refactor: remove duplicate get_padding

TTS/tts/layers/vits/networks.py

@@ -10,10 +10,6 @@ from TTS.tts.utils.helpers import sequence_mask
 LRELU_SLOPE = 0.1
 
 
-def get_padding(kernel_size, dilation=1):
-    return int((kernel_size * dilation - dilation) / 2)
-
-
 class TextEncoder(nn.Module):
     def __init__(
         self,

TTS/tts/layers/xtts/hifigan_decoder.py

@@ -9,16 +9,13 @@ from torch.nn.utils.parametrizations import weight_norm
 from torch.nn.utils.parametrize import remove_parametrizations
 
 from TTS.utils.io import load_fsspec
+from TTS.vocoder.models.hifigan_generator import get_padding
 
 logger = logging.getLogger(__name__)
 
 LRELU_SLOPE = 0.1
 
 
-def get_padding(k, d):
-    return int((k * d - d) / 2)
-
-
 class ResBlock1(torch.nn.Module):
     """Residual Block Type 1. It has 3 convolutional layers in each convolutional block.
 

TTS/vc/models/freevc.py

@@ -19,10 +19,11 @@ from TTS.tts.utils.speakers import SpeakerManager
 from TTS.utils.io import load_fsspec
 from TTS.vc.configs.freevc_config import FreeVCConfig
 from TTS.vc.models.base_vc import BaseVC
-from TTS.vc.modules.freevc.commons import get_padding, init_weights
+from TTS.vc.modules.freevc.commons import init_weights
 from TTS.vc.modules.freevc.mel_processing import mel_spectrogram_torch
 from TTS.vc.modules.freevc.speaker_encoder.speaker_encoder import SpeakerEncoder as SpeakerEncoderEx
 from TTS.vc.modules.freevc.wavlm import get_wavlm
+from TTS.vocoder.models.hifigan_generator import get_padding
 
 logger = logging.getLogger(__name__)
 

TTS/vc/modules/freevc/commons.py

@@ -12,10 +12,6 @@ def init_weights(m: torch.nn.Module, mean: float = 0.0, std: float = 0.01) -> No
         m.weight.data.normal_(mean, std)
 
 
-def get_padding(kernel_size, dilation=1):
-    return int((kernel_size * dilation - dilation) / 2)
-
-
 def intersperse(lst, item):
     result = [item] * (len(lst) * 2 + 1)
     result[1::2] = lst

TTS/vc/modules/freevc/modules.py

@@ -7,7 +7,8 @@ from torch.nn.utils.parametrize import remove_parametrizations
 
 import TTS.vc.modules.freevc.commons as commons
 from TTS.tts.layers.generic.normalization import LayerNorm2
-from TTS.vc.modules.freevc.commons import get_padding, init_weights
+from TTS.vc.modules.freevc.commons import init_weights
+from TTS.vocoder.models.hifigan_generator import get_padding
 
 LRELU_SLOPE = 0.1
 

TTS/vocoder/models/hifigan_discriminator.py

@@ -3,6 +3,8 @@ import torch
 from torch import nn
 from torch.nn import functional as F
 
+from TTS.vocoder.models.hifigan_generator import get_padding
+
 LRELU_SLOPE = 0.1
 
 
@@ -29,7 +31,6 @@ class DiscriminatorP(torch.nn.Module):
     def __init__(self, period, kernel_size=5, stride=3, use_spectral_norm=False):
         super().__init__()
         self.period = period
-        get_padding = lambda k, d: int((k * d - d) / 2)
         norm_f = nn.utils.spectral_norm if use_spectral_norm else nn.utils.parametrizations.weight_norm
         self.convs = nn.ModuleList(
             [

TTS/vocoder/models/hifigan_generator.py

@@ -15,8 +15,8 @@ logger = logging.getLogger(__name__)
 LRELU_SLOPE = 0.1
 
 
-def get_padding(k, d):
-    return int((k * d - d) / 2)
+def get_padding(kernel_size: int, dilation: int = 1) -> int:
+    return int((kernel_size * dilation - dilation) / 2)
 
 
 class ResBlock1(torch.nn.Module):