def test_throws_if_scalar_freq_and_amp_with_no_length(self):
dummy_sr = 8
dummy_freq = 3
dummy_amp = 0.7
operator = Operator(sr=dummy_sr)
with self.assertRaises(ValueError):
operator.forward(dummy_freq, dummy_amp)
def test_throws_if_tensor_freq_and_amp_are_different_lengths(self):
dummy_sr = 8
dummy_freq = torch.Tensor([0.5, 0.1, 0.3])
dummy_amp = torch.Tensor([12, 0.1, 0.2, 0.5])
operator = Operator(sr=dummy_sr)
with self.assertRaises(ValueError):
operator.forward(dummy_freq, dummy_amp)
def test_throws_if_tensor_freq_and_amp_have_more_than_batch_and_time_dims(
self):
dummy_sr = 8
dummy_freq = torch.Tensor([[[0.1]]])
dummy_amp = torch.Tensor([[[0.5]]])
operator = Operator(sr=dummy_sr)
with self.assertRaises(ValueError):
operator.forward(dummy_freq, dummy_amp)
def test_can_synthesise_sinusoid(self):
dummy_sr = 2
dummy_freq = 1
sample_length = 4
expected_output = torch.Tensor([1, -1, 1, -1])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq, length=sample_length)
torch.testing.assert_allclose(actual_output, expected_output)
def test_can_set_tensor_frequency(self):
dummy_sr = 8
dummy_freq = torch.Tensor([2, 0, 4, 2])
sample_length = 4
expected_output = torch.Tensor([1, 1, -1, 0])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq, length=sample_length)
torch.testing.assert_allclose(actual_output, expected_output)
def test_throws_if_phase_mod_has_more_than_batch_and_time_dims(self):
dummy_sr = 8
dummy_freq = 4
dummy_amp = 0.5
dummy_phase = torch.Tensor([[[0.1]]])
operator = Operator(sr=dummy_sr)
with self.assertRaises(ValueError):
operator.forward(dummy_freq,
dummy_amp,
length=4,
phase_mod=dummy_phase)
def test_can_set_tensor_amplitude(self):
dummy_sr = 8
dummy_freq = 4
dummy_amplitude = torch.Tensor([0, 0.4, 0.9, 13.0])
sample_length = 4
expected_output = torch.Tensor([0.0, -0.4, 0.9, -13.0])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq,
amplitude=dummy_amplitude,
length=sample_length)
torch.testing.assert_allclose(actual_output, expected_output)
def test_can_set_amplitude(self):
dummy_sr = 4
dummy_freq = 2
dummy_amplitude = 0.5
sample_length = 4
expected_output = torch.Tensor([0.5, -0.5, 0.5, -0.5])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq,
amplitude=dummy_amplitude,
length=sample_length)
torch.testing.assert_allclose(actual_output, expected_output)
def test_can_use_freq_with_multiple_batches_but_one_time_step(self):
dummy_sr = 4
dummy_freq = torch.Tensor([[1], [0]])
dummy_amp = 1
sample_length = 3
expected_output = torch.Tensor([[1, 0, -1], [1, 1, 1]])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq,
dummy_amp,
length=sample_length)
torch.testing.assert_allclose(actual_output, expected_output)
def test_can_modulate_phase(self):
dummy_sr = 4
dummy_freq = 1
dummy_phase_mod = torch.Tensor([math.pi, 0, -math.pi, math.pi])
sample_length = 4
expected_output = torch.Tensor([-1, 0, 1, 0])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq,
phase_mod=dummy_phase_mod,
length=sample_length)
torch.testing.assert_allclose(actual_output, expected_output)
def test_can_set_initial_phase(self):
dummy_sr = 4
dummy_freq = 1
dummy_amp = 0.8
dummy_initial_phase = 0.5 * math.pi
sample_length = 4
expected_output = torch.Tensor([0, -0.8, 0, 0.8])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq,
dummy_amp,
phase_mod=dummy_initial_phase,
length=sample_length)
torch.testing.assert_allclose(actual_output, expected_output)
def test_can_use_phase_with_multiple_batches_but_one_time_step(self):
dummy_sr = 4
dummy_freq = 1
dummy_amp = 1
dummy_phase = torch.Tensor([[0], [math.pi]])
sample_length = 3
expected_output = torch.Tensor([[1, 0, -1], [-1, 0, 1]])
operator = Operator(sr=dummy_sr)
actual_output = operator.forward(dummy_freq,
dummy_amp,
length=sample_length,
phase_mod=dummy_phase)
torch.testing.assert_allclose(actual_output, expected_output)
def test_can_create(self):
operator = Operator()
self.assertIsInstance(operator, Operator)
def test_is_nn_module(self):
operator = Operator()
self.assertIsInstance(operator, torch.nn.Module)