Add files via upload

GPT_SoVITS/AR/utils/__init__.py

@@ -0,0 +1,37 @@
+import re
+
+
+def str2bool(str):
+    return True if str.lower() == 'true' else False
+
+
+def get_newest_ckpt(string_list):
+    # 定义一个正则表达式模式，用于匹配字符串中的数字
+    pattern = r'epoch=(\d+)-step=(\d+)\.ckpt'
+
+    # 使用正则表达式提取每个字符串中的数字信息，并创建一个包含元组的列表
+    extracted_info = []
+    for string in string_list:
+        match = re.match(pattern, string)
+        if match:
+            epoch = int(match.group(1))
+            step = int(match.group(2))
+            extracted_info.append((epoch, step, string))
+    # 按照 epoch 后面的数字和 step 后面的数字进行排序
+    sorted_info = sorted(
+        extracted_info, key=lambda x: (x[0], x[1]), reverse=True)
+    # 获取最新的 ckpt 文件名
+    newest_ckpt = sorted_info[0][2]
+    return newest_ckpt
+
+
+# 文本存在且不为空时 return True
+def check_txt_file(file_path):
+    try:
+        with open(file_path, 'r') as file:
+            text = file.readline().strip()
+        assert text.strip() != ''
+        return text
+    except Exception:
+        return False
+    return False

GPT_SoVITS/AR/utils/initialize.py

@@ -0,0 +1,38 @@
+#!/usr/bin/env python3
+"""Initialize modules for espnet2 neural networks."""
+import torch
+from typeguard import check_argument_types
+
+
+def initialize(model: torch.nn.Module, init: str):
+    """Initialize weights of a neural network module.
+
+    Parameters are initialized using the given method or distribution.
+
+    Custom initialization routines can be implemented into submodules
+    as function `espnet_initialization_fn` within the custom module.
+
+    Args:
+        model: Target.
+        init: Method of initialization.
+    """
+    assert check_argument_types()
+    print("init with", init)
+
+    # weight init
+    for p in model.parameters():
+        if p.dim() > 1:
+            if init == "xavier_uniform":
+                torch.nn.init.xavier_uniform_(p.data)
+            elif init == "xavier_normal":
+                torch.nn.init.xavier_normal_(p.data)
+            elif init == "kaiming_uniform":
+                torch.nn.init.kaiming_uniform_(p.data, nonlinearity="relu")
+            elif init == "kaiming_normal":
+                torch.nn.init.kaiming_normal_(p.data, nonlinearity="relu")
+            else:
+                raise ValueError("Unknown initialization: " + init)
+    # bias init
+    for name, p in model.named_parameters():
+        if ".bias" in name and p.dim() == 1:
+            p.data.zero_()

GPT_SoVITS/AR/utils/io.py

@@ -0,0 +1,32 @@
+import sys
+
+import torch
+import yaml
+
+
+def load_yaml_config(path):
+    with open(path) as f:
+        config = yaml.full_load(f)
+    return config
+
+
+def save_config_to_yaml(config, path):
+    assert path.endswith('.yaml')
+    with open(path, 'w') as f:
+        f.write(yaml.dump(config))
+        f.close()
+
+
+def write_args(args, path):
+    args_dict = dict((name, getattr(args, name)) for name in dir(args)
+                     if not name.startswith('_'))
+    with open(path, 'a') as args_file:
+        args_file.write('==> torch version: {}\n'.format(torch.__version__))
+        args_file.write(
+            '==> cudnn version: {}\n'.format(torch.backends.cudnn.version()))
+        args_file.write('==> Cmd:\n')
+        args_file.write(str(sys.argv))
+        args_file.write('\n==> args:\n')
+        for k, v in sorted(args_dict.items()):
+            args_file.write('  %s: %s\n' % (str(k), str(v)))
+        args_file.close()