deviandice/DL-Art-School
1
0
Fork 0
forked from mrq/DL-Art-School
Code Issues Pull Requests Packages Projects Releases Wiki Activity

flat0 work

Browse Source
This commit is contained in:
James Betker 2022-03-24 11:46:53 -06:00
parent b0d2827fad
commit cc5fc91562
1 changed files with 15 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
codes/models/audio/tts/unet_diffusion_tts_flat0.py
View File

@ -153,14 +153,15 @@ class DiffusionTtsFlat(nn.Module):
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
)
self.code_norm = normalization(model_channels)
self.latent_converter = nn.Conv1d(in_latent_channels, model_channels, 1)
self.contextual_embedder = nn.Sequential(nn.Conv1d(in_channels,model_channels,3,padding=1,stride=2),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True))
self.conditioning_conv = nn.Conv1d(model_channels*2, model_channels, 1)
nn.Conv1d(model_channels, model_channels*2,3,padding=1,stride=2),
AttentionBlock(model_channels*2, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels*2, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels*2, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels*2, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels*2, num_heads, relative_pos_embeddings=True))
self.unconditioned_embedding = nn.Parameter(torch.randn(1,model_channels,1))
self.conditioning_timestep_integrator = TimestepEmbedSequential(
DiffusionLayer(model_channels, dropout, num_heads),
@ -207,9 +208,13 @@ class DiffusionTtsFlat(nn.Module):
code_emb = self.unconditioned_embedding.repeat(x.shape[0], 1, 1)
else:
unused_params.append(self.unconditioned_embedding)
cond_emb = self.contextual_embedder(conditioning_input)
if len(cond_emb.shape) == 3: # Just take the first element.
cond_emb = cond_emb[:, :, 0]
speech_conditioning_input = conditioning_input.unsqueeze(1) if len(conditioning_input.shape) == 3 else conditioning_input
conds = []
for j in range(speech_conditioning_input.shape[1]):
conds.append(self.contextual_embedder(speech_conditioning_input[:, j]))
conds = torch.cat(conds, dim=-1)
cond_emb = conds.mean(dim=-1)
cond_scale, cond_shift = torch.chunk(cond_emb, 2, dim=1)
if is_latent(aligned_conditioning):
code_emb = self.latent_converter(aligned_conditioning)
unused_params.extend(list(self.code_converter.parameters()) + list(self.code_embedding.parameters()))
@ -217,8 +222,7 @@ class DiffusionTtsFlat(nn.Module):
code_emb = self.code_embedding(aligned_conditioning).permute(0,2,1)
code_emb = self.code_converter(code_emb)
unused_params.extend(list(self.latent_converter.parameters()))
cond_emb_spread = cond_emb.unsqueeze(-1).repeat(1, 1, code_emb.shape[-1])
code_emb = self.conditioning_conv(torch.cat([cond_emb_spread, code_emb], dim=1))
code_emb = self.code_norm(code_emb) * (1 + cond_scale.unsqueeze(-1)) + cond_shift.unsqueeze(-1)
# Mask out the conditioning branch for whole batch elements, implementing something similar to classifier-free guidance.
if self.training and self.unconditioned_percentage > 0:
unconditioned_batches = torch.rand((code_emb.shape[0], 1, 1),