Replace pytest's paremeterization with parameterized

[mthrok]

If you are interested working on this, please leave a comment here (so that other people will not do a duplicated work), then create a PR and tag @mthrok.

---

For historical and technical reasons, torchaudio's test suite cannot use pytest as a module. (we still use it as a runner in CircleCI, which is fine.)

There are couple of pytest utilities used in torchaudio's test code.

audio/test/torchaudio_unittest/functional/functional_cpu_test.py

Lines 115 to 163 in 6b07bcf

	@pytest.mark.parametrize('complex_tensor', [
	torch.randn(1, 2, 1025, 400, 2),
	torch.randn(1025, 400, 2)
	])
	@pytest.mark.parametrize('power', [1, 2, 0.7])
	def test_complex_norm(complex_tensor, power):
	expected_norm_tensor = complex_tensor.pow(2).sum(-1).pow(power / 2)
	norm_tensor = F.complex_norm(complex_tensor, power)
	torch.testing.assert_allclose(norm_tensor, expected_norm_tensor, atol=1e-5, rtol=1e-5)
	@pytest.mark.parametrize('specgram', [
	torch.randn(2, 1025, 400),
	torch.randn(1, 201, 100)
	])
	@pytest.mark.parametrize('mask_param', [100])
	@pytest.mark.parametrize('mask_value', [0., 30.])
	@pytest.mark.parametrize('axis', [1, 2])
	def test_mask_along_axis(specgram, mask_param, mask_value, axis):
	mask_specgram = F.mask_along_axis(specgram, mask_param, mask_value, axis)
	other_axis = 1 if axis == 2 else 2
	masked_columns = (mask_specgram == mask_value).sum(other_axis)
	num_masked_columns = (masked_columns == mask_specgram.size(other_axis)).sum()
	num_masked_columns //= mask_specgram.size(0)
	assert mask_specgram.size() == specgram.size()
	assert num_masked_columns < mask_param
	@pytest.mark.parametrize('mask_param', [100])
	@pytest.mark.parametrize('mask_value', [0., 30.])
	@pytest.mark.parametrize('axis', [2, 3])
	def test_mask_along_axis_iid(mask_param, mask_value, axis):
	torch.random.manual_seed(42)
	specgrams = torch.randn(4, 2, 1025, 400)
	mask_specgrams = F.mask_along_axis_iid(specgrams, mask_param, mask_value, axis)
	other_axis = 2 if axis == 3 else 3
	masked_columns = (mask_specgrams == mask_value).sum(other_axis)
	num_masked_columns = (masked_columns == mask_specgrams.size(other_axis)).sum(-1)
	assert mask_specgrams.size() == specgrams.size()
	assert (num_masked_columns < mask_param).sum() == num_masked_columns.numel()

We would like to replace pytest.mark.parametrize with parameterized module like the following

audio/test/torchaudio_unittest/sox_compatibility_test.py

Lines 305 to 310 in 3250d3d

	@parameterized.expand([
	('q', 'quarter_sine'),
	('h', 'half_sine'),
	('t', 'linear'),
	])
	def test_fade(self, fade_shape_sox, fade_shape):

Steps

1. Convert the test function into test class using TorchaudioTestCase
2. In test method, use parameterized.expand
3. Additionally, when generating a tensor with random values, do it in the method and set the seed value first so that the test is more deterministic.

[krishnakalyan3]

@mthrok I would be happy to work on this.

[mthrok]

@krishnakalyan3 sure. thanks!