_argument('-c', '--config', type=str, default="./configs/base.json",
+                      help='JSON file for configuration')
+  parser.add_argument('-m', '--model', type=str, required=True,
+                      help='Model name')
+
+  args = parser.parse_args()
+  model_dir = os.path.join("./logs", args.model)
+
+  if not os.path.exists(model_dir):
+    os.makedirs(model_dir)
+
+  config_path = args.config
+  config_save_path = os.path.join(model_dir, "config.json")
+  if init:
+    with open(config_path, "r") as f:
+      data = f.read()
+    with open(config_save_path, "w") as f:
+      f.write(data)
+  else:
+    with open(config_save_path, "r") as f:
+      data = f.read()
+  config = json.loads(data)
+
+  hparams = HParams(**config)
+  hparams.model_dir = model_dir
+  return hparams
+
+
+def get_hparams_from_dir(model_dir):
+  config_save_path = os.path.join(model_dir, "config.json")
+  with open(config_save_path, "r") as f:
+    data = f.read()
+  config = json.loads(data)
+
+  hparams =HParams(**config)
+  hparams.model_dir = model_dir
+  return hparams
+
+
+def get_hparams_from_file(config_path):
+  with open(config_path, "r") as f:
+    data = f.read()
+  config = json.loads(data)
+
+  hparams =HParams(**config)
+  return hparams
+
+
+def check_git_hash(model_dir):
+  source_dir = os.path.dirname(os.path.realpath(__file__))
+  if not os.path.exists(os.path.join(source_dir, ".git")):
+    logger.warn("{} is not a git repository, therefore hash value comparison will be ignored.".format(
+      source_dir
+    ))
+    return
+
+  cur_hash = subprocess.getoutput("git rev-parse HEAD")
+
+  path = os.path.join(model_dir, "githash")
+  if os.path.exists(path):
+    saved_hash = open(path).read()
+    if saved_hash != cur_hash:
+      logger.warn("git hash values are different. {}(saved) != {}(current)".format(
+        saved_hash[:8], cur_hash[:8]))
+  else:
+    open(path, "w").write(cur_hash)
+
+
+def get_logger(model_dir, filename="train.log"):
+  global logger
+  logger = logging.getLogger(os.path.basename(model_dir))
+  logger.setLevel(logging.DEBUG)
+
+  formatter = logging.Formatter("%(asctime)s\t%(name)s\t%(levelname)s\t%(message)s")
+  if not os.path.exists(model_dir):
+    os.makedirs(model_dir)
+  h = logging.FileHandler(os.path.join(model_dir, filename))
+  h.setLevel(logging.DEBUG)
+  h.setFormatter(formatter)
+  logger.addHandler(h)
+  return logger
+
+
+class HParams():
+  def __init__(self, **kwargs):
+    for k, v in kwargs.items():
+      if type(v) == dict:
+        v = HParams(**v)
+      self[k] = v
+
+  def keys(self):
+    return self.__dict__.keys()
+
+  def items(self):
+    return self.__dict__.items()
+
+  def values(self):
+    return self.__dict__.values()
+
+  def __len__(self):
+    return len(self.__dict__)
+
+  def __getitem__(self, key):
+    return getattr(self, key)
+
+  def __setitem__(self, key, value):
+    return setattr(self, key, value)
+
+  def __contains__(self, key):
+    return key in self.__dict__
+
+  def __repr__(self):
+    return self.__dict__.__repr__()