mrq/DL-Art-School
1
3
Fork 5
Code Issues Pull Requests Packages Projects Releases Wiki Activity

deeper resblock encoder

Browse Source
This commit is contained in:
James Betker 2022-05-23 11:46:40 -06:00
parent bb36bb4db5
commit f432bdf7ae
1 changed files with 18 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
codes/models/audio/mel2vec.py
View File

@ -389,6 +389,24 @@ class Mel2Vec(nn.Module):
ResBlock(dims=1, channels=inner_dim, dropout=dropout),
)
self.dim_reduction_mult = 4
elif feature_producer_type == 'deep_residual':
self.input_blocks = nn.Sequential(nn.Conv1d(mel_input_channels, inner_dim//2, kernel_size=5, padding=2, stride=2),
nn.GroupNorm(num_groups=8, num_channels=inner_dim//2, affine=True),
nn.GELU(),
ResBlock(dims=1, channels=inner_dim//2, dropout=dropout),
ResBlock(dims=1, channels=inner_dim//2, dropout=dropout),
ResBlock(dims=1, channels=inner_dim//2, dropout=dropout),
ResBlock(dims=1, channels=inner_dim//2, dropout=dropout),
ResBlock(dims=1, channels=inner_dim//2, dropout=dropout),
nn.Conv1d(inner_dim//2, inner_dim, kernel_size=3, padding=1, stride=2),
nn.GELU(),
ResBlock(dims=1, channels=inner_dim, dropout=dropout),
ResBlock(dims=1, channels=inner_dim, dropout=dropout),
ResBlock(dims=1, channels=inner_dim, dropout=dropout),
ResBlock(dims=1, channels=inner_dim, dropout=dropout),
ResBlock(dims=1, channels=inner_dim, dropout=dropout),
)
self.dim_reduction_mult = 4
elif feature_producer_type == 'voice_8x':
self.input_blocks = nn.Sequential(nn.Conv1d(mel_input_channels, inner_dim//4, kernel_size=5, padding=2, stride=2),
nn.GroupNorm(num_groups=8, num_channels=inner_dim//4, affine=True),