deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-05 18:09:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

update

Browse Source
This commit is contained in:
lyuxiang.lx
2025-08-19 16:23:59 +08:00
parent cf615011ce
commit e3c2400abb
12 changed files with 891 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
cosyvoice/hifigan/f0_predictor.py
View File

@@ -17,6 +17,7 @@ try:
from torch.nn.utils.parametrizations import weight_norm
except ImportError:
from torch.nn.utils import weight_norm
from cosyvoice.transformer.convolution import CausalConv1d
class ConvRNNF0Predictor(nn.Module):
@@ -56,3 +57,47 @@ class ConvRNNF0Predictor(nn.Module):
x = self.condnet(x)
x = x.transpose(1, 2)
return torch.abs(self.classifier(x).squeeze(-1))
class CausalConvRNNF0Predictor(nn.Module):
def __init__(self,
num_class: int = 1,
in_channels: int = 80,
cond_channels: int = 512
):
super().__init__()
self.num_class = num_class
self.condnet = nn.Sequential(
weight_norm(
CausalConv1d(in_channels, cond_channels, kernel_size=4, causal_type='right')
),
nn.ELU(),
weight_norm(
CausalConv1d(cond_channels, cond_channels, kernel_size=3, causal_type='left')
),
nn.ELU(),
weight_norm(
CausalConv1d(cond_channels, cond_channels, kernel_size=3, causal_type='left')
),
nn.ELU(),
weight_norm(
CausalConv1d(cond_channels, cond_channels, kernel_size=3, causal_type='left')
),
nn.ELU(),
weight_norm(
CausalConv1d(cond_channels, cond_channels, kernel_size=3, causal_type='left')
),
nn.ELU(),
)
self.classifier = nn.Linear(in_features=cond_channels, out_features=self.num_class)
def forward(self, x: torch.Tensor, finalize: bool = True) -> torch.Tensor:
if finalize is True:
x = self.condnet[0](x)
else:
x = self.condnet[0](x[:, :, :-self.condnet[0].causal_padding], x[:, :, -self.condnet[0].causal_padding:])
for i in range(1, len(self.condnet)):
x = self.condnet[i](x)
x = x.transpose(1, 2)
return torch.abs(self.classifier(x).squeeze(-1))