Medix-AI/coqui-tts
Medix-AI
/
coqui-tts
mirror of https://github.com/coqui-ai/TTS.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fix Speaker Consistency Loss (SCL) (#2364)

This commit is contained in:
Edresson Casanova 2023-02-27 03:14:00 -03:00 committed by GitHub
parent a365a7e888
commit 16b9862252
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 7 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
TTS/encoder/models/resnet.py
View File

@ -161,16 +161,14 @@ class ResNetSpeakerEncoder(BaseEncoder):
Shapes: Shapes:
- x: :math:`(N, 1, T_{in})` or :math:`(N, D_{spec}, T_{in})` - x: :math:`(N, 1, T_{in})` or :math:`(N, D_{spec}, T_{in})`
""" """
with torch.no_grad(): x.squeeze_(1)
with torch.cuda.amp.autocast(enabled=False): # if you torch spec compute it otherwise use the mel spec computed by the AP
x.squeeze_(1) if self.use_torch_spec:
# if you torch spec compute it otherwise use the mel spec computed by the AP x = self.torch_spec(x)
if self.use_torch_spec:
x = self.torch_spec(x)
if self.log_input: if self.log_input:
x = (x + 1e-6).log() x = (x + 1e-6).log()
x = self.instancenorm(x).unsqueeze(1) x = self.instancenorm(x).unsqueeze(1)
x = self.conv1(x) x = self.conv1(x)
x = self.relu(x) x = self.relu(x)