change imports

GPT_SoVITS/AR/data/data_module.py

@@ -1,8 +1,8 @@
 # modified from https://github.com/yangdongchao/SoundStorm/blob/master/soundstorm/s1/AR/data/data_module.py
 # reference: https://github.com/lifeiteng/vall-e
 from pytorch_lightning import LightningDataModule
-from AR.data.bucket_sampler import DistributedBucketSampler
+from GPT_SoVITS.AR.data.bucket_sampler import DistributedBucketSampler
-from AR.data.dataset import Text2SemanticDataset
+from GPT_SoVITS.AR.data.dataset import Text2SemanticDataset
 from torch.utils.data import DataLoader
 
 

GPT_SoVITS/AR/models/t2s_lightning_module.py

@@ -8,9 +8,9 @@ from typing import Dict
 
 import torch
 from pytorch_lightning import LightningModule
-from AR.models.t2s_model import Text2SemanticDecoder
+from GPT_SoVITS.AR.models.t2s_model import Text2SemanticDecoder
-from AR.modules.lr_schedulers import WarmupCosineLRSchedule
+from GPT_SoVITS.AR.modules.lr_schedulers import WarmupCosineLRSchedule
-from AR.modules.optim import ScaledAdam
+from GPT_SoVITS.AR.modules.optim import ScaledAdam
 
 class Text2SemanticLightningModule(LightningModule):
     def __init__(self, config, output_dir, is_train=True):

GPT_SoVITS/AR/models/t2s_lightning_module_onnx.py

@@ -8,9 +8,9 @@ from typing import Dict
 
 import torch
 from pytorch_lightning import LightningModule
-from AR.models.t2s_model_onnx import Text2SemanticDecoder
+from GPT_SoVITS.AR.models.t2s_model_onnx import Text2SemanticDecoder
-from AR.modules.lr_schedulers import WarmupCosineLRSchedule
+from GPT_SoVITS.AR.modules.lr_schedulers import WarmupCosineLRSchedule
-from AR.modules.optim import ScaledAdam
+from GPT_SoVITS.AR.modules.optim import ScaledAdam
 
 
 class Text2SemanticLightningModule(LightningModule):

GPT_SoVITS/AR/models/t2s_model.py

@@ -5,8 +5,8 @@ from typing import List, Optional
 import torch
 from tqdm import tqdm
 
-from AR.models.utils import make_pad_mask
+from GPT_SoVITS.AR.models.utils import make_pad_mask
-from AR.models.utils import (
+from GPT_SoVITS.AR.models.utils import (
     topk_sampling,
     sample,
     logits_to_probs,
@@ -15,11 +15,11 @@ from AR.models.utils import (
     make_reject_y,
     get_batch_logps
 )
-from AR.modules.embedding import SinePositionalEmbedding
+from GPT_SoVITS.AR.modules.embedding import SinePositionalEmbedding
-from AR.modules.embedding import TokenEmbedding
+from GPT_SoVITS.AR.modules.embedding import TokenEmbedding
-from AR.modules.transformer import LayerNorm
+from GPT_SoVITS.AR.modules.transformer import LayerNorm
-from AR.modules.transformer import TransformerEncoder
+from GPT_SoVITS.AR.modules.transformer import TransformerEncoder
-from AR.modules.transformer import TransformerEncoderLayer
+from GPT_SoVITS.AR.modules.transformer import TransformerEncoderLayer
 from torch import nn
 from torch.nn import functional as F
 from torchmetrics.classification import MulticlassAccuracy

GPT_SoVITS/AR/models/t2s_model_onnx.py

@@ -3,11 +3,11 @@
 import torch
 from tqdm import tqdm
 
-from AR.modules.embedding_onnx import SinePositionalEmbedding
+from GPT_SoVITS.AR.modules.embedding_onnx import SinePositionalEmbedding
-from AR.modules.embedding_onnx import TokenEmbedding
+from GPT_SoVITS.AR.modules.embedding_onnx import TokenEmbedding
-from AR.modules.transformer_onnx import LayerNorm
+from GPT_SoVITS.AR.modules.transformer_onnx import LayerNorm
-from AR.modules.transformer_onnx import TransformerEncoder
+from GPT_SoVITS.AR.modules.transformer_onnx import TransformerEncoder
-from AR.modules.transformer_onnx import TransformerEncoderLayer
+from GPT_SoVITS.AR.modules.transformer_onnx import TransformerEncoderLayer
 from torch import nn
 from torch.nn import functional as F
 from torchmetrics.classification import MulticlassAccuracy

GPT_SoVITS/AR/modules/activation.py

@@ -12,7 +12,7 @@ from torch.nn.modules.linear import NonDynamicallyQuantizableLinear
 from torch.nn.parameter import Parameter
 
 from torch.nn import functional as F
-from AR.modules.patched_mha_with_cache import multi_head_attention_forward_patched
+from GPT_SoVITS.AR.modules.patched_mha_with_cache import multi_head_attention_forward_patched
 
 F.multi_head_attention_forward = multi_head_attention_forward_patched
 
@@ -152,14 +152,14 @@ class MultiheadAttention(Module):
                 self.in_proj_linear = linear1_cls(
                     embed_dim, 3 * embed_dim, bias=bias, **factory_kwargs
                 )
-                self.in_proj_weight = self.in_proj_linear.weight
+                self.in_proj_weight = self.in_proj_lineGPT_SoVITS.AR.weight
 
                 self.register_parameter("q_proj_weight", None)
                 self.register_parameter("k_proj_weight", None)
                 self.register_parameter("v_proj_weight", None)
 
                 if bias:
-                    self.in_proj_bias = self.in_proj_linear.bias
+                    self.in_proj_bias = self.in_proj_lineGPT_SoVITS.AR.bias
                 else:
                     self.register_parameter("in_proj_bias", None)
 

GPT_SoVITS/AR/modules/activation_onnx.py

@@ -12,7 +12,7 @@ from torch.nn.modules.linear import NonDynamicallyQuantizableLinear
 from torch.nn.parameter import Parameter
 
 from torch.nn import functional as F
-from AR.modules.patched_mha_with_cache_onnx import multi_head_attention_forward_patched
+from GPT_SoVITS.AR.modules.patched_mha_with_cache_onnx import multi_head_attention_forward_patched
 
 
 class MultiheadAttention(Module):
@@ -95,14 +95,14 @@ class MultiheadAttention(Module):
                 self.in_proj_linear = linear1_cls(
                     embed_dim, 3 * embed_dim, bias=bias, **factory_kwargs
                 )
-                self.in_proj_weight = self.in_proj_linear.weight
+                self.in_proj_weight = self.in_proj_lineGPT_SoVITS.AR.weight
 
                 self.register_parameter("q_proj_weight", None)
                 self.register_parameter("k_proj_weight", None)
                 self.register_parameter("v_proj_weight", None)
 
                 if bias:
-                    self.in_proj_bias = self.in_proj_linear.bias
+                    self.in_proj_bias = self.in_proj_lineGPT_SoVITS.AR.bias
                 else:
                     self.register_parameter("in_proj_bias", None)
 

GPT_SoVITS/AR/modules/transformer.py

@@ -10,8 +10,8 @@ from typing import Tuple
 from typing import Union
 
 import torch
-from AR.modules.activation import MultiheadAttention
+from GPT_SoVITS.AR.modules.activation import MultiheadAttention
-from AR.modules.scaling import BalancedDoubleSwish
+from GPT_SoVITS.AR.modules.scaling import BalancedDoubleSwish
 from torch import nn
 from torch import Tensor
 from torch.nn import functional as F

GPT_SoVITS/AR/modules/transformer_onnx.py

@@ -10,8 +10,8 @@ from typing import Tuple
 from typing import Union
 
 import torch
-from AR.modules.activation_onnx import MultiheadAttention
+from GPT_SoVITS.AR.modules.activation_onnx import MultiheadAttention
-from AR.modules.scaling import BalancedDoubleSwish
+from GPT_SoVITS.AR.modules.scaling import BalancedDoubleSwish
 from torch import nn
 from torch import Tensor
 from torch.nn import functional as F

GPT_SoVITS/AR/text_processing/phonemizer.py

@@ -9,7 +9,7 @@ import regex
 from gruut import sentences
 from gruut.const import Sentence
 from gruut.const import Word
-from AR.text_processing.symbols import SYMBOL_TO_ID
+from GPT_SoVITS.AR.text_processing.symbols import SYMBOL_TO_ID
 
 
 class GruutPhonemizer:

GPT_SoVITS/TTS_infer_pack/TTS.py

@@ -25,15 +25,15 @@ from typing import Generator, List, Tuple, Union
 from tqdm import tqdm
 
 
-from AR.models.t2s_lightning_module import Text2SemanticLightningModule
+from GPT_SoVITS.AR.models.t2s_lightning_module import Text2SemanticLightningModule
-from feature_extractor.cnhubert import CNHubert
+from GPT_SoVITS.feature_extractor.cnhubert import CNHubert
-from module.models import SynthesizerTrn
+from GPT_SoVITS.module.models import SynthesizerTrn
-from module.mel_processing import spectrogram_torch
+from GPT_SoVITS.module.mel_processing import spectrogram_torch
-from tools.i18n.i18n import I18nAuto, scan_language_list
+from GPT_SoVITS.tools.i18n.i18n import I18nAuto, scan_language_list
-from tools.my_utils import load_audio
+from GPT_SoVITS.tools.my_utils import load_audio
 from transformers import AutoModelForMaskedLM, AutoTokenizer
-from TTS_infer_pack.text_segmentation_method import splits
+from GPT_SoVITS.TTS_infer_pack.text_segmentation_method import splits
-from TTS_infer_pack.TextPreprocessor import TextPreprocessor
+from GPT_SoVITS.TTS_infer_pack.TextPreprocessor import TextPreprocessor
 
 language=os.environ.get("language","Auto")
 language=sys.argv[-1] if sys.argv[-1] in scan_language_list() else language

GPT_SoVITS/export_torch_script.py

@@ -13,7 +13,7 @@ from torch.nn import functional as F
 from transformers import AutoModelForMaskedLM, AutoTokenizer
 from feature_extractor import cnhubert
 
-from AR.models.t2s_lightning_module import Text2SemanticLightningModule
+from GPT_SoVITS.AR.models.t2s_lightning_module import Text2SemanticLightningModule
 from module.models_onnx import SynthesizerTrn
 
 from inference_webui import get_phones_and_bert

GPT_SoVITS/inference_gui.py

@@ -299,7 +299,7 @@ class GPTSoVITSGUI(QMainWindow):
 
             result = "Audio saved to " + output_wav_path
 
-        self.status_bar.showMessage("合成完成！输出路径：" + output_wav_path, 5000)
+        self.status_bGPT_SoVITS.AR.showMessage("合成完成！输出路径：" + output_wav_path, 5000)
         self.output_text.append("处理结果：\n" + result)
 
 

GPT_SoVITS/inference_webui.py

@@ -84,7 +84,7 @@ from feature_extractor import cnhubert
 cnhubert.cnhubert_base_path = cnhubert_base_path
 
 from module.models import SynthesizerTrn
-from AR.models.t2s_lightning_module import Text2SemanticLightningModule
+from GPT_SoVITS.AR.models.t2s_lightning_module import Text2SemanticLightningModule
 from text import cleaned_text_to_sequence
 from text.cleaner import clean_text
 from time import time as ttime

GPT_SoVITS/module/attentions.py

@@ -365,7 +365,7 @@ class MultiHeadAttention(nn.Module):
     def _attention_bias_proximal(self, length):
         """Bias for self-attention to encourage attention to close positions.
         Args:
-          length: an integer scalar.
+          length: an integer scalGPT_SoVITS.AR.
         Returns:
           a Tensor with shape [1, 1, length, length]
         """

GPT_SoVITS/module/attentions_onnx.py

@@ -303,7 +303,7 @@ class MultiHeadAttention(nn.Module):
     def _attention_bias_proximal(self, length):
         """Bias for self-attention to encourage attention to close positions.
         Args:
-          length: an integer scalar.
+          length: an integer scalGPT_SoVITS.AR.
         Returns:
           a Tensor with shape [1, 1, length, length]
         """

GPT_SoVITS/onnx_export.py

@@ -1,5 +1,5 @@
 from module.models_onnx import SynthesizerTrn, symbols_v1, symbols_v2
-from AR.models.t2s_lightning_module_onnx import Text2SemanticLightningModule
+from GPT_SoVITS.AR.models.t2s_lightning_module_onnx import Text2SemanticLightningModule
 import torch
 import torchaudio
 from torch import nn

GPT_SoVITS/s1_train.py

@@ -14,14 +14,14 @@ from pytorch_lightning import Trainer
 from pytorch_lightning.callbacks import ModelCheckpoint
 from pytorch_lightning.loggers import TensorBoardLogger  # WandbLogger
 from pytorch_lightning.strategies import DDPStrategy
-from AR.data.data_module import Text2SemanticDataModule
+from GPT_SoVITS.AR.data.data_module import Text2SemanticDataModule
-from AR.models.t2s_lightning_module import Text2SemanticLightningModule
+from GPT_SoVITS.AR.models.t2s_lightning_module import Text2SemanticLightningModule
-from AR.utils.io import load_yaml_config
+from GPT_SoVITS.AR.utils.io import load_yaml_config
 
 logging.getLogger("numba").setLevel(logging.WARNING)
 logging.getLogger("matplotlib").setLevel(logging.WARNING)
 torch.set_float32_matmul_precision("high")
-from AR.utils import get_newest_ckpt
+from GPT_SoVITS.AR.utils import get_newest_ckpt
 
 from collections import OrderedDict
 from time import time as ttime

GPT_SoVITS/tools/i18n/locale/pt_BR.json

@@ -106,7 +106,7 @@
     "开启一键三连": "Ativar um clique",
     "开启打标WebUI": "Abrir Labeling WebUI",
     "开启文本获取": "Ativar obtenção de texto",
-    "开启无参考文本模式。不填参考文本亦相当于开启。": "Ativar o modo sem texto de referência. Não preencher o texto de referência também equivale a ativar.",
+    "开启无参考文本模式。不填参考文本亦相当于开启。": "Ativar o modo sem texto de referência. Não preencher o texto de referência também equivale a ativGPT_SoVITS.AR.",
     "开启离线批量ASR": "Ativar ASR offline em lote",
     "开启语义token提取": "Ativar extração de token semântico",
     "开启语音切割": "Ativar corte de voz",

GPT_SoVITS/tools/uvr5/bsroformer.py

@@ -62,7 +62,7 @@ class BsRoformer_Loader:
         length_init = mix.shape[-1]
 
         progress_bar = tqdm(total=length_init // step + 1)
-        progress_bar.set_description("Processing")
+        progress_bGPT_SoVITS.AR.set_description("Processing")
 
         # Do pad from the beginning and end to account floating window results better
         if length_init > 2 * border and (border > 0):
@@ -102,7 +102,7 @@ class BsRoformer_Loader:
                     batch_data.append(part)
                     batch_locations.append((i, length))
                     i += step
-                    progress_bar.update(1)
+                    progress_bGPT_SoVITS.AR.update(1)
 
                     if len(batch_data) >= batch_size or (i >= mix.shape[1]):
                         arr = torch.stack(batch_data, dim=0)
@@ -131,7 +131,7 @@ class BsRoformer_Loader:
                     # Remove pad
                     estimated_sources = estimated_sources[..., border:-border]
 
-        progress_bar.close()
+        progress_bGPT_SoVITS.AR.close()
 
         return {k: v for k, v in zip(['vocals', 'other'], estimated_sources)}
 

GPT_SoVITS/tools/uvr5/mdxnet.py

@@ -143,7 +143,7 @@ class Predictor:
     def demix_base(self, mixes, margin_size):
         chunked_sources = []
         progress_bar = tqdm(total=len(mixes))
-        progress_bar.set_description("Processing")
+        progress_bGPT_SoVITS.AR.set_description("Processing")
         for mix in mixes:
             cmix = mixes[mix]
             sources = []
@@ -188,12 +188,12 @@ class Predictor:
                     end = None
                 sources.append(tar_signal[:, start:end])
 
-                progress_bar.update(1)
+                progress_bGPT_SoVITS.AR.update(1)
 
             chunked_sources.append(sources)
         _sources = np.concatenate(chunked_sources, axis=-1)
         # del self.model
-        progress_bar.close()
+        progress_bGPT_SoVITS.AR.close()
         return _sources
 
     def prediction(self, m, vocal_root, others_root, format):

api.py

@@ -163,7 +163,7 @@ import numpy as np
 from feature_extractor import cnhubert
 from io import BytesIO
 from module.models import SynthesizerTrn
-from AR.models.t2s_lightning_module import Text2SemanticLightningModule
+from GPT_SoVITS.AR.models.t2s_lightning_module import Text2SemanticLightningModule
 from text import cleaned_text_to_sequence
 from text.cleaner import clean_text
 from module.mel_processing import spectrogram_torch