Format import using isort --profile black --gitignore --lai 2 .

config.py

@@ -11,6 +11,7 @@ import yaml
 
 from style_bert_vits2.logging import logger
 
+
 # If not cuda available, set possible devices to cpu
 cuda_available = torch.cuda.is_available()
 

data_utils.py

@@ -15,6 +15,7 @@ from style_bert_vits2.models.hyper_parameters import HyperParametersData
 from style_bert_vits2.models.utils import load_filepaths_and_text, load_wav_to_torch
 from style_bert_vits2.nlp import cleaned_text_to_sequence
 
+
 """Multi speaker version"""
 
 

default_style.py

@@ -1,9 +1,10 @@
+import json
 import os
-from style_bert_vits2.constants import DEFAULT_STYLE
-from style_bert_vits2.logging import logger
 
 import numpy as np
-import json
+
+from style_bert_vits2.constants import DEFAULT_STYLE
+from style_bert_vits2.logging import logger
 
 
 def set_style_config(json_path, output_path):

gen_yaml.py

@@ -1,6 +1,7 @@
 import argparse
 import os
 import shutil
+
 import yaml
 
 

mel_processing.py

@@ -1,7 +1,9 @@
+import warnings
+
 import torch
 import torch.utils.data
 from librosa.filters import mel as librosa_mel_fn
-import warnings
+
 
 # warnings.simplefilter(action='ignore', category=FutureWarning)
 warnings.filterwarnings(action="ignore")

preprocess_all.py

@@ -1,7 +1,9 @@
 import argparse
-from webui.train import preprocess_all
 from multiprocessing import cpu_count
 
+from webui.train import preprocess_all
+
+
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
     parser.add_argument(

preprocess_text.py

@@ -12,6 +12,7 @@ from style_bert_vits2.logging import logger
 from style_bert_vits2.nlp import clean_text
 from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT
 
+
 preprocess_text_config = config.preprocess_text_config
 
 

resample.py

@@ -7,9 +7,10 @@ import pyloudnorm as pyln
 import soundfile
 from tqdm import tqdm
 
+from config import config
 from style_bert_vits2.logging import logger
 from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT
-from config import config
+
 
 DEFAULT_BLOCK_SIZE: float = 0.400  # seconds
 

server_fastapi.py

@@ -38,6 +38,7 @@ from style_bert_vits2.nlp import bert_models
 from style_bert_vits2.nlp.japanese import pyopenjtalk_worker as pyopenjtalk
 from style_bert_vits2.tts_model import TTSModel, TTSModelHolder
 
+
 ln = config.server_config.language
 
 

slice.py

@@ -11,6 +11,7 @@ from tqdm import tqdm
 from style_bert_vits2.logging import logger
 from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT
 
+
 vad_model, utils = torch.hub.load(
     repo_or_dir="snakers4/silero-vad",
     model="silero_vad",

speech_mos.py

@@ -14,6 +14,7 @@ from config import config
 from style_bert_vits2.logging import logger
 from style_bert_vits2.tts_model import TTSModel
 
+
 warnings.filterwarnings("ignore")
 
 mos_result_dir = Path("mos_results")

style_bert_vits2/models/attentions.py

@@ -1,6 +1,6 @@
+import math
 from typing import Any, Optional
 
-import math
 import torch
 from torch import nn
 from torch.nn import functional as F

style_bert_vits2/models/commons.py

@@ -3,9 +3,10 @@
 コードと完全に一致している保証はない。あくまで参考程度とすること。
 """
 
+from typing import Any, Optional, Union
+
 import torch
 from torch.nn import functional as F
-from typing import Any, Optional, Union
 
 
 def init_weights(m: torch.nn.Module, mean: float = 0.0, std: float = 0.01) -> None:

style_bert_vits2/models/infer.py

@@ -1,12 +1,11 @@
-from typing import Any, cast, Optional, Union
+from typing import Any, Optional, Union, cast
 
 import torch
 from numpy.typing import NDArray
 
 from style_bert_vits2.constants import Languages
 from style_bert_vits2.logging import logger
-from style_bert_vits2.models import commons
+from style_bert_vits2.models import commons, utils
-from style_bert_vits2.models import utils
 from style_bert_vits2.models.hyper_parameters import HyperParameters
 from style_bert_vits2.models.models import SynthesizerTrn
 from style_bert_vits2.models.models_jp_extra import (

style_bert_vits2/models/models.py

@@ -7,10 +7,7 @@ from torch.nn import Conv1d, Conv2d, ConvTranspose1d
 from torch.nn import functional as F
 from torch.nn.utils import remove_weight_norm, spectral_norm, weight_norm
 
-from style_bert_vits2.models import attentions
+from style_bert_vits2.models import attentions, commons, modules, monotonic_alignment
-from style_bert_vits2.models import commons
-from style_bert_vits2.models import modules
-from style_bert_vits2.models import monotonic_alignment
 from style_bert_vits2.nlp.symbols import NUM_LANGUAGES, NUM_TONES, SYMBOLS
 
 

style_bert_vits2/models/models_jp_extra.py

@@ -7,10 +7,7 @@ from torch.nn import Conv1d, Conv2d, ConvTranspose1d
 from torch.nn import functional as F
 from torch.nn.utils import remove_weight_norm, spectral_norm, weight_norm
 
-from style_bert_vits2.models import attentions
+from style_bert_vits2.models import attentions, commons, modules, monotonic_alignment
-from style_bert_vits2.models import commons
-from style_bert_vits2.models import modules
-from style_bert_vits2.models import monotonic_alignment
 from style_bert_vits2.nlp.symbols import NUM_LANGUAGES, NUM_TONES, SYMBOLS
 
 

style_bert_vits2/models/monotonic_alignment.py

@@ -3,10 +3,11 @@
 コードと完全に一致している保証はない。あくまで参考程度とすること。
 """
 
+from typing import Any
+
 import numba
 import torch
-from numpy import int32, float32, zeros
+from numpy import float32, int32, zeros
-from typing import Any
 
 
 def maximum_path(neg_cent: torch.Tensor, mask: torch.Tensor) -> torch.Tensor:

style_bert_vits2/models/utils/__init__.py

@@ -4,7 +4,7 @@ import os
 import re
 import subprocess
 from pathlib import Path
-from typing import Any, Optional, Union, TYPE_CHECKING
+from typing import TYPE_CHECKING, Any, Optional, Union
 
 import numpy as np
 import torch
@@ -15,6 +15,7 @@ from style_bert_vits2.logging import logger
 from style_bert_vits2.models.utils import checkpoints  # type: ignore
 from style_bert_vits2.models.utils import safetensors  # type: ignore
 
+
 if TYPE_CHECKING:
     # tensorboard はライブラリとしてインストールされている場合は依存関係に含まれないため、型チェック時のみインポートする
     from torch.utils.tensorboard import SummaryWriter

style_bert_vits2/nlp/__init__.py

@@ -1,4 +1,4 @@
-from typing import Optional, TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 from style_bert_vits2.constants import Languages
 from style_bert_vits2.nlp.symbols import (
@@ -7,6 +7,7 @@ from style_bert_vits2.nlp.symbols import (
     SYMBOLS,
 )
 
+
 # __init__.py は配下のモジュールをインポートした時点で実行される
 # PyTorch のインポートは重いので、型チェック時以外はインポートしない
 if TYPE_CHECKING:
@@ -99,10 +100,10 @@ def cleaned_text_to_sequence(
     cleaned_phones: list[str], tones: list[int], language: Languages
 ) -> tuple[list[int], list[int], list[int]]:
     """
-    テキスト文字列を、テキスト内の記号に対応する一連の ID に変換する
+    音素リスト・アクセントリスト・言語を、テキスト内の対応する ID に変換する
 
     Args:
-        cleaned_phones (list[str]): clean_text() でクリーニングされた音素のリスト (?)
+        cleaned_phones (list[str]): clean_text() でクリーニングされた音素のリスト
         tones (list[int]): 各音素のアクセント
         language (Languages): テキストの言語
 

style_bert_vits2/nlp/bert_models.py

@@ -9,7 +9,7 @@ Style-Bert-VITS2 の学習・推論に必要な各言語ごとの BERT モデル
 """
 
 import gc
-from typing import cast, Optional, Union
+from typing import Optional, Union, cast
 
 import torch
 from transformers import (

style_bert_vits2/nlp/chinese/g2p.py

@@ -2,7 +2,7 @@ import re
 from pathlib import Path
 
 import jieba.posseg as psg
-from pypinyin import lazy_pinyin, Style
+from pypinyin import Style, lazy_pinyin
 
 from style_bert_vits2.nlp.chinese.tone_sandhi import ToneSandhi
 from style_bert_vits2.nlp.symbols import PUNCTUATIONS

style_bert_vits2/nlp/chinese/tone_sandhi.py

@@ -13,8 +13,7 @@
 # limitations under the License.
 
 import jieba
-from pypinyin import lazy_pinyin
+from pypinyin import Style, lazy_pinyin
-from pypinyin import Style
 
 
 class ToneSandhi:

style_bert_vits2/nlp/japanese/normalizer.py

@@ -1,5 +1,6 @@
 import re
 import unicodedata
+
 from num2words import num2words
 
 from style_bert_vits2.nlp.symbols import PUNCTUATIONS

style_bert_vits2/nlp/japanese/user_dict/__init__.py

@@ -17,12 +17,13 @@ from fastapi import HTTPException
 
 from style_bert_vits2.constants import DEFAULT_USER_DICT_DIR
 from style_bert_vits2.nlp.japanese import pyopenjtalk_worker as pyopenjtalk
-from style_bert_vits2.nlp.japanese.user_dict.word_model import UserDictWord, WordTypes
 from style_bert_vits2.nlp.japanese.user_dict.part_of_speech_data import (
     MAX_PRIORITY,
     MIN_PRIORITY,
     part_of_speech_data,
 )
+from style_bert_vits2.nlp.japanese.user_dict.word_model import UserDictWord, WordTypes
+
 
 # root_dir = engine_root()
 # save_dir = get_save_dir()

style_bert_vits2/nlp/japanese/user_dict/part_of_speech_data.py

@@ -14,6 +14,7 @@ from style_bert_vits2.nlp.japanese.user_dict.word_model import (
     WordTypes,
 )
 
+
 MIN_PRIORITY = USER_DICT_MIN_PRIORITY
 MAX_PRIORITY = USER_DICT_MAX_PRIORITY
 

style_bert_vits2/nlp/japanese/user_dict/word_model.py

@@ -11,6 +11,7 @@ from typing import List, Optional
 
 from pydantic import BaseModel, Field, validator
 
+
 USER_DICT_MIN_PRIORITY = 0
 USER_DICT_MAX_PRIORITY = 10
 

style_gen.py

@@ -1,6 +1,6 @@
 import argparse
-from concurrent.futures import ThreadPoolExecutor
 import warnings
+from concurrent.futures import ThreadPoolExecutor
 
 import numpy as np
 import torch
@@ -11,9 +11,11 @@ from style_bert_vits2.logging import logger
 from style_bert_vits2.models.hyper_parameters import HyperParameters
 from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT
 
+
 warnings.filterwarnings("ignore", category=UserWarning)
 from pyannote.audio import Inference, Model
 
+
 model = Model.from_pretrained("pyannote/wespeaker-voxceleb-resnet34-LM")
 inference = Inference(model, window="whole")
 device = torch.device(config.style_gen_config.device)

train_ms.py

@@ -24,8 +24,7 @@ from data_utils import (
 from losses import discriminator_loss, feature_loss, generator_loss, kl_loss
 from mel_processing import mel_spectrogram_torch, spec_to_mel_torch
 from style_bert_vits2.logging import logger
-from style_bert_vits2.models import commons
+from style_bert_vits2.models import commons, utils
-from style_bert_vits2.models import utils
 from style_bert_vits2.models.hyper_parameters import HyperParameters
 from style_bert_vits2.models.models import (
     DurationDiscriminator,
@@ -35,6 +34,7 @@ from style_bert_vits2.models.models import (
 from style_bert_vits2.nlp.symbols import SYMBOLS
 from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT
 
+
 torch.backends.cuda.matmul.allow_tf32 = True
 torch.backends.cudnn.allow_tf32 = (
     True  # If encontered training problem,please try to disable TF32.

train_ms_jp_extra.py

@@ -24,8 +24,7 @@ from data_utils import (
 from losses import WavLMLoss, discriminator_loss, feature_loss, generator_loss, kl_loss
 from mel_processing import mel_spectrogram_torch, spec_to_mel_torch
 from style_bert_vits2.logging import logger
-from style_bert_vits2.models import commons
+from style_bert_vits2.models import commons, utils
-from style_bert_vits2.models import utils
 from style_bert_vits2.models.hyper_parameters import HyperParameters
 from style_bert_vits2.models.models_jp_extra import (
     DurationDiscriminator,
@@ -36,6 +35,7 @@ from style_bert_vits2.models.models_jp_extra import (
 from style_bert_vits2.nlp.symbols import SYMBOLS
 from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT
 
+
 torch.backends.cuda.matmul.allow_tf32 = True
 torch.backends.cudnn.allow_tf32 = (
     True  # If encontered training problem,please try to disable TF32.