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

testing seq_len_norm

This commit is contained in:
root 2020-01-27 16:02:34 +01:00
parent 7a616aa9ef
commit f4678cbd6b
4 changed files with 41 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
layers/losses.py
View File

@ -35,7 +35,7 @@ class L1LossMasked(nn.Module):
mask = mask.expand_as(x)
loss = functional.l1_loss(
x * mask, target * mask, reduction='none')
loss = loss.mul(out_weights.cuda()).sum()
loss = loss.mul(out_weights.to(loss.device)).sum()
else:
mask = mask.expand_as(x)
loss = functional.l1_loss(
@ -74,7 +74,7 @@ class MSELossMasked(nn.Module):
mask = mask.expand_as(x)
loss = functional.mse_loss(
x * mask, target * mask, reduction='none')
loss = loss.mul(out_weights.cuda()).sum()
loss = loss.mul(out_weights.to(loss.device)).sum()
else:
mask = mask.expand_as(x)
loss = functional.mse_loss(

41
tests/test_layers.py
View File

@ -119,7 +119,7 @@ class EncoderTests(unittest.TestCase):
class L1LossMaskedTests(unittest.TestCase):
def test_in_out(self):
# test input == target
layer = L1LossMasked()
layer = L1LossMasked(seq_len_norm=False)
dummy_input = T.ones(4, 8, 128).float()
dummy_target = T.ones(4, 8, 128).float()
dummy_length = (T.ones(4) * 8).long()
@ -131,7 +131,7 @@ class L1LossMaskedTests(unittest.TestCase):
dummy_target = T.zeros(4, 8, 128).float()
dummy_length = (T.ones(4) * 8).long()
output = layer(dummy_input, dummy_target, dummy_length)
assert output.item() == 1.0, "1.0 vs {}".format(output.data[0])
assert output.item() == 1.0, "1.0 vs {}".format(output.item())
# test if padded values of input makes any difference
dummy_input = T.ones(4, 8, 128).float()
@ -140,7 +140,7 @@ class L1LossMaskedTests(unittest.TestCase):
mask = (
(sequence_mask(dummy_length).float() - 1.0) * 100.0).unsqueeze(2)
output = layer(dummy_input + mask, dummy_target, dummy_length)
assert output.item() == 1.0, "1.0 vs {}".format(output.data[0])
assert output.item() == 1.0, "1.0 vs {}".format(output.item())
dummy_input = T.rand(4, 8, 128).float()
dummy_target = dummy_input.detach()
@ -148,4 +148,37 @@ class L1LossMaskedTests(unittest.TestCase):
mask = (
(sequence_mask(dummy_length).float() - 1.0) * 100.0).unsqueeze(2)
output = layer(dummy_input + mask, dummy_target, dummy_length)
assert output.item() == 0, "0 vs {}".format(output.data[0])
assert output.item() == 0, "0 vs {}".format(output.item())
# seq_len_norm = True
# test input == target
layer = L1LossMasked(seq_len_norm=True)
dummy_input = T.ones(4, 8, 128).float()
dummy_target = T.ones(4, 8, 128).float()
dummy_length = (T.ones(4) * 8).long()
output = layer(dummy_input, dummy_target, dummy_length)
assert output.item() == 0.0
# test input != target
dummy_input = T.ones(4, 8, 128).float()
dummy_target = T.zeros(4, 8, 128).float()
dummy_length = (T.ones(4) * 8).long()
output = layer(dummy_input, dummy_target, dummy_length)
assert output.item() == 1.0, "1.0 vs {}".format(output.item())
# test if padded values of input makes any difference
dummy_input = T.ones(4, 8, 128).float()
dummy_target = T.zeros(4, 8, 128).float()
dummy_length = (T.arange(5, 9)).long()
mask = (
(sequence_mask(dummy_length).float() - 1.0) * 100.0).unsqueeze(2)
output = layer(dummy_input + mask, dummy_target, dummy_length)
assert abs(output.item() - 1.0) < 1e-5, "1.0 vs {}".format(output.item())
dummy_input = T.rand(4, 8, 128).float()
dummy_target = dummy_input.detach()
dummy_length = (T.arange(5, 9)).long()
mask = (
(sequence_mask(dummy_length).float() - 1.0) * 100.0).unsqueeze(2)
output = layer(dummy_input + mask, dummy_target, dummy_length)
assert output.item() == 0, "0 vs {}".format(output.item())

2
tests/test_tacotron2_model.py
View File

@ -38,7 +38,7 @@ class TacotronTrainTest(unittest.TestCase):
stop_targets.size(1) // c.r, -1)
stop_targets = (stop_targets.sum(2) > 0.0).unsqueeze(2).float().squeeze()
criterion = MSELossMasked().to(device)
criterion = MSELossMasked(seq_len_norm=False).to(device)
criterion_st = nn.BCEWithLogitsLoss().to(device)
model = Tacotron2(num_chars=24, r=c.r, num_speakers=5).to(device)
model.train()

2
tests/test_tacotron_model.py
View File

@ -106,7 +106,7 @@ class TacotronGSTTrainTest(unittest.TestCase):
stop_targets = (stop_targets.sum(2) >
0.0).unsqueeze(2).float().squeeze()
criterion = L1LossMasked().to(device)
criterion = L1LossMasked(seq_len_norm=False).to(device)
criterion_st = nn.BCEWithLogitsLoss().to(device)
model = Tacotron(
num_chars=32,