def test_tasnet_forward_backward(
N,
L,
B,
H,
P,
X,
R,
num_spk,
norm_type,
causal,
mask_nonlinear,
):
model = TasNet(
N=N,
L=L,
B=B,
H=H,
P=P,
X=X,
R=R,
num_spk=num_spk,
norm_type=norm_type,
causal=causal,
mask_nonlinear=mask_nonlinear,
)
est_speech, *_ = model(
torch.randn(2, 100, requires_grad=True), ilens=torch.LongTensor([100, 80])
)
loss = sum([est.mean() for est in est_speech])
loss.backward()
def test_tasnet_cosistency(
N,
L,
B,
H,
P,
X,
R,
num_spk,
norm_type,
causal,
mask_nonlinear,
):
model = TasNet(
N=N,
L=L,
B=B,
H=H,
P=P,
X=X,
R=R,
num_spk=num_spk,
norm_type=norm_type,
causal=causal,
mask_nonlinear=mask_nonlinear,
)
model.eval()
random_input_numpy = np.random.randn(2, 100) # np.float64
random_input_torch = (
torch.from_numpy(random_input_numpy - 1.0).float() + 1.0
) # torch.float32
random_input_numpy = torch.from_numpy(
random_input_numpy.astype("float32")
) # np.float64-->np.float32-->torch.float32
est_speech_numpy, *_ = model(random_input_numpy, ilens=torch.LongTensor([100, 80]))
est_speech_torch, *_ = model(random_input_torch, ilens=torch.LongTensor([100, 80]))
assert (est_speech_torch[0] - est_speech_numpy[0]).abs().mean() < 1e-5
assert (
np.abs((est_speech_torch[-1] - est_speech_numpy[-1]).detach().numpy()).mean()
< 1e-5
)
def test_tasnet_output(
N,
L,
B,
H,
P,
X,
R,
num_spk,
norm_type,
causal,
mask_nonlinear,
):
inputs = torch.randn(2, 160)
ilens = torch.LongTensor([160, 120])
for num_spk in range(1, 3):
model = TasNet(
N=N,
L=L,
B=B,
H=H,
P=P,
X=X,
R=R,
num_spk=num_spk,
norm_type=norm_type,
causal=causal,
mask_nonlinear=mask_nonlinear,
)
specs, _, masks = model(inputs, ilens)
assert isinstance(specs, list)
assert isinstance(masks, dict)
for n in range(num_spk):
assert "spk{}".format(n + 1) in masks
assert specs[n].shape == masks["spk{}".format(n + 1)].shape
assert specs[n].shape == (2, 160)
def test_tasnet_invalid_mask_nonlinear():
with pytest.raises(ValueError):
TasNet(5, 20, 5, 10, 3, 8, 4, 2, mask_nonlinear="fff")
def test_tasnet_invalid_norm_type():
with pytest.raises(ValueError):
TasNet(5, 20, 5, 10, 3, 8, 4, 2, norm_type="fff")
def test_tasnet_invalid_mask_nonlinear():
with pytest.raises(ValueError):
TasNet(2, 40, 2, 2, 3, 3, 2, 2, mask_nonlinear="fff")
def test_tasnet_invalid_norm_type():
with pytest.raises(ValueError):
TasNet(2, 40, 2, 2, 3, 3, 2, 2, norm_type="fff")