diff --git a/.gitattributes b/.gitattributes
new file mode 100644
index 0000000000000000000000000000000000000000..207cbccd8f3a9c851465d6ddb3cbe501b3012b40
--- /dev/null
+++ b/.gitattributes
@@ -0,0 +1,48 @@
+*.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
+*.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
+models/HolasoyGerman/HolasoyGerman/added_IVF3117_Flat_nprobe_1_HolasoyGerman_v2.index filter=lfs diff=lfs merge=lfs -text
+models/Homer/Homer/added_IVF360_Flat_nprobe_1_HomerEsp_v1.index filter=lfs diff=lfs merge=lfs -text
+models/Ibai/Ibai/added_IVF4601_Flat_nprobe_1_Ibai_v2.index filter=lfs diff=lfs merge=lfs -text
+models/IlloJuan/IlloJuan/added_IVF593_Flat_nprobe_1_IlloJuan_v2.index filter=lfs diff=lfs merge=lfs -text
+models/Joseju/Joseju/added_IVF256_Flat_nprobe_1_Joseju_v2.index filter=lfs diff=lfs merge=lfs -text
+models/Quevedo/Quevedo/added_IVF2301_Flat_nprobe_10.index filter=lfs diff=lfs merge=lfs -text
+models/Shadoune666/Shadoune666/added_IVF716_Flat_nprobe_1_Shadoune666_v2.index filter=lfs diff=lfs merge=lfs -text
+models/Spreen/Spreen/added_IVF4737_Flat_nprobe_1_Spreen_v2.index filter=lfs diff=lfs merge=lfs -text
+models/Totakeke/Totakeke/added_IVF256_Flat_nprobe_1_full_totakeke_v2.index filter=lfs diff=lfs merge=lfs -text
+models/Vegetta777/Vegetta777/added_IVF2021_Flat_nprobe_1_Vegetta777_v2.index filter=lfs diff=lfs merge=lfs -text
+models/Villager/Villager/added_IVF81_Flat_nprobe_1_v2.index filter=lfs diff=lfs merge=lfs -text
+models/WalterWhite/WalterWhite/added_IVF458_Flat_nprobe_1_ww2_v2.index filter=lfs diff=lfs merge=lfs -text
+models/FernandoAlonso/FernandoAlonso/added_IVF582_Flat_nprobe_1_fernando2_v2.index filter=lfs diff=lfs merge=lfs -text
diff --git a/README.md b/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..11f4ae415ea820a8c82c69da45fe909c104c5611
--- /dev/null
+++ b/README.md
@@ -0,0 +1,14 @@
+---
+title: RVC TTS Demo
+emoji: 🚀
+colorFrom: red
+colorTo: pink
+sdk: gradio
+sdk_version: 3.36.1
+app_file: app.py
+pinned: false
+license: gpl-3.0
+duplicated_from: ImPavloh/RVC-TTS-Demo
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
\ No newline at end of file
diff --git a/app.py b/app.py
new file mode 100644
index 0000000000000000000000000000000000000000..4c8bce2c69cfd2e7b7f55a3053738fe604c6a86a
--- /dev/null
+++ b/app.py
@@ -0,0 +1,200 @@
+import os
+import json
+import torch
+import asyncio
+import librosa
+import hashlib
+import edge_tts
+import gradio as gr
+from config import Config
+from vc_infer_pipeline import VC
+from fairseq import checkpoint_utils
+from lib.infer_pack.models import (SynthesizerTrnMs256NSFsid, SynthesizerTrnMs256NSFsid_nono, SynthesizerTrnMs768NSFsid, SynthesizerTrnMs768NSFsid_nono,)
+
+config = Config()
+
+def load_json_file(filepath):
+ with open(filepath, "r", encoding="utf-8") as f: content = json.load(f)
+ return content
+
+def file_checksum(file_path):
+ with open(file_path, 'rb') as f:
+ file_data = f.read()
+ return hashlib.md5(file_data).hexdigest()
+
+def get_existing_model_info(category_directory):
+ model_info_path = os.path.join(category_directory, 'model_info.json')
+ if os.path.exists(model_info_path):
+ with open(model_info_path, 'r') as f: return json.load(f)
+ return None
+
+def generate_model_info_files():
+ folder_info = {}
+ model_directory = "models/"
+ for category_name in os.listdir(model_directory):
+ category_directory = os.path.join(model_directory, category_name)
+ if not os.path.isdir(category_directory): continue
+
+ folder_info[category_name] = {"title": category_name, "folder_path": category_name}
+ existing_model_info = get_existing_model_info(category_directory)
+ model_info = {}
+ regenerate_model_info = False
+
+ for model_name in os.listdir(category_directory):
+ model_path = os.path.join(category_directory, model_name)
+ if not os.path.isdir(model_path): continue
+
+ model_data, regenerate = gather_model_info(category_directory, model_name, model_path, existing_model_info)
+ if model_data is not None:
+ model_info[model_name] = model_data
+ regenerate_model_info |= regenerate
+
+ if regenerate_model_info:
+ with open(os.path.join(category_directory, 'model_info.json'), 'w') as f: json.dump(model_info, f, indent=4)
+
+ folder_info_path = os.path.join(model_directory, 'folder_info.json')
+ with open(folder_info_path, 'w') as f: json.dump(folder_info, f, indent=4)
+
+def should_regenerate_model_info(existing_model_info, model_name, pth_checksum, index_checksum):
+ if existing_model_info is None or model_name not in existing_model_info: return True
+ return (existing_model_info[model_name]['model_path_checksum'] != pth_checksum or existing_model_info[model_name]['index_path_checksum'] != index_checksum)
+
+def get_model_files(model_path): return [f for f in os.listdir(model_path) if f.endswith('.pth') or f.endswith('.index')]
+
+def gather_model_info(category_directory, model_name, model_path, existing_model_info):
+ model_files = get_model_files(model_path)
+ if len(model_files) != 2: return None, False
+
+ pth_file = [f for f in model_files if f.endswith('.pth')][0]
+ index_file = [f for f in model_files if f.endswith('.index')][0]
+ pth_checksum = file_checksum(os.path.join(model_path, pth_file))
+ index_checksum = file_checksum(os.path.join(model_path, index_file))
+ regenerate = should_regenerate_model_info(existing_model_info, model_name, pth_checksum, index_checksum)
+
+ return {"title": model_name, "model_path": pth_file, "feature_retrieval_library": index_file, "model_path_checksum": pth_checksum, "index_path_checksum": index_checksum}, regenerate
+
+def create_vc_fn(model_name, tgt_sr, net_g, vc, if_f0, version, file_index):
+ def vc_fn(tts_text, tts_voice):
+ try:
+ if len(tts_text) > 100: return None
+ if tts_text is None or tts_voice is None: return None
+ asyncio.run(edge_tts.Communicate(tts_text, "-".join(tts_voice.split('-')[:-1])).save("tts.mp3"))
+ audio, sr = librosa.load("tts.mp3", sr=16000, mono=True)
+ vc_input = "tts.mp3"
+ times = [0, 0, 0]
+ audio_opt = vc.pipeline(hubert_model, net_g, 0, audio, vc_input, times, 0, "pm", file_index, 0.7, if_f0, 3, tgt_sr, 0, 1, version, 0.5, f0_file=None)
+ return (tgt_sr, audio_opt)
+ except Exception: return None
+ return vc_fn
+
+def load_model_parameters(category_folder, character_name, info):
+ model_index = f"models/{category_folder}/{character_name}/{info['feature_retrieval_library']}"
+ cpt = torch.load(f"models/{category_folder}/{character_name}/{info['model_path']}", map_location="cpu")
+ return model_index, cpt
+
+def select_net_g(cpt, version, if_f0):
+ if version == "v1":
+ if if_f0 == 1: net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=config.is_half)
+ else: net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"])
+ elif version == "v2":
+ if if_f0 == 1: net_g = SynthesizerTrnMs768NSFsid(*cpt["config"], is_half=config.is_half)
+ else: net_g = SynthesizerTrnMs768NSFsid_nono(*cpt["config"])
+ return net_g
+
+def load_model_and_prepare(cpt, net_g):
+ del net_g.enc_q
+ net_g.load_state_dict(cpt["weight"], strict=False)
+ net_g.eval().to(config.device)
+ net_g = net_g.half() if config.is_half else net_g.float()
+ return net_g
+
+def create_and_append_model(models, model_functions, character_name, model_title, version, vc_fn):
+ models.append((character_name, model_title, version, vc_fn))
+ model_functions[character_name] = vc_fn
+ return models, model_functions
+
+def load_model():
+ categories = []
+ model_functions = {}
+ folder_info = load_json_file("models/folder_info.json")
+ for category_name, category_info in folder_info.items():
+ models = []
+ models_info = load_json_file(f"models/{category_info['folder_path']}/model_info.json")
+ for character_name, info in models_info.items():
+ model_index, cpt = load_model_parameters(category_info['folder_path'], character_name, info)
+ net_g = select_net_g(cpt, cpt.get("version", "v1"), cpt.get("f0", 1))
+ cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0]
+ net_g = load_model_and_prepare(cpt, net_g)
+ vc = VC(cpt["config"][-1], config)
+ vc_fn = create_vc_fn(info['model_path'], cpt["config"][-1], net_g, vc, cpt.get("f0", 1), cpt.get("version", "v1"), model_index)
+ models, model_functions = create_and_append_model(models, model_functions, character_name, info['title'], cpt.get("version", "v1"), vc_fn)
+ categories.append([category_info['title'], category_info['folder_path'], models])
+ return categories, model_functions
+
+generate_model_info_files()
+
+css = """
+.gradio-container { font-family: 'IBM Plex Sans', sans-serif; }
+footer { visibility: hidden; display: none; }
+.center-container { display: flex; flex-direction: column; align-items: center; justify-content: center;}
+"""
+
+if __name__ == '__main__':
+ global hubert_model
+ models, _, _ = checkpoint_utils.load_model_ensemble_and_task(["hubert_base.pt"], suffix="")
+ hubert_model = models[0]
+ hubert_model = hubert_model.to(config.device)
+ hubert_model = hubert_model.half() if config.is_half else hubert_model.float()
+ hubert_model.eval()
+ categories, model_functions = load_model()
+ tts_voice_list = asyncio.get_event_loop().run_until_complete(edge_tts.list_voices())
+ voices = [f"{v['ShortName']}-{v['Gender']}" for v in tts_voice_list]
+ with gr.Blocks(css=css, title="Demo RVC TTS - Pavloh", theme=gr.themes.Soft(primary_hue="cyan", secondary_hue="blue", radius_size="lg", text_size="lg")
+ .set(loader_color="#0B0F19", shadow_drop='*shadow_drop_lg', block_border_width="3px")) as pavloh:
+ gr.HTML("""
+
+ """)
+
+ with gr.Row():
+ with gr.Column():
+ m1 = gr.Dropdown(label="📦 Voice Model", choices=list(model_functions.keys()), allow_custom_value=False, value="Ibai")
+ t2 = gr.Dropdown(label="⚙️ Voice style and language [Edge-TTS]", choices=voices, allow_custom_value=False, value="es-ES-AlvaroNeural-Male")
+ t1 = gr.Textbox(label="📝 Text to convert")
+ c1 = gr.Button("Convert", variant="primary")
+ a1 = gr.Audio(label="🔉 Converted Text", interactive=False)
+
+ def call_selected_model_fn(selected_model, t1, t2):
+ vc_fn = model_functions[selected_model]
+ return vc_fn(t1, t2)
+
+ c1.click(fn=call_selected_model_fn, inputs=[m1, t1, t2], outputs=[a1])
+
+ gr.HTML("""
+
+ By using this website, you agree to the license.
+
+ """)
+
+pavloh.queue(concurrency_count=1).launch()
\ No newline at end of file
diff --git a/config.py b/config.py
new file mode 100644
index 0000000000000000000000000000000000000000..ddd0c75c4b71f44d12ce02e99646bd8c34eb8a84
--- /dev/null
+++ b/config.py
@@ -0,0 +1,31 @@
+import argparse
+from multiprocessing import cpu_count
+
+class Config:
+ def __init__(self):
+ self.device = "cpu"
+ self.is_half = False
+ self.n_cpu = cpu_count()
+ (self.python_cmd, self.colab, self.noparallel, self.noautoopen, self.api) = self.arg_parse()
+ self.listen_port = 7860
+ self.x_pad, self.x_query, self.x_center, self.x_max = self.device_config()
+
+ @staticmethod
+ def arg_parse() -> tuple:
+ parser = argparse.ArgumentParser()
+ parser.add_argument("--pycmd", type=str, default="python")
+ parser.add_argument("--colab", action="store_true")
+ parser.add_argument("--noparallel", action="store_true")
+ parser.add_argument("--noautoopen", action="store_true")
+ parser.add_argument("--api", action="store_true")
+ cmd_opts = parser.parse_args()
+
+ return (cmd_opts.pycmd, cmd_opts.colab, cmd_opts.noparallel, cmd_opts.noautoopen, cmd_opts.api)
+
+ def device_config(self) -> tuple:
+ x_pad = 1
+ x_query = 6
+ x_center = 38
+ x_max = 41
+
+ return x_pad, x_query, x_center, x_max
\ No newline at end of file
diff --git a/hubert_base.pt b/hubert_base.pt
new file mode 100644
index 0000000000000000000000000000000000000000..72f47ab58564f01d5cc8b05c63bdf96d944551ff
--- /dev/null
+++ b/hubert_base.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f54b40fd2802423a5643779c4861af1e9ee9c1564dc9d32f54f20b5ffba7db96
+size 189507909
diff --git a/lib/infer_pack/__pycache__/attentions.cpython-310.pyc b/lib/infer_pack/__pycache__/attentions.cpython-310.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..8449e194fefec43e1b0afe0af05ca18572584aac
Binary files /dev/null and b/lib/infer_pack/__pycache__/attentions.cpython-310.pyc differ
diff --git a/lib/infer_pack/__pycache__/commons.cpython-310.pyc b/lib/infer_pack/__pycache__/commons.cpython-310.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..20fce537a6e0448755f6a06aeb1995241eb7d15f
Binary files /dev/null and b/lib/infer_pack/__pycache__/commons.cpython-310.pyc differ
diff --git a/lib/infer_pack/__pycache__/models.cpython-310.pyc b/lib/infer_pack/__pycache__/models.cpython-310.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..9d46ea578021180ad17cf0200dd2251048fff0ca
Binary files /dev/null and b/lib/infer_pack/__pycache__/models.cpython-310.pyc differ
diff --git a/lib/infer_pack/__pycache__/modules.cpython-310.pyc b/lib/infer_pack/__pycache__/modules.cpython-310.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..51565c81478f93cf86f6cfba1c421f48013b6e5b
Binary files /dev/null and b/lib/infer_pack/__pycache__/modules.cpython-310.pyc differ
diff --git a/lib/infer_pack/__pycache__/transforms.cpython-310.pyc b/lib/infer_pack/__pycache__/transforms.cpython-310.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..846fbb824c3dc8d3b24997fc34ba6497dea6a335
Binary files /dev/null and b/lib/infer_pack/__pycache__/transforms.cpython-310.pyc differ
diff --git a/lib/infer_pack/attentions.py b/lib/infer_pack/attentions.py
new file mode 100644
index 0000000000000000000000000000000000000000..da4937f7d6ba428fe3d74d3e228056abe7cb998d
--- /dev/null
+++ b/lib/infer_pack/attentions.py
@@ -0,0 +1,253 @@
+import copy
+import math
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import functional as F
+from lib.infer_pack import commons
+from lib.infer_pack import modules
+from lib.infer_pack.modules import LayerNorm
+
+class Encoder(nn.Module):
+ def __init__(self, hidden_channels, filter_channels, n_heads, n_layers, kernel_size=1, p_dropout=0.0, window_size=10, **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 _ 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=0.0, 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 _ 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):
+ self_attn_mask = commons.subsequent_mask(x_mask.size(2)).to(device=x.device, dtype=x.dtype)
+ encdec_attn_mask = h_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+ x = x * x_mask
+ for i in range(self.n_layers):
+ y = self.self_attn_layers[i](x, x, self_attn_mask)
+ y = self.drop(y)
+ x = self.norm_layers_0[i](x + y)
+
+ y = self.encdec_attn_layers[i](x, h, encdec_attn_mask)
+ y = self.drop(y)
+ x = self.norm_layers_1[i](x + y)
+
+ y = self.ffn_layers[i](x, x_mask)
+ y = self.drop(y)
+ x = self.norm_layers_2[i](x + y)
+ x = x * x_mask
+ return x
+
+
+class MultiHeadAttention(nn.Module):
+ def __init__(self, channels, out_channels, n_heads, p_dropout=0.0, window_size=None, heads_share=True, block_length=None, proximal_bias=False, proximal_init=False):
+ super().__init__()
+ assert channels % n_heads == 0
+
+ self.channels = channels
+ self.out_channels = out_channels
+ self.n_heads = n_heads
+ self.p_dropout = p_dropout
+ self.window_size = window_size
+ self.heads_share = heads_share
+ self.block_length = block_length
+ self.proximal_bias = proximal_bias
+ self.proximal_init = proximal_init
+ self.attn = None
+
+ self.k_channels = channels // n_heads
+ self.conv_q = nn.Conv1d(channels, channels, 1)
+ self.conv_k = nn.Conv1d(channels, channels, 1)
+ self.conv_v = nn.Conv1d(channels, channels, 1)
+ self.conv_o = nn.Conv1d(channels, out_channels, 1)
+ self.drop = nn.Dropout(p_dropout)
+
+ if window_size is not None:
+ n_heads_rel = 1 if heads_share else n_heads
+ rel_stddev = self.k_channels**-0.5
+ self.emb_rel_k = nn.Parameter(
+ torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels)
+ * rel_stddev
+ )
+ self.emb_rel_v = nn.Parameter(
+ torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels)
+ * rel_stddev
+ )
+
+ nn.init.xavier_uniform_(self.conv_q.weight)
+ nn.init.xavier_uniform_(self.conv_k.weight)
+ nn.init.xavier_uniform_(self.conv_v.weight)
+ if proximal_init:
+ with torch.no_grad():
+ self.conv_k.weight.copy_(self.conv_q.weight)
+ self.conv_k.bias.copy_(self.conv_q.bias)
+
+ def forward(self, x, c, attn_mask=None):
+ q = self.conv_q(x)
+ k = self.conv_k(c)
+ v = self.conv_v(c)
+
+ x, self.attn = self.attention(q, k, v, mask=attn_mask)
+
+ x = self.conv_o(x)
+ return x
+
+ def attention(self, query, key, value, mask=None):
+ 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)
+ 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))
+ return output, p_attn
+
+ def _matmul_with_relative_values(self, x, y): return torch.matmul(x, y.unsqueeze(0))
+ def _matmul_with_relative_keys(self, x, y): return torch.matmul(x, y.unsqueeze(0).transpose(-2, -1))
+
+ def _get_relative_embeddings(self, relative_embeddings, length):
+ max_relative_position = 2 * self.window_size + 1
+ 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
+ return padded_relative_embeddings[:, slice_start_position:slice_end_position]
+
+ def _relative_position_to_absolute_position(self, x):
+ batch, heads, length, _ = x.size()
+ x = F.pad(x, commons.convert_pad_shape([[0, 0], [0, 0], [0, 0], [0, 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]]))
+
+ return x_flat.view([batch, heads, length + 1, 2 * length - 1])[:, :, :length, length - 1 :]
+
+ def _absolute_position_to_relative_position(self, x):
+ batch, heads, length, _ = x.size()
+ 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)])
+ x_flat = F.pad(x_flat, commons.convert_pad_shape([[0, 0], [0, 0], [length, 0]]))
+ return x_flat.view([batch, heads, length, 2 * length])[:, :, :, 1:]
+
+ def _attention_bias_proximal(self, length):
+ r = torch.arange(length, dtype=torch.float32)
+ diff = torch.unsqueeze(r, 0) - torch.unsqueeze(r, 1)
+ return torch.unsqueeze(torch.unsqueeze(-torch.log1p(torch.abs(diff)), 0), 0)
+
+class FFN(nn.Module):
+ def __init__(self, in_channels, out_channels, filter_channels, kernel_size, p_dropout=0.0, activation=None, causal=False):
+ super().__init__()
+ self.in_channels = in_channels
+ self.out_channels = out_channels
+ self.filter_channels = filter_channels
+ self.kernel_size = kernel_size
+ self.p_dropout = p_dropout
+ self.activation = activation
+ self.causal = causal
+ self.padding = self._causal_padding if causal else 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
+ return self._extracted_from__same_padding_5(pad_l, pad_r, 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
+ return self._extracted_from__same_padding_5(pad_l, pad_r, x)
+
+ def _extracted_from__same_padding_5(self, pad_l, pad_r, x):
+ padding = [[0, 0], [0, 0], [pad_l, pad_r]]
+ x = F.pad(x, commons.convert_pad_shape(padding))
+ return x
\ No newline at end of file
diff --git a/lib/infer_pack/commons.py b/lib/infer_pack/commons.py
new file mode 100644
index 0000000000000000000000000000000000000000..29d5c1c54e35f96a8548bac4b503af7ce39e0d5a
--- /dev/null
+++ b/lib/infer_pack/commons.py
@@ -0,0 +1,112 @@
+import math
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+def init_weights(m, mean=0.0, std=0.01):
+ classname = m.__class__.__name__
+ if "Conv" in classname: m.weight.data.normal_(mean, std)
+
+def get_padding(kernel_size, dilation=1):
+ return (kernel_size * dilation - dilation) // 2
+
+def convert_pad_shape(pad_shape):
+ l = pad_shape[::-1]
+ return [item for sublist in l for item in sublist]
+
+def kl_divergence(m_p, logs_p, m_q, logs_q):
+ kl = logs_q - logs_p - 0.5
+ kl += 0.5 * (torch.exp(2.0 * logs_p) + (m_p - m_q) ** 2) * torch.exp(-2.0 * logs_q)
+ return kl
+
+def rand_gumbel(shape):
+ return -torch.log(-torch.log(torch.rand(shape) + 1e-5))
+
+def rand_gumbel_like(x):
+ return rand_gumbel(x.size()).to(dtype=x.dtype, device=x.device)
+
+def slice_segments(x, ids_str, segment_size=4, slice_dim=2):
+ if slice_dim == 1: ret = torch.zeros_like(x[:, :segment_size])
+ else: 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 apply_timing_signal_1d(x, operation='add', min_timescale=1.0, max_timescale=1.0e4):
+ b, channels, length = x.size()
+ signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+ signal = signal.to(dtype=x.dtype, device=x.device)
+
+ if operation == 'add': return x + signal
+ elif operation == 'cat': return torch.cat([x, signal], axis=1)
+
+def subsequent_mask(length):
+ return torch.tril(torch.ones(length, length)).unsqueeze(0).unsqueeze(0)
+
+@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:, :])
+ return t_act * s_act
+
+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):
+ 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 ** (1.0 / norm_type)
+ return total_norm
\ No newline at end of file
diff --git a/lib/infer_pack/models.py b/lib/infer_pack/models.py
new file mode 100644
index 0000000000000000000000000000000000000000..fbcac8deb5fe6fe2c77752ca5b10459ce71ea43b
--- /dev/null
+++ b/lib/infer_pack/models.py
@@ -0,0 +1,711 @@
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+from lib.infer_pack import modules
+from lib.infer_pack import attentions
+from lib.infer_pack import commons
+from lib.infer_pack.commons import init_weights, get_padding
+from torch.nn import Conv1d, ConvTranspose1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm, spectral_norm
+from lib.infer_pack.commons import init_weights
+import numpy as np
+from lib.infer_pack import commons
+
+class TextEncoder256(nn.Module):
+ def __init__(self, out_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout, f0=True):
+ super().__init__()
+ 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_phone = nn.Linear(256, hidden_channels)
+ self.lrelu = nn.LeakyReLU(0.1, inplace=True)
+ if f0 == True: self.emb_pitch = nn.Embedding(256, hidden_channels)
+ 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, phone, pitch, lengths):
+ if pitch is None: x = self.emb_phone(phone)
+ else: x = self.emb_phone(phone) + self.emb_pitch(pitch)
+ x = x * math.sqrt(self.hidden_channels)
+ x = self.lrelu(x)
+ x = torch.transpose(x, 1, -1)
+ x_mask = torch.unsqueeze(commons.sequence_mask(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 m, logs, x_mask
+
+class TextEncoder768(nn.Module):
+ def __init__(self, out_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout, f0=True):
+ super().__init__()
+ 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_phone = nn.Linear(768, hidden_channels)
+ self.lrelu = nn.LeakyReLU(0.1, inplace=True)
+ if f0 == True: self.emb_pitch = nn.Embedding(256, hidden_channels)
+ 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, phone, pitch, lengths):
+ if pitch is None: x = self.emb_phone(phone)
+ else: x = self.emb_phone(phone) + self.emb_pitch(pitch)
+ x = x * math.sqrt(self.hidden_channels)
+ x = self.lrelu(x)
+ x = torch.transpose(x, 1, -1)
+ x_mask = torch.unsqueeze(commons.sequence_mask(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 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 _ 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
+
+ def remove_weight_norm(self):
+ for i in range(self.n_flows): self.flows[i * 2].remove_weight_norm()
+
+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
+
+ def remove_weight_norm(self):
+ self.enc.remove_weight_norm()
+
+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 k, d in 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):
+ for l in self.ups: remove_weight_norm(l)
+ for l in self.resblocks: l.remove_weight_norm()
+
+class SineGen(torch.nn.Module):
+ def __init__(self, samp_rate, harmonic_num=0, sine_amp=0.1, noise_std=0.003, voiced_threshold=0):
+ super(SineGen, self).__init__()
+ self.sine_amp = sine_amp
+ self.noise_std = noise_std
+ self.harmonic_num = harmonic_num
+ self.dim = self.harmonic_num + 1
+ self.sampling_rate = samp_rate
+ self.voiced_threshold = voiced_threshold
+
+ def _f02uv(self, f0):
+ uv = torch.ones_like(f0)
+ uv = uv * (f0 > self.voiced_threshold)
+ return uv
+
+ def forward(self, f0, upp):
+ with torch.no_grad():
+ f0 = f0[:, None].transpose(1, 2)
+ f0_buf = torch.zeros(f0.shape[0], f0.shape[1], self.dim, device=f0.device)
+ f0_buf[:, :, 0] = f0[:, :, 0]
+ for idx in np.arange(self.harmonic_num): f0_buf[:, :, idx + 1] = f0_buf[:, :, 0] * (idx + 2)
+ rad_values = (f0_buf / self.sampling_rate) % 1
+ rand_ini = torch.rand(f0_buf.shape[0], f0_buf.shape[2], device=f0_buf.device)
+ rand_ini[:, 0] = 0
+ rad_values[:, 0, :] = rad_values[:, 0, :] + rand_ini
+ tmp_over_one = torch.cumsum(rad_values, 1)
+ tmp_over_one *= upp
+ tmp_over_one = F.interpolate(tmp_over_one.transpose(2, 1), scale_factor=upp, mode="linear", align_corners=True).transpose(2, 1)
+ rad_values = F.interpolate(rad_values.transpose(2, 1), scale_factor=upp, mode="nearest").transpose(2, 1)
+ tmp_over_one %= 1
+ tmp_over_one_idx = (tmp_over_one[:, 1:, :] - tmp_over_one[:, :-1, :]) < 0
+ cumsum_shift = torch.zeros_like(rad_values)
+ cumsum_shift[:, 1:, :] = tmp_over_one_idx * -1.0
+ sine_waves = torch.sin(torch.cumsum(rad_values + cumsum_shift, dim=1) * 2 * np.pi)
+ sine_waves = sine_waves * self.sine_amp
+ uv = self._f02uv(f0)
+ uv = F.interpolate(uv.transpose(2, 1), scale_factor=upp, mode="nearest").transpose(2, 1)
+ noise_amp = uv * self.noise_std + (1 - uv) * self.sine_amp / 3
+ noise = noise_amp * torch.randn_like(sine_waves)
+ sine_waves = sine_waves * uv + noise
+ return sine_waves, uv, noise
+
+
+class SourceModuleHnNSF(torch.nn.Module):
+ def __init__(
+ self,
+ sampling_rate,
+ harmonic_num=0,
+ sine_amp=0.1,
+ add_noise_std=0.003,
+ voiced_threshod=0,
+ is_half=True,
+ ):
+ super(SourceModuleHnNSF, self).__init__()
+
+ self.sine_amp = sine_amp
+ self.noise_std = add_noise_std
+ self.is_half = is_half
+ self.l_sin_gen = SineGen(sampling_rate, harmonic_num, sine_amp, add_noise_std, voiced_threshod)
+ self.l_linear = torch.nn.Linear(harmonic_num + 1, 1)
+ self.l_tanh = torch.nn.Tanh()
+
+ def forward(self, x, upp=None):
+ sine_wavs, uv, _ = self.l_sin_gen(x, upp)
+ if self.is_half: sine_wavs = sine_wavs.half()
+ sine_merge = self.l_tanh(self.l_linear(sine_wavs))
+ return sine_merge, None, None
+
+
+class GeneratorNSF(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,
+ sr,
+ is_half=False,
+ ):
+ super(GeneratorNSF, self).__init__()
+ self.num_kernels = len(resblock_kernel_sizes)
+ self.num_upsamples = len(upsample_rates)
+ self.f0_upsamp = torch.nn.Upsample(scale_factor=np.prod(upsample_rates))
+ self.m_source = SourceModuleHnNSF(sampling_rate=sr, harmonic_num=0, is_half=is_half)
+ self.noise_convs = nn.ModuleList()
+ 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)):
+ c_cur = upsample_initial_channel // (2 ** (i + 1))
+ self.ups.append(
+ weight_norm(ConvTranspose1d(upsample_initial_channel // (2**i), upsample_initial_channel // (2 ** (i + 1)), k, u, padding=(k - u) // 2)))
+ if i + 1 < len(upsample_rates):
+ stride_f0 = np.prod(upsample_rates[i + 1 :])
+ self.noise_convs.append(Conv1d(1, c_cur, kernel_size=stride_f0 * 2, stride=stride_f0, padding=stride_f0 // 2))
+ else: self.noise_convs.append(Conv1d(1, c_cur, kernel_size=1))
+
+ self.resblocks = nn.ModuleList()
+ for i in range(len(self.ups)):
+ ch = upsample_initial_channel // (2 ** (i + 1))
+ for k, d in 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)
+
+ self.upp = np.prod(upsample_rates)
+
+ def forward(self, x, f0, g=None):
+ har_source, noi_source, uv = self.m_source(f0, self.upp)
+ har_source = har_source.transpose(1, 2)
+ 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)
+ x_source = self.noise_convs[i](har_source)
+ x = x + x_source
+ 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):
+ for l in self.ups: remove_weight_norm(l)
+ for l in self.resblocks: l.remove_weight_norm()
+
+sr2sr = {
+ "32k": 32000,
+ "40k": 40000,
+ "48k": 48000,
+}
+
+class SynthesizerTrnMs256NSFsid(nn.Module):
+ def __init__(
+ self,
+ 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,
+ spk_embed_dim,
+ gin_channels,
+ sr,
+ **kwargs
+ ):
+ super().__init__()
+ if type(sr) == type("strr"): sr = sr2sr[sr]
+ 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.gin_channels = gin_channels
+ self.spk_embed_dim = spk_embed_dim
+ self.enc_p = TextEncoder256(inter_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout)
+ self.dec = GeneratorNSF(inter_channels, resblock, resblock_kernel_sizes, resblock_dilation_sizes, upsample_rates, upsample_initial_channel, upsample_kernel_sizes, gin_channels=gin_channels, sr=sr, is_half=kwargs["is_half"])
+ 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, 3, gin_channels=gin_channels)
+ self.emb_g = nn.Embedding(self.spk_embed_dim, gin_channels)
+
+ def remove_weight_norm(self):
+ self.dec.remove_weight_norm()
+ self.flow.remove_weight_norm()
+ self.enc_q.remove_weight_norm()
+
+ def forward(self, phone, phone_lengths, pitch, pitchf, y, y_lengths, ds):
+ g = self.emb_g(ds).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+ z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
+ z_p = self.flow(z, y_mask, g=g)
+ z_slice, ids_slice = commons.rand_slice_segments(z, y_lengths, self.segment_size)
+ pitchf = commons.slice_segments2(pitchf, ids_slice, self.segment_size)
+ o = self.dec(z_slice, pitchf, g=g)
+ return o, ids_slice, x_mask, y_mask, (z, z_p, m_p, logs_p, m_q, logs_q)
+
+ def infer(self, phone, phone_lengths, pitch, nsff0, sid, max_len=None):
+ g = self.emb_g(sid).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+ z_p = (m_p + torch.exp(logs_p) * torch.randn_like(m_p) * 0.66666) * x_mask
+ z = self.flow(z_p, x_mask, g=g, reverse=True)
+ o = self.dec((z * x_mask)[:, :, :max_len], nsff0, g=g)
+ return o, x_mask, (z, z_p, m_p, logs_p)
+
+
+class SynthesizerTrnMs768NSFsid(nn.Module):
+ def __init__(
+ self,
+ 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,
+ spk_embed_dim,
+ gin_channels,
+ sr,
+ **kwargs
+ ):
+ super().__init__()
+ if type(sr) == type("strr"): sr = sr2sr[sr]
+ 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.gin_channels = gin_channels
+ self.spk_embed_dim = spk_embed_dim
+ self.enc_p = TextEncoder768(inter_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout)
+ self.dec = GeneratorNSF(inter_channels, resblock, resblock_kernel_sizes, resblock_dilation_sizes, upsample_rates, upsample_initial_channel, upsample_kernel_sizes, gin_channels=gin_channels, sr=sr, is_half=kwargs["is_half"])
+ 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, 3, gin_channels=gin_channels)
+ self.emb_g = nn.Embedding(self.spk_embed_dim, gin_channels)
+
+ def remove_weight_norm(self):
+ self.dec.remove_weight_norm()
+ self.flow.remove_weight_norm()
+ self.enc_q.remove_weight_norm()
+
+ def forward(self, phone, phone_lengths, pitch, pitchf, y, y_lengths, ds):
+ g = self.emb_g(ds).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+ z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
+ z_p = self.flow(z, y_mask, g=g)
+ z_slice, ids_slice = commons.rand_slice_segments(z, y_lengths, self.segment_size)
+ pitchf = commons.slice_segments2(pitchf, ids_slice, self.segment_size)
+ o = self.dec(z_slice, pitchf, g=g)
+ return o, ids_slice, x_mask, y_mask, (z, z_p, m_p, logs_p, m_q, logs_q)
+
+ def infer(self, phone, phone_lengths, pitch, nsff0, sid, max_len=None):
+ g = self.emb_g(sid).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+ z_p = (m_p + torch.exp(logs_p) * torch.randn_like(m_p) * 0.66666) * x_mask
+ z = self.flow(z_p, x_mask, g=g, reverse=True)
+ o = self.dec((z * x_mask)[:, :, :max_len], nsff0, g=g)
+ return o, x_mask, (z, z_p, m_p, logs_p)
+
+class SynthesizerTrnMs256NSFsid_nono(nn.Module):
+ def __init__(
+ self,
+ 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,
+ spk_embed_dim,
+ gin_channels,
+ sr=None,
+ **kwargs
+ ):
+ super().__init__()
+ 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.gin_channels = gin_channels
+ self.spk_embed_dim = spk_embed_dim
+ self.enc_p = TextEncoder256(inter_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout, f0=False)
+ 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, 3, gin_channels=gin_channels)
+ self.emb_g = nn.Embedding(self.spk_embed_dim, gin_channels)
+
+ def remove_weight_norm(self):
+ self.dec.remove_weight_norm()
+ self.flow.remove_weight_norm()
+ self.enc_q.remove_weight_norm()
+
+ def forward(self, phone, phone_lengths, y, y_lengths, ds):
+ g = self.emb_g(ds).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, None, phone_lengths)
+ z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
+ z_p = self.flow(z, y_mask, g=g)
+ z_slice, ids_slice = commons.rand_slice_segments(z, y_lengths, self.segment_size)
+ o = self.dec(z_slice, g=g)
+ return o, ids_slice, x_mask, y_mask, (z, z_p, m_p, logs_p, m_q, logs_q)
+
+ def infer(self, phone, phone_lengths, sid, max_len=None):
+ g = self.emb_g(sid).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, None, phone_lengths)
+ z_p = (m_p + torch.exp(logs_p) * torch.randn_like(m_p) * 0.66666) * x_mask
+ z = self.flow(z_p, x_mask, g=g, reverse=True)
+ o = self.dec((z * x_mask)[:, :, :max_len], g=g)
+ return o, x_mask, (z, z_p, m_p, logs_p)
+
+class SynthesizerTrnMs768NSFsid_nono(nn.Module):
+ def __init__(
+ self,
+ 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,
+ spk_embed_dim,
+ gin_channels,
+ sr=None,
+ **kwargs
+ ):
+ super().__init__()
+ 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.gin_channels = gin_channels
+ self.spk_embed_dim = spk_embed_dim
+ self.enc_p = TextEncoder768(inter_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout, f0=False)
+ 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, 3, gin_channels=gin_channels)
+ self.emb_g = nn.Embedding(self.spk_embed_dim, gin_channels)
+
+ def remove_weight_norm(self):
+ self.dec.remove_weight_norm()
+ self.flow.remove_weight_norm()
+ self.enc_q.remove_weight_norm()
+
+ def forward(self, phone, phone_lengths, y, y_lengths, ds):
+ g = self.emb_g(ds).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, None, phone_lengths)
+ z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
+ z_p = self.flow(z, y_mask, g=g)
+ z_slice, ids_slice = commons.rand_slice_segments(z, y_lengths, self.segment_size)
+ o = self.dec(z_slice, g=g)
+ return o, ids_slice, x_mask, y_mask, (z, z_p, m_p, logs_p, m_q, logs_q)
+
+ def infer(self, phone, phone_lengths, sid, max_len=None):
+ g = self.emb_g(sid).unsqueeze(-1)
+ m_p, logs_p, x_mask = self.enc_p(phone, None, phone_lengths)
+ z_p = (m_p + torch.exp(logs_p) * torch.randn_like(m_p) * 0.66666) * x_mask
+ z = self.flow(z_p, x_mask, g=g, reverse=True)
+ o = self.dec((z * x_mask)[:, :, :max_len], g=g)
+ return o, x_mask, (z, z_p, m_p, logs_p)
+
+class MultiPeriodDiscriminator(torch.nn.Module):
+ def __init__(self, use_spectral_norm=False):
+ super(MultiPeriodDiscriminator, self).__init__()
+ periods = [2, 3, 5, 7, 11, 17]
+ discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
+ 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 d in 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 MultiPeriodDiscriminatorV2(torch.nn.Module):
+ def __init__(self, use_spectral_norm=False):
+ super(MultiPeriodDiscriminatorV2, self).__init__()
+ periods = [2, 3, 5, 7, 11, 17, 23, 37]
+
+ discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
+ 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 d in 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 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 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 = []
+ b, c, t = x.shape
+ if t % self.period != 0:
+ 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
\ No newline at end of file
diff --git a/lib/infer_pack/models_onnx.py b/lib/infer_pack/models_onnx.py
new file mode 100644
index 0000000000000000000000000000000000000000..1bbfb69e458245dc7215dfce1daa38827a4a3069
--- /dev/null
+++ b/lib/infer_pack/models_onnx.py
@@ -0,0 +1,582 @@
+import math
+import torch
+import numpy as np
+from torch import nn
+from torch.nn import functional as F
+from torch.nn import Conv1d, ConvTranspose1d, Conv2d
+from lib.infer_pack import modules, attentions, commons
+from torch.nn.utils import weight_norm, remove_weight_norm
+from lib.infer_pack.commons import init_weights, get_padding
+from lib.infer_pack.commons import init_weights, sequence_mask
+from torch.nn.utils import weight_norm, remove_weight_norm, spectral_norm
+from lib.infer_pack.modules import ResidualCouplingLayer, WN, ResBlock1, ResBlock2, LRELU_SLOPE
+
+class TextEncoder(nn.Module):
+ def __init__(
+ self,
+ input_dim,
+ out_channels,
+ hidden_channels,
+ filter_channels,
+ n_heads,
+ n_layers,
+ kernel_size,
+ p_dropout,
+ f0=True,
+ ):
+ super().__init__()
+ 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_phone = nn.Linear(input_dim, hidden_channels)
+ self.lrelu = nn.LeakyReLU(0.1, inplace=True)
+
+ if f0:self.emb_pitch = nn.Embedding(256, hidden_channels)
+ 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, phone, pitch, lengths):
+ x = self.emb_phone(phone) + self.emb_pitch(pitch) if pitch is not None else self.emb_phone(phone)
+ x *= math.sqrt(self.hidden_channels)
+ x = self.lrelu(x)
+ x = torch.transpose(x, 1, -1)
+ x_mask = torch.unsqueeze(sequence_mask(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 m, logs, x_mask
+
+class TextEncoder768(nn.Module):
+ def __init__(
+ self,
+ out_channels,
+ hidden_channels,
+ filter_channels,
+ n_heads,
+ n_layers,
+ kernel_size,
+ p_dropout,
+ f0=True,
+ ):
+ super().__init__()
+ 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_phone = nn.Linear(768, hidden_channels)
+ self.lrelu = nn.LeakyReLU(0.1, inplace=True)
+ if f0 == True:self.emb_pitch = nn.Embedding(256, hidden_channels)
+ 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, phone, pitch, lengths):
+ if pitch is None: x = self.emb_phone(phone)
+ else: x = self.emb_phone(phone) + self.emb_pitch(pitch)
+ x = x * math.sqrt(self.hidden_channels)
+ x = self.lrelu(x)
+ x = torch.transpose(x, 1, -1)
+ x_mask = torch.unsqueeze(commons.sequence_mask(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 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 _ 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
+
+ def remove_weight_norm(self):
+ for i in range(self.n_flows):
+ self.flows[i * 2].remove_weight_norm()
+
+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
+
+ def remove_weight_norm(self):
+ self.enc.remove_weight_norm()
+
+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 k, d in 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):
+ for l in self.ups: remove_weight_norm(l)
+ for l in self.resblocks: l.remove_weight_norm()
+
+class SineGen(torch.nn.Module):
+ def __init__(
+ self,
+ samp_rate,
+ harmonic_num=0,
+ sine_amp=0.1,
+ noise_std=0.003,
+ voiced_threshold=0,
+ ):
+ super(SineGen, self).__init__()
+ self.sine_amp = sine_amp
+ self.noise_std = noise_std
+ self.harmonic_num = harmonic_num
+ self.dim = self.harmonic_num + 1
+ self.sampling_rate = samp_rate
+ self.voiced_threshold = voiced_threshold
+
+ def _f02uv(self, f0):
+ uv = torch.ones_like(f0)
+ uv = uv * (f0 > self.voiced_threshold)
+ return uv
+
+ def forward(self, f0, upp):
+ with torch.no_grad():
+ f0 = f0[:, None].transpose(1, 2)
+ f0_buf = torch.zeros(f0.shape[0], f0.shape[1], self.dim, device=f0.device)
+ f0_buf[:, :, 0] = f0[:, :, 0]
+ for idx in np.arange(self.harmonic_num): f0_buf[:, :, idx + 1] = f0_buf[:, :, 0] * (idx + 2)
+ rad_values = (f0_buf / self.sampling_rate) % 1
+ rand_ini = torch.rand(f0_buf.shape[0], f0_buf.shape[2], device=f0_buf.device)
+ rand_ini[:, 0] = 0
+ rad_values[:, 0, :] = rad_values[:, 0, :] + rand_ini
+ tmp_over_one = torch.cumsum(rad_values, 1)
+ tmp_over_one *= upp
+ tmp_over_one = F.interpolate(tmp_over_one.transpose(2, 1), scale_factor=upp, mode="linear", align_corners=True).transpose(2, 1)
+ rad_values = F.interpolate(rad_values.transpose(2, 1), scale_factor=upp, mode="nearest").transpose(2, 1)
+ tmp_over_one %= 1
+ tmp_over_one_idx = (tmp_over_one[:, 1:, :] - tmp_over_one[:, :-1, :]) < 0
+ cumsum_shift = torch.zeros_like(rad_values)
+ cumsum_shift[:, 1:, :] = tmp_over_one_idx * -1.0
+ sine_waves = torch.sin(torch.cumsum(rad_values + cumsum_shift, dim=1) * 2 * np.pi)
+ sine_waves = sine_waves * self.sine_amp
+ uv = self._f02uv(f0)
+ uv = F.interpolate(uv.transpose(2, 1), scale_factor=upp, mode="nearest").transpose(2, 1)
+ noise_amp = uv * self.noise_std + (1 - uv) * self.sine_amp / 3
+ noise = noise_amp * torch.randn_like(sine_waves)
+ sine_waves = sine_waves * uv + noise
+ return sine_waves, uv, noise
+
+class SourceModuleHnNSF(torch.nn.Module):
+ def __init__(
+ self,
+ sampling_rate,
+ harmonic_num=0,
+ sine_amp=0.1,
+ add_noise_std=0.003,
+ voiced_threshod=0,
+ is_half=True,
+ ):
+ super(SourceModuleHnNSF, self).__init__()
+
+ self.sine_amp = sine_amp
+ self.noise_std = add_noise_std
+ self.is_half = is_half
+ self.l_sin_gen = SineGen(sampling_rate, harmonic_num, sine_amp, add_noise_std, voiced_threshod)
+ self.l_linear = torch.nn.Linear(harmonic_num + 1, 1)
+ self.l_tanh = torch.nn.Tanh()
+
+ def forward(self, x, upp=None):
+ sine_wavs, uv, _ = self.l_sin_gen(x, upp)
+ if self.is_half: sine_wavs = sine_wavs.half()
+ sine_merge = self.l_tanh(self.l_linear(sine_wavs))
+ return sine_merge, None, None
+
+
+class GeneratorNSF(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,
+ sr,
+ is_half=False,
+ ):
+ super(GeneratorNSF, self).__init__()
+ self.num_kernels = len(resblock_kernel_sizes)
+ self.num_upsamples = len(upsample_rates)
+ self.f0_upsamp = torch.nn.Upsample(scale_factor=np.prod(upsample_rates))
+ self.m_source = SourceModuleHnNSF(sampling_rate=sr, harmonic_num=0, is_half=is_half)
+ self.noise_convs = nn.ModuleList()
+ 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)):
+ c_cur = upsample_initial_channel // (2 ** (i + 1))
+ self.ups.append(weight_norm(ConvTranspose1d(upsample_initial_channel // (2**i), upsample_initial_channel // (2 ** (i + 1)), k, u, padding=(k - u) // 2,)))
+ if i + 1 < len(upsample_rates):
+ stride_f0 = np.prod(upsample_rates[i + 1 :])
+ self.noise_convs.append(Conv1d(1, c_cur, kernel_size=stride_f0 * 2, stride=stride_f0, padding=stride_f0 // 2,))
+ else: self.noise_convs.append(Conv1d(1, c_cur, kernel_size=1))
+
+ self.resblocks = nn.ModuleList()
+ for i in range(len(self.ups)):
+ ch = upsample_initial_channel // (2 ** (i + 1))
+ for k, d in 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)
+
+ self.upp = np.prod(upsample_rates)
+
+ def forward(self, x, f0, g=None):
+ har_source, noi_source, uv = self.m_source(f0, self.upp)
+ har_source = har_source.transpose(1, 2)
+ 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)
+ x_source = self.noise_convs[i](har_source)
+ x = x + x_source
+ 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):
+ for l in self.ups:
+ remove_weight_norm(l)
+ for l in self.resblocks:
+ l.remove_weight_norm()
+
+sr2sr = {"32k": 32000,"40k": 40000,"48k": 48000,}
+
+class SynthesizerTrnMsNSFsidM(nn.Module):
+ def __init__(
+ self,
+ 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,
+ spk_embed_dim,
+ gin_channels,
+ sr,
+ version,
+ **kwargs
+ ):
+ super().__init__()
+ if type(sr) == type("strr"): sr = sr2sr[sr]
+ 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.gin_channels = gin_channels
+ self.spk_embed_dim = spk_embed_dim
+ if version == "v1": self.enc_p = TextEncoder(inter_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout)
+ else: self.enc_p = TextEncoder768(inter_channels, hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout)
+ self.dec = GeneratorNSF(inter_channels, resblock, resblock_kernel_sizes, resblock_dilation_sizes, upsample_rates, upsample_initial_channel, upsample_kernel_sizes, gin_channels=gin_channels, sr=sr, is_half=kwargs["is_half"])
+ 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, 3, gin_channels=gin_channels)
+ self.emb_g = nn.Embedding(self.spk_embed_dim, gin_channels)
+ self.speaker_map = None
+
+ def remove_weight_norm(self):
+ self.dec.remove_weight_norm()
+ self.flow.remove_weight_norm()
+ self.enc_q.remove_weight_norm()
+
+ def construct_spkmixmap(self, n_speaker):
+ self.speaker_map = torch.zeros((n_speaker, 1, 1, self.gin_channels))
+ for i in range(n_speaker): self.speaker_map[i] = self.emb_g(torch.LongTensor([[i]]))
+ self.speaker_map = self.speaker_map.unsqueeze(0)
+
+ def forward(self, phone, phone_lengths, pitch, nsff0, g, rnd, max_len=None):
+ if self.speaker_map is not None:
+ g = g.reshape((g.shape[0], g.shape[1], 1, 1, 1))
+ g = g * self.speaker_map
+ g = torch.sum(g, dim=1)
+ g = g.transpose(0, -1).transpose(0, -2).squeeze(0)
+ else:
+ g = g.unsqueeze(0)
+ g = self.emb_g(g).transpose(1, 2)
+
+ m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+ z_p = (m_p + torch.exp(logs_p) * rnd) * x_mask
+ z = self.flow(z_p, x_mask, g=g, reverse=True)
+ return self.dec((z * x_mask)[:, :, :max_len], nsff0, g=g)
+
+class MultiPeriodDiscriminator(torch.nn.Module):
+ def __init__(self, use_spectral_norm=False):
+ super(MultiPeriodDiscriminator, self).__init__()
+ periods = [2, 3, 5, 7, 11, 17]
+
+ discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
+ 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 d in 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 MultiPeriodDiscriminatorV2(torch.nn.Module):
+ def __init__(self, use_spectral_norm=False):
+ super(MultiPeriodDiscriminatorV2, self).__init__()
+ periods = [2, 3, 5, 7, 11, 17, 23, 37]
+ discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
+ 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 d in 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 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 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 = []
+ b, c, t = x.shape
+ if t % self.period != 0:
+ 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
\ No newline at end of file
diff --git a/lib/infer_pack/modules.py b/lib/infer_pack/modules.py
new file mode 100644
index 0000000000000000000000000000000000000000..9e87efaec1cef72aac3e7e7a23fda26b0bb75ea7
--- /dev/null
+++ b/lib/infer_pack/modules.py
@@ -0,0 +1,315 @@
+import copy
+import math
+import numpy as np
+import scipy
+import torch
+from torch import nn
+from torch.nn import functional as F
+from torch.nn import Conv1d, ConvTranspose1d, AvgPool1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm
+from lib.infer_pack import commons
+from lib.infer_pack.commons import init_weights, get_padding
+from lib.infer_pack.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):
+ def __init__(self, channels, kernel_size, n_layers, p_dropout=0.0):
+ super().__init__()
+ self.channels = channels
+ self.kernel_size = kernel_size
+ self.n_layers = n_layers
+ self.p_dropout = p_dropout
+ self.drop = nn.Dropout(p_dropout)
+ self.convs_sep = nn.ModuleList()
+ self.convs_1x1 = nn.ModuleList()
+ self.norms_1 = nn.ModuleList()
+ self.norms_2 = nn.ModuleList()
+ for i in range(n_layers):
+ dilation = kernel_size**i
+ padding = (kernel_size * dilation - dilation)
+ 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)
+
+ 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 reverse: return x
+ logdet = torch.zeros(x.size(0)).to(dtype=x.dtype, device=x.device)
+ return x, logdet
+
+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
+
+ def remove_weight_norm(self): self.enc.remove_weight_norm()
+
+class ConvFlow(nn.Module):
+ def __init__(self, in_channels, filter_channels, kernel_size, n_layers, num_bins=10, tail_bound=5.0):
+ super().__init__()
+ self.in_channels = in_channels
+ self.filter_channels = filter_channels
+ self.kernel_size = kernel_size
+ self.n_layers = n_layers
+ self.num_bins = num_bins
+ self.tail_bound = tail_bound
+ self.half_channels = in_channels // 2
+
+ self.pre = nn.Conv1d(self.half_channels, filter_channels, 1)
+ self.convs = DDSConv(filter_channels, kernel_size, n_layers, p_dropout=0.0)
+ self.proj = nn.Conv1d(filter_channels, self.half_channels * (num_bins * 3 - 1), 1)
+ self.proj.weight.data.zero_()
+ self.proj.bias.data.zero_()
+
+ def forward(self, x, x_mask, g=None, reverse=False):
+ x0, x1 = torch.split(x, [self.half_channels] * 2, 1)
+ h = self.pre(x0)
+ h = self.convs(h, x_mask, g=g)
+ h = self.proj(h) * x_mask
+
+ b, c, t = x0.shape
+ h = h.reshape(b, c, -1, t).permute(0, 1, 3, 2)
+
+ 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])
+ return (x, logdet) if not reverse else x
\ No newline at end of file
diff --git a/lib/infer_pack/onnx_inference.py b/lib/infer_pack/onnx_inference.py
new file mode 100644
index 0000000000000000000000000000000000000000..a94b71dbe32fe6b339c9098363257b6f8ef9e62a
--- /dev/null
+++ b/lib/infer_pack/onnx_inference.py
@@ -0,0 +1,67 @@
+import onnxruntime
+import librosa
+import numpy as np
+from lib.infer_pack.modules.F0Predictor.PMF0Predictor import PMF0Predictor
+
+VEC_PATH = "pretrained/vec-768-layer-12.onnx"
+MAX_LENGTH = 50.0
+F0_MIN = 50
+F0_MAX = 1100
+F0_MEL_MIN = 1127 * np.log(1 + F0_MIN / 700)
+F0_MEL_MAX = 1127 * np.log(1 + F0_MAX / 700)
+RESAMPLING_RATE = 16000
+
+class ContentVectorModel:
+ def __init__(self, vector_path=VEC_PATH):
+ providers = ["CPUExecutionProvider"]
+ self.model = onnxruntime.InferenceSession(vector_path, providers=providers)
+
+ def __call__(self, audio_wave): return self.process_audio(audio_wave)
+
+ def process_audio(self, audio_wave):
+ features = audio_wave.mean(-1) if audio_wave.ndim == 2 else audio_wave
+ features = np.expand_dims(np.expand_dims(features, 0), 0)
+ onnx_input = {self.model.get_inputs()[0].name: features}
+ logits = self.model.run(None, onnx_input)[0]
+ return logits.transpose(0, 2, 1)
+
+
+class OnnxRVC:
+ def __init__(self, model_path, sampling_rate=40000, hop_size=512, vector_path=VEC_PATH):
+ self.vec_model = ContentVectorModel(f"{vector_path}.onnx")
+ providers = ["CPUExecutionProvider"]
+ self.model = onnxruntime.InferenceSession(model_path, providers=providers)
+ self.sampling_rate = sampling_rate
+ self.hop_size = hop_size
+
+ def forward(self, hubert, hubert_length, pitch, pitchf, ds, rnd):
+ onnx_input = {self.model.get_inputs()[0].name: hubert, self.model.get_inputs()[1].name: hubert_length, self.model.get_inputs()[2].name: pitch, self.model.get_inputs()[3].name: pitchf, self.model.get_inputs()[4].name: ds, self.model.get_inputs()[5].name: rnd}
+ return (self.model.run(None, onnx_input)[0] * 32767).astype(np.int16)
+
+ def inference(self, raw_path, sid, f0_method="pm", f0_up_key=0, pad_time=0.5, cr_threshold=0.02):
+ f0_predictor = PMF0Predictor(hop_length=self.hop_size, sampling_rate=self.sampling_rate, threshold=cr_threshold)
+ wav, sr = librosa.load(raw_path, sr=self.sampling_rate)
+ org_length = len(wav)
+ if org_length / sr > MAX_LENGTH: raise RuntimeError("Reached Max Length")
+
+ wav16k = librosa.resample(wav, orig_sr=sr, target_sr=RESAMPLING_RATE)
+ hubert = self.vec_model(wav16k)
+ hubert = np.repeat(hubert, 2, axis=2).transpose(0, 2, 1).astype(np.float32)
+ hubert_length = hubert.shape[1]
+
+ pitchf = f0_predictor.compute_f0(wav, hubert_length)
+ pitchf *= 2 ** (f0_up_key / 12)
+ pitch = pitchf.copy()
+ f0_mel = 1127 * np.log(1 + pitch / 700)
+ f0_mel = np.clip(f0_mel - F0_MEL_MIN, 0, None) * 254 / (F0_MEL_MAX - F0_MEL_MIN) + 1
+ pitch = np.rint(f0_mel).astype(np.int64)
+
+ pitchf = pitchf.reshape(1, -1).astype(np.float32)
+ pitch = pitch.reshape(1, -1)
+ ds = np.array([sid]).astype(np.int64)
+ rnd = np.random.randn(1, 192, hubert_length).astype(np.float32)
+ hubert_length = np.array([hubert_length]).astype(np.int64)
+
+ out_wav = self.forward(hubert, hubert_length, pitch, pitchf, ds, rnd).squeeze()
+ out_wav = np.pad(out_wav, (0, 2 * self.hop_size), "constant")
+ return out_wav[:org_length]
\ No newline at end of file
diff --git a/lib/infer_pack/transforms.py b/lib/infer_pack/transforms.py
new file mode 100644
index 0000000000000000000000000000000000000000..4de15fa1d9e8297bf2ee91d7cbf457617cc1ecef
--- /dev/null
+++ b/lib/infer_pack/transforms.py
@@ -0,0 +1,115 @@
+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=1.0, 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}
+
+ return 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)
+
+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=1.0, min_bin_width=DEFAULT_MIN_BIN_WIDTH, min_bin_height=DEFAULT_MIN_BIN_HEIGHT, min_derivative=DEFAULT_MIN_DERIVATIVE):
+ if tails != "linear": raise RuntimeError(f"{tails} tails are not implemented.")
+
+ 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
+
+ inside_interval_mask = (inputs >= -tail_bound) & (inputs <= tail_bound)
+ outside_interval_mask = ~inside_interval_mask
+ outputs = torch.where(outside_interval_mask, inputs, torch.zeros_like(inputs))
+ logabsdet = torch.zeros_like(inputs)
+
+ inside_outputs, inside_logabsdet = 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)
+
+ outputs[inside_interval_mask] = inside_outputs
+ logabsdet[inside_interval_mask] = inside_logabsdet
+
+ return outputs, logabsdet
+
+def rational_quadratic_spline(inputs, unnormalized_widths, unnormalized_heights, unnormalized_derivatives, inverse=False, left=0.0, right=1.0, bottom=0.0, top=1.0, min_bin_width=DEFAULT_MIN_BIN_WIDTH, min_bin_height=DEFAULT_MIN_BIN_HEIGHT, min_derivative=DEFAULT_MIN_DERIVATIVE):
+ 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, heights = compute_widths_and_heights(unnormalized_widths, unnormalized_heights, min_bin_width, min_bin_height, num_bins, left, right, bottom, top)
+ cumwidths, cumheights = widths.cumsum(dim=-1), heights.cumsum(dim=-1)
+ cumwidths[..., 0] = left
+ cumwidths[..., -1] = right
+ cumheights[..., 0] = bottom
+ cumheights[..., -1] = top
+ widths, heights = cumwidths[..., 1:] - cumwidths[..., :-1], cumheights[..., 1:] - cumheights[..., :-1]
+
+ derivatives = min_derivative + F.softplus(unnormalized_derivatives)
+
+ if inverse: bin_idx = searchsorted(cumheights, inputs)[..., None]
+ else: bin_idx = searchsorted(cumwidths, inputs)[..., None]
+
+ gather_args = (-1, bin_idx)
+ input_cumwidths, input_bin_widths, input_cumheights, input_delta, input_derivatives, input_derivatives_plus_one, input_heights = map(
+ lambda tensor: tensor.gather(*gather_args)[..., 0],
+ (cumwidths, widths, cumheights, heights / widths, derivatives, derivatives[..., 1:], heights))
+
+ if inverse: outputs, logabsdet = inverse_rational_quadratic_spline(inputs, input_cumheights, input_heights, input_derivatives, input_derivatives_plus_one, input_delta, input_bin_widths, input_cumwidths)
+ else: outputs, logabsdet = direct_rational_quadratic_spline(inputs, input_cumwidths, input_bin_widths, input_cumheights, input_heights, input_derivatives, input_derivatives_plus_one, input_delta)
+
+ return outputs, logabsdet
+
+def compute_widths_and_heights(unnormalized_widths, unnormalized_heights, min_bin_width, min_bin_height, num_bins, left, right, bottom, top):
+ widths = F.softmax(unnormalized_widths, dim=-1)
+ widths = min_bin_width + (1 - min_bin_width * num_bins) * widths
+ widths = (right - left) * widths + left
+
+ heights = F.softmax(unnormalized_heights, dim=-1)
+ heights = min_bin_height + (1 - min_bin_height * num_bins) * heights
+ heights = (top - bottom) * heights + bottom
+
+ return widths, heights
+
+def inverse_rational_quadratic_spline(inputs, input_cumheights, input_heights, input_derivatives, input_derivatives_plus_one, input_delta, input_bin_widths, input_cumwidths):
+ 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
+
+def direct_rational_quadratic_spline(inputs, input_cumwidths, input_bin_widths, input_cumheights, input_heights, input_derivatives, input_derivatives_plus_one, input_delta):
+ 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
\ No newline at end of file
diff --git a/models/FernandoAlonso/FernandoAlonso/added_IVF582_Flat_nprobe_1_fernando2_v2.index b/models/FernandoAlonso/FernandoAlonso/added_IVF582_Flat_nprobe_1_fernando2_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..7a96c205c5475b987f31ebe0d71a1b7919942d87
--- /dev/null
+++ b/models/FernandoAlonso/FernandoAlonso/added_IVF582_Flat_nprobe_1_fernando2_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5d70bed6e4bed0b54eaef9c1f2dd180fda94035287baafadb99a456b8cbd52
+size 71801099
diff --git a/models/FernandoAlonso/FernandoAlonso/fernando2.pth b/models/FernandoAlonso/FernandoAlonso/fernando2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3ed6f842549a0a69d8d3bad38c3461d4b741de0e
--- /dev/null
+++ b/models/FernandoAlonso/FernandoAlonso/fernando2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3926c2925ebe47d0c7a8d5b947813f0866a08d2970a10cb69d64f1b6e2371bbc
+size 55224197
diff --git a/models/HolasoyGerman/HolasoyGerman/HolasoyGerman.pth b/models/HolasoyGerman/HolasoyGerman/HolasoyGerman.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5c578cd14f44b8f610d4a150f8eb1d46f2b4859c
--- /dev/null
+++ b/models/HolasoyGerman/HolasoyGerman/HolasoyGerman.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db9c812c5ecd6e7afa5b0bee0300741ba87731097c6449b64924d46c10644247
+size 55226033
diff --git a/models/HolasoyGerman/HolasoyGerman/added_IVF3117_Flat_nprobe_1_HolasoyGerman_v2.index b/models/HolasoyGerman/HolasoyGerman/added_IVF3117_Flat_nprobe_1_HolasoyGerman_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..95a7d812a4bf7678e6cd83cf5e1312a08d7edc8b
--- /dev/null
+++ b/models/HolasoyGerman/HolasoyGerman/added_IVF3117_Flat_nprobe_1_HolasoyGerman_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a11850d919513a315e2a4ec41586a74c650f0ccaf3fc16d8aebff4e63ce07dc4
+size 384023779
diff --git a/models/HolasoyGerman/model_info.json b/models/HolasoyGerman/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..77858dc2ddd3e8d80c9cf475ae27c44a53457149
--- /dev/null
+++ b/models/HolasoyGerman/model_info.json
@@ -0,0 +1,9 @@
+{
+ "HolasoyGerman": {
+ "title": "HolasoyGerman",
+ "model_path": "HolasoyGerman.pth",
+ "feature_retrieval_library": "added_IVF3117_Flat_nprobe_1_HolasoyGerman_v2.index",
+ "model_path_checksum": "74633371eb2c2a75292633ee8922fddc",
+ "index_path_checksum": "febffc21f866063b40900cd068d379b9"
+ }
+}
\ No newline at end of file
diff --git a/models/Homer/Homer/HomerEsp.pth b/models/Homer/Homer/HomerEsp.pth
new file mode 100644
index 0000000000000000000000000000000000000000..32a6486a5df9c5eef70d9ca1944f692e922b910f
--- /dev/null
+++ b/models/Homer/Homer/HomerEsp.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:309244d07dd5f611f4bca2415a933a6bf499d1a803cd8df45e005d8256373b65
+size 55027130
diff --git a/models/Homer/Homer/added_IVF360_Flat_nprobe_1_HomerEsp_v1.index b/models/Homer/Homer/added_IVF360_Flat_nprobe_1_HomerEsp_v1.index
new file mode 100644
index 0000000000000000000000000000000000000000..37c6e9d8fb91934df637ec94b5dc30d943dbc7e9
--- /dev/null
+++ b/models/Homer/Homer/added_IVF360_Flat_nprobe_1_HomerEsp_v1.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3b3375a8e01675a1a46439af33ff4941f52eb9592b8784c154dcd55cf9eba03
+size 14861971
diff --git a/models/Homer/model_info.json b/models/Homer/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..3223a251e58e445390fe55af30744d4bb4acada8
--- /dev/null
+++ b/models/Homer/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Homer": {
+ "title": "Homer",
+ "model_path": "HomerEsp.pth",
+ "feature_retrieval_library": "added_IVF360_Flat_nprobe_1_HomerEsp_v1.index",
+ "model_path_checksum": "7df9a9cdf205fe13f5ea6d73d7e9fd41",
+ "index_path_checksum": "5d554ed74ef44a057d4f997dabe8c3ea"
+ }
+}
\ No newline at end of file
diff --git a/models/Ibai/Ibai/Ibai.pth b/models/Ibai/Ibai/Ibai.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fbee77e780c25b3ccb7b3245cf7744d909879afa
--- /dev/null
+++ b/models/Ibai/Ibai/Ibai.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa102e749f66ede51b3bc43a4cceaeeaac7abd26c9f20348a233d042349d0467
+size 55227410
diff --git a/models/Ibai/Ibai/added_IVF4601_Flat_nprobe_1_Ibai_v2.index b/models/Ibai/Ibai/added_IVF4601_Flat_nprobe_1_Ibai_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..c8385992ace5610ef984d1f4dc5c1d864aa98648
--- /dev/null
+++ b/models/Ibai/Ibai/added_IVF4601_Flat_nprobe_1_Ibai_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:177232ae8f7428fbc925719e11fcfd73a48b0919c3f94fd64eae6ff9f815c440
+size 566889539
diff --git a/models/Ibai/model_info.json b/models/Ibai/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..bc35ddb4bb5bee2a11cb49aa471b1b09ef75feb8
--- /dev/null
+++ b/models/Ibai/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Ibai": {
+ "title": "Ibai",
+ "model_path": "Ibai.pth",
+ "feature_retrieval_library": "added_IVF4601_Flat_nprobe_1_Ibai_v2.index",
+ "model_path_checksum": "098e96b7158edd654d0e269959ac9a20",
+ "index_path_checksum": "b16da2fa68a6341f8fc409a83db68dcf"
+ }
+}
\ No newline at end of file
diff --git a/models/IlloJuan/IlloJuan/IlloJuan.pth b/models/IlloJuan/IlloJuan/IlloJuan.pth
new file mode 100644
index 0000000000000000000000000000000000000000..25d419497c44f61ec13510d8fc987598fde07fc3
--- /dev/null
+++ b/models/IlloJuan/IlloJuan/IlloJuan.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80c3229a3b5ee888ffb9f8b85385de9705966b156ea4471619f25186318f55b4
+size 55223738
diff --git a/models/IlloJuan/IlloJuan/added_IVF593_Flat_nprobe_1_IlloJuan_v2.index b/models/IlloJuan/IlloJuan/added_IVF593_Flat_nprobe_1_IlloJuan_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..7bd424e273276a2b29034b9f91e61f875d4fafe1
--- /dev/null
+++ b/models/IlloJuan/IlloJuan/added_IVF593_Flat_nprobe_1_IlloJuan_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3bce6d0a1f554f5b5e38bb8f5d72e15a05c2383261f72e4fb5c06c2d714763f
+size 73162459
diff --git a/models/IlloJuan/model_info.json b/models/IlloJuan/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..2a5327087761dff0a1de98bc6f933e1b902d3ff3
--- /dev/null
+++ b/models/IlloJuan/model_info.json
@@ -0,0 +1,9 @@
+{
+ "IlloJuan": {
+ "title": "IlloJuan",
+ "model_path": "IlloJuan.pth",
+ "feature_retrieval_library": "added_IVF593_Flat_nprobe_1_IlloJuan_v2.index",
+ "model_path_checksum": "06a20ffd281284950537ebdbd227b8a9",
+ "index_path_checksum": "fcae1daa14cc24d862ca09216a02595c"
+ }
+}
\ No newline at end of file
diff --git a/models/Quevedo/Quevedo/added_IVF2301_Flat_nprobe_10.index b/models/Quevedo/Quevedo/added_IVF2301_Flat_nprobe_10.index
new file mode 100644
index 0000000000000000000000000000000000000000..489b1685d1b6fb487c0dfad5ffa5ac7b02dbf752
--- /dev/null
+++ b/models/Quevedo/Quevedo/added_IVF2301_Flat_nprobe_10.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fdfabf3c2a559c0c5649f9991e667157bb14440a49860dce02e06ec26a34846
+size 95005027
diff --git a/models/Quevedo/Quevedo/quevedo.pth b/models/Quevedo/Quevedo/quevedo.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5484624ec4a64440e1f4e0875fd8b7c267ad88d3
--- /dev/null
+++ b/models/Quevedo/Quevedo/quevedo.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:071f9b5e30ae0c219b47de25a9300b4097f5a7d8025946088ce5d505d9ae6227
+size 55026095
diff --git a/models/Quevedo/model_info.json b/models/Quevedo/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..7ce685bad91702640a9e99cdfd3eed42c13e4d69
--- /dev/null
+++ b/models/Quevedo/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Quevedo": {
+ "title": "Quevedo",
+ "model_path": "quevedo.pth",
+ "feature_retrieval_library": "added_IVF2301_Flat_nprobe_10.index",
+ "model_path_checksum": "6325e08a29a9cbfd0625dfe34be24a0c",
+ "index_path_checksum": "924db9553a0a70538a53e625fbadde40"
+ }
+}
\ No newline at end of file
diff --git a/models/Shadoune666/Shadoune666/Shadoune666.pth b/models/Shadoune666/Shadoune666/Shadoune666.pth
new file mode 100644
index 0000000000000000000000000000000000000000..d10297f15e2db1b4454c9bdd661fe8827e2cbb7e
--- /dev/null
+++ b/models/Shadoune666/Shadoune666/Shadoune666.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1447dd914d38c33b377b913c5ebc5f9bddf5a71195e2bb9b400af9bde6735e91
+size 55225115
diff --git a/models/Shadoune666/Shadoune666/added_IVF716_Flat_nprobe_1_Shadoune666_v2.index b/models/Shadoune666/Shadoune666/added_IVF716_Flat_nprobe_1_Shadoune666_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..75efbdb85bfc2fdf7205ec0afbf77e487acfdafc
--- /dev/null
+++ b/models/Shadoune666/Shadoune666/added_IVF716_Flat_nprobe_1_Shadoune666_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf519d09b556d5052f68a0be8628d10071e294de284291810b9ed88d2f8e5981
+size 88328379
diff --git a/models/Shadoune666/model_info.json b/models/Shadoune666/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..802bfdf0183648d070621a8fd8bb55e49d2776bb
--- /dev/null
+++ b/models/Shadoune666/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Shadoune666": {
+ "title": "Shadoune666",
+ "model_path": "Shadoune666.pth",
+ "feature_retrieval_library": "added_IVF716_Flat_nprobe_1_Shadoune666_v2.index",
+ "model_path_checksum": "7bca067c9055104ebea2bf3f6c7bfde4",
+ "index_path_checksum": "b23d61f502025b4f9d967e4de1935d0b"
+ }
+}
\ No newline at end of file
diff --git a/models/Spreen/Spreen/Spreen.pth b/models/Spreen/Spreen/Spreen.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b45005cc214452fc8e114abb313157eb6a3f8e46
--- /dev/null
+++ b/models/Spreen/Spreen/Spreen.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4be95ea3bb27866c12fafd46ee9997165d025f50e35260b23e2c175c942f6cff
+size 55216420
diff --git a/models/Spreen/Spreen/added_IVF4737_Flat_nprobe_1_Spreen_v2.index b/models/Spreen/Spreen/added_IVF4737_Flat_nprobe_1_Spreen_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..3d23bde2676752c69bf885aa63e3063248ce28ff
--- /dev/null
+++ b/models/Spreen/Spreen/added_IVF4737_Flat_nprobe_1_Spreen_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:669ca397e95bf51ff0e3f1aabe1ca60c83a5a6c85653ba0f1e550089c63b4305
+size 583660139
diff --git a/models/Spreen/model_info.json b/models/Spreen/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..75d081ba28fc3d14125b5859d11ee85451386815
--- /dev/null
+++ b/models/Spreen/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Spreen": {
+ "title": "Spreen",
+ "model_path": "Spreen.pth",
+ "feature_retrieval_library": "added_IVF4737_Flat_nprobe_1_Spreen_v2.index",
+ "model_path_checksum": "dbdb20d98928308c2d048444efb34b77",
+ "index_path_checksum": "213bbb9c387426fadc97c41e97f34b0a"
+ }
+}
\ No newline at end of file
diff --git a/models/Totakeke/Totakeke/added_IVF256_Flat_nprobe_1_full_totakeke_v2.index b/models/Totakeke/Totakeke/added_IVF256_Flat_nprobe_1_full_totakeke_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..5e57f260360da5260b043305e2bfafab4e93cd28
--- /dev/null
+++ b/models/Totakeke/Totakeke/added_IVF256_Flat_nprobe_1_full_totakeke_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b88bd6d822fa8391fa2ca6dd6e0c6eac8ac031385519da3fd87d2a6e8f272151
+size 31588619
diff --git a/models/Totakeke/Totakeke/full_totakeke_e150_s44400.pth b/models/Totakeke/Totakeke/full_totakeke_e150_s44400.pth
new file mode 100644
index 0000000000000000000000000000000000000000..265450b2e2419ff10fdefd7a8f073b5c399bf36c
--- /dev/null
+++ b/models/Totakeke/Totakeke/full_totakeke_e150_s44400.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3fb774cbbb74ae6005a87e370fc01337d3a24f7c5bb8c0962b335e4bd69488b
+size 57590901
diff --git a/models/Totakeke/model_info.json b/models/Totakeke/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..018fd930592d9d14256a03ad45a4426966540f6d
--- /dev/null
+++ b/models/Totakeke/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Totakeke": {
+ "title": "Totakeke",
+ "model_path": "full_totakeke_e150_s44400.pth",
+ "feature_retrieval_library": "added_IVF256_Flat_nprobe_1_full_totakeke_v2.index",
+ "model_path_checksum": "ccb3a07fd141c6e918eca8a969d95c5d",
+ "index_path_checksum": "d4b2d2662efdc0bcb8f3610a686525c9"
+ }
+}
\ No newline at end of file
diff --git a/models/Vegetta777/Vegetta777/Vegetta777.pth b/models/Vegetta777/Vegetta777/Vegetta777.pth
new file mode 100644
index 0000000000000000000000000000000000000000..acce06d49d6e5b680b0040afcd6a960b0be33ef3
--- /dev/null
+++ b/models/Vegetta777/Vegetta777/Vegetta777.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f13483615d05c1b31a4eb0fbee35ac5dca9f9a48a36a0165b947c08303ce75e5
+size 55218256
diff --git a/models/Vegetta777/Vegetta777/added_IVF2021_Flat_nprobe_1_Vegetta777_v2.index b/models/Vegetta777/Vegetta777/added_IVF2021_Flat_nprobe_1_Vegetta777_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..5f91aa9d368cf03c85d8659f0b042b42000c0363
--- /dev/null
+++ b/models/Vegetta777/Vegetta777/added_IVF2021_Flat_nprobe_1_Vegetta777_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd713395db0957eb626a40d096ee29ec8bebe19fac3d9ce356a3e587920c71d8
+size 248990419
diff --git a/models/Vegetta777/model_info.json b/models/Vegetta777/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a779126927076aadcef4e9532fa15126956d87d
--- /dev/null
+++ b/models/Vegetta777/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Vegetta777": {
+ "title": "Vegetta777",
+ "model_path": "Vegetta777.pth",
+ "feature_retrieval_library": "added_IVF2021_Flat_nprobe_1_Vegetta777_v2.index",
+ "model_path_checksum": "9bc80abfc506e98056e1db7b10000dea",
+ "index_path_checksum": "5a53e6f6a997e892e5fb26a7a6959550"
+ }
+}
\ No newline at end of file
diff --git a/models/Villager/Villager/added_IVF81_Flat_nprobe_1_v2.index b/models/Villager/Villager/added_IVF81_Flat_nprobe_1_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..a5c85b58f78306a29e0051e1833ae87a751cd8bd
--- /dev/null
+++ b/models/Villager/Villager/added_IVF81_Flat_nprobe_1_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8b8f5b445b1a1c3ee7155f3a121075399ca5b5a01032c459168b739c7c633e5
+size 10013219
diff --git a/models/Villager/Villager/villager.pth b/models/Villager/Villager/villager.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fdb494ec8d7baf037c7a85671f36edb96c071651
--- /dev/null
+++ b/models/Villager/Villager/villager.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2d94a1084bc9d65cd60ab7fb229a83476e79f8b75ec89e7343d0b9352ea1739
+size 55223738
diff --git a/models/Villager/model_info.json b/models/Villager/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..fcad95eda470beb5594903c7ba7a69d0218b89ba
--- /dev/null
+++ b/models/Villager/model_info.json
@@ -0,0 +1,9 @@
+{
+ "Villager": {
+ "title": "Villager",
+ "model_path": "villager.pth",
+ "feature_retrieval_library": "added_IVF81_Flat_nprobe_1_v2.index",
+ "model_path_checksum": "587b2be8f6fa60f7533dddf1a007e2fa",
+ "index_path_checksum": "13ef69f0f2181f77d45c4d36d7364f2b"
+ }
+}
\ No newline at end of file
diff --git a/models/WalterWhite/WalterWhite/added_IVF458_Flat_nprobe_1_ww2_v2.index b/models/WalterWhite/WalterWhite/added_IVF458_Flat_nprobe_1_ww2_v2.index
new file mode 100644
index 0000000000000000000000000000000000000000..fbe46fd844efe40a4f3588e24b05387dc0de51c9
--- /dev/null
+++ b/models/WalterWhite/WalterWhite/added_IVF458_Flat_nprobe_1_ww2_v2.index
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2030573a1685644ec51b816bb93194d813a30aa7e066fe27b849dc1e80f7706
+size 56527379
diff --git a/models/WalterWhite/WalterWhite/ww2.pth b/models/WalterWhite/WalterWhite/ww2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..bf8a9a587daa8aa96953e5d4d03bffe12ed2ca77
--- /dev/null
+++ b/models/WalterWhite/WalterWhite/ww2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82e6f75403e2cb820df885fbebef827973494d55176d8e523f60e209d7cc03ea
+size 57551491
diff --git a/models/WalterWhite/model_info.json b/models/WalterWhite/model_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..394bbc37497a18bfccc59045f2a390c603d8f47a
--- /dev/null
+++ b/models/WalterWhite/model_info.json
@@ -0,0 +1,9 @@
+{
+ "WalterWhite": {
+ "title": "WalterWhite",
+ "model_path": "ww2.pth",
+ "feature_retrieval_library": "added_IVF458_Flat_nprobe_1_ww2_v2.index",
+ "model_path_checksum": "dca6c0e3977e4fa61edb7e7324cf882a",
+ "index_path_checksum": "88108c501ab0ff5c0cc13c3d6df15374"
+ }
+}
\ No newline at end of file
diff --git a/models/folder_info.json b/models/folder_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..9fa042322316a50cb8d393a862430acfbc8af0f3
--- /dev/null
+++ b/models/folder_info.json
@@ -0,0 +1,54 @@
+{
+ "HolasoyGerman": {
+ "title": "HolasoyGerman",
+ "folder_path": "HolasoyGerman"
+ },
+ "Homer": {
+ "title": "Homer",
+ "folder_path": "Homer"
+ },
+ "Ibai": {
+ "title": "Ibai",
+ "folder_path": "Ibai"
+ },
+ "IlloJuan": {
+ "title": "IlloJuan",
+ "folder_path": "IlloJuan"
+ },
+ "Joseju": {
+ "title": "Joseju",
+ "folder_path": "Joseju"
+ },
+ "Pavloh": {
+ "title": "Pavloh",
+ "folder_path": "Pavloh"
+ },
+ "Quevedo": {
+ "title": "Quevedo",
+ "folder_path": "Quevedo"
+ },
+ "Shadoune666": {
+ "title": "Shadoune666",
+ "folder_path": "Shadoune666"
+ },
+ "Spreen": {
+ "title": "Spreen",
+ "folder_path": "Spreen"
+ },
+ "Totakeke": {
+ "title": "Totakeke",
+ "folder_path": "Totakeke"
+ },
+ "Vegetta777": {
+ "title": "Vegetta777",
+ "folder_path": "Vegetta777"
+ },
+ "Villager": {
+ "title": "Villager",
+ "folder_path": "Villager"
+ },
+ "WalterWhite": {
+ "title": "WalterWhite",
+ "folder_path": "WalterWhite"
+ }
+}
\ No newline at end of file
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..f92047aedaa2077a82279ce1b2fea77921258a9d
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,25 @@
+wheel
+PyNaCl
+demucs
+ffmpeg
+asyncio
+discord
+requests
+edge-tts
+setuptools
+torchcrepe
+discord.py
+onnxruntime
+tensorboard
+configparser
+tensorboardX
+scipy==1.9.3
+httpx==0.23.0
+numba==0.56.4
+numpy==1.23.5
+librosa==0.9.1
+fairseq==0.12.2
+faiss-cpu==1.7.3
+pyworld>=0.3.2
+soundfile>=0.12.1
+praat-parselmouth>=0.4.2
\ No newline at end of file
diff --git a/vc_infer_pipeline.py b/vc_infer_pipeline.py
new file mode 100644
index 0000000000000000000000000000000000000000..2e5d17bdc522e3d2757f2accd17258994b40e613
--- /dev/null
+++ b/vc_infer_pipeline.py
@@ -0,0 +1,230 @@
+from scipy import signal
+from functools import lru_cache
+import torch.nn.functional as F
+import numpy as np, parselmouth, torch
+import pyworld, os, traceback, faiss, librosa
+
+bh, ah = signal.butter(N=5, Wn=48, btype="high", fs=16000)
+
+@lru_cache
+def cache_harvest_f0(input_audio_path, fs, f0max, f0min, frame_period, input_audio_path2wav):
+ audio = input_audio_path2wav[input_audio_path]
+ f0, t = pyworld.harvest(
+ audio, fs=fs, f0_ceil=f0max, f0_floor=f0min, frame_period=frame_period
+ )
+ f0 = pyworld.stonemask(audio, f0, t, fs)
+ return f0
+
+def change_rms(data1, sr1, data2, sr2, rate):
+ rms1 = librosa.feature.rms(y=data1, frame_length=sr1 // 2 * 2, hop_length=sr1 // 2)
+ rms2 = librosa.feature.rms(y=data2, frame_length=sr2 // 2 * 2, hop_length=sr2 // 2)
+ rms1 = torch.from_numpy(rms1).unsqueeze(0)
+ rms2 = torch.from_numpy(rms2).unsqueeze(0)
+ rms1 = F.interpolate(rms1, size=data2.shape[0], mode="linear").squeeze()
+ rms2 = F.interpolate(rms2, size=data2.shape[0], mode="linear").squeeze()
+ rms2 = torch.max(rms2, torch.zeros_like(rms2) + 1e-6)
+ data2 *= (torch.pow(rms1, 1 - rate) * torch.pow(rms2, rate - 1)).numpy()
+
+ return data2
+
+class VC:
+ def __init__(self, tgt_sr, config):
+ self.x_pad = config.x_pad
+ self.x_query = config.x_query
+ self.x_center = config.x_center
+ self.x_max = config.x_max
+ self.is_half = config.is_half
+ self.sr = 16000
+ self.window = 160
+ self.t_pad = self.sr * self.x_pad
+ self.t_pad_tgt = tgt_sr * self.x_pad
+ self.t_pad2 = self.t_pad * 2
+ self.t_query = self.sr * self.x_query
+ self.t_center = self.sr * self.x_center
+ self.t_max = self.sr * self.x_max
+ self.device = config.device
+
+ def get_f0(self, input_audio_path, x, p_len, f0_up_key, f0_method, filter_radius, inp_f0=None):
+ global input_audio_path2wav
+ time_step = self.window / self.sr * 1000
+ f0_min = 50
+ f0_max = 1100
+ f0_mel_min = 1127 * np.log(1 + f0_min / 700)
+ f0_mel_max = 1127 * np.log(1 + f0_max / 700)
+
+ if f0_method == "pm":
+ f0 = (parselmouth.Sound(x, self.sr)
+ .to_pitch_ac(time_step=time_step / 1000, voicing_threshold=0.6, pitch_floor=f0_min,
+ pitch_ceiling=f0_max,)
+ .selected_array["frequency"])
+ pad_size = (p_len - len(f0) + 1) // 2
+ if pad_size > 0 or p_len - len(f0) - pad_size > 0:
+ f0 = np.pad(f0, [[pad_size, p_len - len(f0) - pad_size]], mode="constant")
+ f0 *= pow(2, f0_up_key / 12)
+ tf0 = self.sr // self.window
+
+ if inp_f0 is not None:
+ delta_t = np.round(
+ (inp_f0[:, 0].max() - inp_f0[:, 0].min()) * tf0 + 1
+ ).astype("int16")
+ replace_f0 = np.interp(list(range(delta_t)), inp_f0[:, 0] * 100, inp_f0[:, 1])
+ shape = f0[self.x_pad * tf0: self.x_pad * tf0 + len(replace_f0)].shape[0]
+ f0[self.x_pad * tf0: self.x_pad * tf0 + len(replace_f0)] = replace_f0[:shape]
+
+ f0bak= f0.copy()
+ f0_mel = 1127 * np.log(1 + f0 / 700)
+ f0_mel[f0_mel > 0] = (f0_mel[f0_mel > 0] - f0_mel_min) * 254 / (f0_mel_max - f0_mel_min) + 1
+ f0_mel[f0_mel <= 1] = 1
+ f0_mel[f0_mel > 255] = 255
+ f0_coarse = np.rint(f0_mel).astype(np.int)
+ return f0_coarse, f0bak
+
+ def vc(self, model, net_g, sid, audio0, pitch, pitchf, times, index, big_npy, index_rate, version, protect):
+ feats = torch.from_numpy(audio0)
+ feats = feats.half() if self.is_half else feats.float()
+
+ if feats.dim() == 2:
+ feats = feats.mean(-1)
+ assert feats.dim() == 1, feats.dim()
+ feats = feats.view(1, -1)
+ padding_mask = torch.BoolTensor(feats.shape).to(self.device).fill_(False)
+
+ inputs = { "source": feats.to(self.device), "padding_mask": padding_mask, "output_layer": 9 if version == "v1" else 12}
+
+ with torch.no_grad():
+ logits = model.extract_features(**inputs)
+ feats = model.final_proj(logits[0]) if version == "v1" else logits[0]
+
+ if protect < 0.5 and pitch is not None and pitchf is not None:
+ feats0 = feats.clone()
+
+ if index is not None and big_npy is not None and index_rate != 0:
+ npy = feats[0].cpu().numpy()
+ if self.is_half:
+ npy = npy.astype("float64")
+
+ score, ix = index.search(npy, k=8)
+ weight = np.square(1 / score)
+ weight /= weight.sum(axis=1, keepdims=True)
+ npy = np.sum(big_npy[ix] * np.expand_dims(weight, axis=2), axis=1)
+
+ if self.is_half:
+ npy = npy.astype("float16")
+ feats = ( torch.from_numpy(npy).unsqueeze(0).to(self.device) * index_rate + (1 - index_rate) * feats)
+
+ feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1)
+ if protect < 0.5 and pitch is not None and pitchf is not None:
+ feats0 = F.interpolate(feats0.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1)
+
+ p_len = audio0.shape[0] // self.window
+ if feats.shape[1] < p_len:
+ p_len = feats.shape[1]
+ if pitch is not None and pitchf is not None:
+ pitch = pitch[:, :p_len]
+ pitchf = pitchf[:, :p_len]
+
+ if protect < 0.5 and pitch is not None and pitchf is not None:
+ pitchff = pitchf.clone()
+ pitchff[pitchf > 0] = 1
+ pitchff[pitchf < 1] = protect
+ pitchff = pitchff.unsqueeze(-1)
+ feats = feats * pitchff + feats0 * (1 - pitchff)
+ feats = feats.to(feats0.dtype)
+ p_len = torch.tensor([p_len], device=self.device).long()
+
+ with torch.no_grad():
+ if pitch is not None and pitchf is not None:
+ audio1 = (
+ (net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0])
+ .data.cpu()
+ .float()
+ .numpy()
+ )
+ else:
+ audio1 = ((net_g.infer(feats, p_len, sid)[0][0, 0]).data.cpu().float().numpy())
+
+ del feats, p_len, padding_mask
+ return audio1
+
+ def pipeline(self,model, net_g, sid, audio, input_audio_path, times, f0_up_key, f0_method, file_index, index_rate, if_f0, filter_radius, tgt_sr, resample_sr, rms_mix_rate, version, protect, f0_file=None,):
+ if (
+ file_index != ""
+ and os.path.exists(file_index) == True
+ and index_rate != 0
+ ):
+ try:
+ index = faiss.read_index(file_index)
+ big_npy = index.reconstruct_n(0, index.ntotal)
+ except:
+ traceback.print_exc()
+ index = big_npy = None
+ else:
+ index = big_npy = None
+ audio = signal.filtfilt(bh, ah, audio)
+ audio_pad = np.pad(audio, (self.window // 2, self.window // 2), mode="reflect")
+ opt_ts = []
+ if audio_pad.shape[0] > self.t_max:
+ audio_sum = np.zeros_like(audio)
+ for i in range(self.window):
+ audio_sum += audio_pad[i : i - self.window]
+ for t in range(self.t_center, audio.shape[0], self.t_center):
+ opt_ts.append(
+ t
+ - self.t_query
+ + np.where(
+ np.abs(audio_sum[t - self.t_query : t + self.t_query])
+ == np.abs(audio_sum[t - self.t_query : t + self.t_query]).min()
+ )[0][0]
+ )
+ s = 0
+ audio_opt = []
+ t = None
+ audio_pad = np.pad(audio, (self.t_pad, self.t_pad), mode="reflect")
+ p_len = audio_pad.shape[0] // self.window
+ inp_f0 = None
+ if hasattr(f0_file, "name") == True:
+ try:
+ with open(f0_file.name, "r") as f:
+ lines = f.read().strip("\n").split("\n")
+ inp_f0 = []
+ for line in lines:
+ inp_f0.append([float(i) for i in line.split(",")])
+ inp_f0 = np.array(inp_f0, dtype="float64")
+ except:
+ traceback.print_exc()
+ sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
+ pitch, pitchf = None, None
+ if if_f0 == 1:
+ pitch, pitchf = self.get_f0(
+ input_audio_path,audio_pad,p_len,f0_up_key,f0_method,filter_radius,inp_f0,
+ )
+ pitch = pitch[:p_len]
+ pitchf = pitchf[:p_len]
+ pitch = torch.tensor(pitch, device=self.device).unsqueeze(0).long()
+ pitchf = torch.tensor(pitchf, device=self.device).unsqueeze(0).float()
+ for t in opt_ts:
+ t = t // self.window * self.window
+ if if_f0 == 1:
+ audio_opt.append(
+ self.vc(model, net_g, sid,audio_pad[s : t + self.t_pad2 + self.window], pitch[:, s // self.window : (t + self.t_pad2) // self.window],pitchf[:, s // self.window : (t + self.t_pad2) // self.window],times,index,big_npy,index_rate,version,protect,)[self.t_pad_tgt : -self.t_pad_tgt])
+ else:
+ audio_opt.append(self.vc(model,net_g,sid,audio_pad[s : t + self.t_pad2 + self.window],None,None,times,index,big_npy,index_rate,version,protect)[self.t_pad_tgt : -self.t_pad_tgt])
+ s = t
+ if if_f0 == 1:
+ audio_opt.append(self.vc(model,net_g,sid,audio_pad[t:],pitch[:, t // self.window :] if t is not None else pitch,pitchf[:, t // self.window :] if t is not None else pitchf,times,index,big_npy,index_rate,version,protect,)[self.t_pad_tgt : -self.t_pad_tgt])
+ else:
+ audio_opt.append(self.vc(model, net_g, sid, audio_pad[t:], None, None, times, index, big_npy, index_rate, version, protect)[self.t_pad_tgt : -self.t_pad_tgt])
+ audio_opt = np.concatenate(audio_opt)
+ if rms_mix_rate != 1:
+ audio_opt = change_rms(audio, 16000, audio_opt, tgt_sr, rms_mix_rate)
+ if resample_sr >= 16000 and tgt_sr != resample_sr:
+ audio_opt = librosa.resample(audio_opt, orig_sr=tgt_sr, target_sr=resample_sr)
+ audio_max = np.abs(audio_opt).max() / 0.99
+ max_int16 = 32768
+ if audio_max > 1:
+ max_int16 /= audio_max
+ audio_opt = (audio_opt * max_int16).astype(np.int16)
+ del pitch, pitchf, sid
+ if torch.cuda.is_available():
+ torch.cuda.empty_cache()
+ return audio_opt
\ No newline at end of file