def test_compatible_with_espnet1():
layer = LogSpectrogram(n_fft=16, hop_length=4)
x = torch.randn(1, 100)
y, _ = layer(x, torch.LongTensor([100]))
y = y.numpy()[0]
y2 = np.log10(spectrogram(x[0].numpy(), n_fft=16, n_shift=4))
np.testing.assert_allclose(y, y2, rtol=0, atol=1e-4)
def test_forward():
layer = LogSpectrogram(n_fft=4, hop_length=1)
x = torch.randn(2, 4, 9)
y, _ = layer(x, torch.LongTensor([4, 3]))
assert y.shape == (2, 5, 9, 3)
def test_output_size():
layer = LogSpectrogram(n_fft=4, hop_length=1)
print(layer.output_size())
def test_get_parameters():
layer = LogSpectrogram(n_fft=4, hop_length=1)
print(layer.get_parameters())
def test_backward_not_leaf_in():
layer = LogSpectrogram(n_fft=4, hop_length=1)
x = torch.randn(2, 4, 9, requires_grad=True)
x = x + 2
y, _ = layer(x, torch.LongTensor([4, 3]))
y.sum().backward()
def test_forward():
layer = LogSpectrogram(n_fft=2)
x = torch.randn(2, 4, 9)
y, _ = layer(x, torch.LongTensor([4, 3]))
assert y.shape == (2, 1, 9, 2)
def test_get_parameters():
layer = LogSpectrogram(n_fft=2)
print(layer.get_parameters())
def test_output_size():
layer = LogSpectrogram(n_fft=2)
print(layer.output_size())
def test_backward_leaf_in():
layer = LogSpectrogram(n_fft=2)
x = torch.randn(2, 4, 9, requires_grad=True)
y, _ = layer(x, torch.LongTensor([4, 3]))
y.sum().backward()