unify code style

app.py

@@ -3,7 +3,6 @@
 import os
 import logging
 
-# Constants
 DEFAULT_PORT = 6969
 MAX_PORT_ATTEMPTS = 10
 

rvc/configs/config.py

@@ -37,7 +37,7 @@ def __init__(self):
         self.gpu_mem = None
         self.x_pad, self.x_query, self.x_center, self.x_max = self.device_config()
 
-    def load_config_json(self) -> dict:
+    def load_config_json(self):
         configs = {}
         for config_file in version_config_paths:
             config_path = os.path.join("rvc", "configs", config_file)
@@ -97,7 +97,7 @@ def get_precision(self):
             print(f"File not found: {full_config_path}")
             return None
 
-    def device_config(self) -> tuple:
+    def device_config(self):
         if self.device.startswith("cuda"):
             self.set_cuda_config()
         else:

rvc/infer/pipeline.py

@@ -21,7 +21,6 @@
 
 logging.getLogger("faiss").setLevel(logging.WARNING)
 
-# Constants for high-pass filter
 FILTER_ORDER = 5
 CUTOFF_FREQUENCY = 48  # Hz
 SAMPLE_RATE = 16000  # Hz
@@ -43,7 +42,7 @@ def change_rms(
         target_audio: np.ndarray,
         target_rate: int,
         rate: float,
-    ) -> np.ndarray:
+    ):
         """
         Adjust the RMS level of target_audio to match the RMS of source_audio, with a given blending rate.
 

rvc/lib/algorithm/commons.py

@@ -157,7 +157,7 @@ def fused_add_tanh_sigmoid_multiply(input_a, input_b, n_channels):
     return acts
 
 
-def convert_pad_shape(pad_shape: List[List[int]]) -> List[int]:
+def convert_pad_shape(pad_shape: List[List[int]]):
     """
     Convert the pad shape to a list of integers.
 

rvc/lib/algorithm/generators.py

@@ -135,7 +135,7 @@ def __init__(
         self.voiced_threshold = voiced_threshold
         self.waveform_dim = self.num_harmonics + 1  # fundamental + harmonics
 
-    def _compute_voiced_unvoiced(self, f0: torch.Tensor) -> torch.Tensor:
+    def _compute_voiced_unvoiced(self, f0: torch.Tensor):
         """
         Generate a binary mask to indicate voiced/unvoiced frames.
 
@@ -145,9 +145,7 @@ def _compute_voiced_unvoiced(self, f0: torch.Tensor) -> torch.Tensor:
         uv_mask = (f0 > self.voiced_threshold).float()
         return uv_mask
 
-    def _generate_sine_wave(
-        self, f0: torch.Tensor, upsampling_factor: int
-    ) -> torch.Tensor:
+    def _generate_sine_wave(self, f0: torch.Tensor, upsampling_factor: int):
         """
         Generate sine waves for the fundamental frequency and its harmonics.
 

rvc/lib/predictors/F0Extractor.py

@@ -30,14 +30,14 @@ def __post_init__(self):
         self.x, self.sample_rate = librosa.load(self.wav_path, sr=self.sample_rate)
 
     @property
-    def hop_size(self) -> float:
+    def hop_size(self):
         return self.hop_length / self.sample_rate
 
     @property
-    def wav16k(self) -> np.ndarray:
+    def wav16k(self):
         return resampy.resample(self.x, self.sample_rate, 16000)
 
-    def extract_f0(self) -> np.ndarray:
+    def extract_f0(self):
         f0 = None
         method = self.method
         if method == "crepe":

rvc/lib/predictors/RMVPE.py

@@ -6,12 +6,10 @@
 from librosa.filters import mel
 from typing import List
 
-# Constants for readability
 N_MELS = 128
 N_CLASS = 360
 
 
-# Define a helper function for creating convolutional blocks
 class ConvBlockRes(nn.Module):
     """
     A convolutional block with residual connection.
@@ -59,7 +57,6 @@ def forward(self, x):
             return self.conv(x) + x
 
 
-# Define a class for residual encoder blocks
 class ResEncoderBlock(nn.Module):
     """
     A residual encoder block.
@@ -94,7 +91,6 @@ def forward(self, x):
             return x
 
 
-# Define a class for the encoder
 class Encoder(nn.Module):
     """
     The encoder part of the DeepUnet.
@@ -146,7 +142,6 @@ def forward(self, x: torch.Tensor):
         return x, concat_tensors
 
 
-# Define a class for the intermediate layer
 class Intermediate(nn.Module):
     """
     The intermediate layer of the DeepUnet.
@@ -177,7 +172,6 @@ def forward(self, x):
         return x
 
 
-# Define a class for residual decoder blocks
 class ResDecoderBlock(nn.Module):
     """
     A residual decoder block.
@@ -220,7 +214,6 @@ def forward(self, x, concat_tensor):
         return x
 
 
-# Define a class for the decoder
 class Decoder(nn.Module):
     """
     The decoder part of the DeepUnet.
@@ -250,7 +243,6 @@ def forward(self, x, concat_tensors):
         return x
 
 
-# Define a class for the DeepUnet architecture
 class DeepUnet(nn.Module):
     """
     The DeepUnet architecture.
@@ -294,7 +286,6 @@ def forward(self, x):
         return x
 
 
-# Define a class for the end-to-end model
 class E2E(nn.Module):
     """
     The end-to-end model.
@@ -348,7 +339,6 @@ def forward(self, mel):
         return x
 
 
-# Define a class for the MelSpectrogram extractor
 class MelSpectrogram(torch.nn.Module):
     """
     Extracts Mel-spectrogram features from audio.
@@ -432,7 +422,6 @@ def forward(self, audio, keyshift=0, speed=1, center=True):
         return log_mel_spec
 
 
-# Define a class for the RMVPE0 predictor
 class RMVPE0Predictor:
     """
     A predictor for fundamental frequency (F0) based on the RMVPE0 model.
@@ -535,7 +524,6 @@ def to_local_average_cents(self, salience, thred=0.05):
         return devided
 
 
-# Define a class for BiGRU (bidirectional GRU)
 class BiGRU(nn.Module):
     """
     A bidirectional GRU layer.

rvc/train/preprocess/preprocess.py

@@ -25,7 +25,6 @@
 logging.getLogger("numba.core.ssa").setLevel(logging.WARNING)
 logging.getLogger("numba.core.interpreter").setLevel(logging.WARNING)
 
-# Constants
 OVERLAP = 0.3
 MAX_AMPLITUDE = 0.9
 ALPHA = 0.75

tabs/extra/f0_extractor/f0_extractor.py

@@ -10,7 +10,7 @@
 i18n = I18nAuto()
 
 
-def extract_f0_curve(audio_path: str, method: str) -> tuple:
+def extract_f0_curve(audio_path: str, method: str):
     print("Extracting F0 Curve...")
     image_path = os.path.join("logs", "f0_plot.png")
     txt_path = os.path.join("logs", "f0_curve.txt")