[BC-Breaking] Remove F.complex_norm and T.ComplexNorm

docs/source/functional.rst

@@ -71,17 +71,6 @@ resample
 
 .. autofunction:: resample
 
-:hidden:`Complex Utility`
-~~~~~~~~~~~~~~~~~~~~~~~~~
-
-Utilities for pseudo complex tensor. This is not for the native complex dtype, such as `cfloat64`, but for tensors with real-value type and have extra dimension at the end for real and imaginary parts.
-
-
-complex_norm
-------------
-
-.. autofunction:: complex_norm
-
 
 :hidden:`Filtering`
 ~~~~~~~~~~~~~~~~~~~

docs/source/transforms.rst

@@ -88,16 +88,6 @@ Transforms are common audio transforms. They can be chained together using :clas
 
   .. automethod:: forward
 
-:hidden:`Complex Utility`
-~~~~~~~~~~~~~~~~~~~~~~~~~
-
-:hidden:`ComplexNorm`
----------------------
-
-.. autoclass:: ComplexNorm
-
-  .. automethod:: forward
-
 :hidden:`Feature Extractions`
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

test/torchaudio_unittest/functional/functional_impl.py

@@ -319,16 +319,6 @@ def test_amplitude_to_DB_top_db_clamp(self, shape):
             f"No values were close to the limit. Did it over-clamp?\n{decibels}"
         )
 
-    @parameterized.expand(
-        list(itertools.product([(1, 2, 1025, 400, 2), (1025, 400, 2)], [1, 2, 0.7]))
-    )
-    def test_complex_norm(self, shape, power):
-        torch.random.manual_seed(42)
-        complex_tensor = torch.randn(*shape, dtype=self.dtype, device=self.device)
-        expected_norm_tensor = complex_tensor.pow(2).sum(-1).pow(power / 2)
-        norm_tensor = F.complex_norm(complex_tensor, power)
-        self.assertEqual(norm_tensor, expected_norm_tensor, atol=1e-5, rtol=1e-5)
-
     @parameterized.expand(
         list(itertools.product([(2, 1025, 400), (1, 201, 100)], [100], [0., 30.], [1, 2]))
     )

test/torchaudio_unittest/functional/torchscript_consistency_impl.py

@@ -223,14 +223,6 @@ def func(tensor):
         tensor = torch.rand((1, 10))
         self._assert_consistency(func, tensor)
 
-    def test_complex_norm(self):
-        def func(tensor):
-            power = 2.
-            return F.complex_norm(tensor, power)
-
-        tensor = torch.randn(1, 2, 1025, 400, 2)
-        self._assert_consistency(func, tensor)
-
     def test_mask_along_axis(self):
         def func(tensor):
             mask_param = 100

test/torchaudio_unittest/transforms/torchscript_consistency_impl.py

@@ -86,10 +86,6 @@ def test_Resample(self):
         tensor = common_utils.get_whitenoise(sample_rate=sr1)
         self._assert_consistency(T.Resample(sr1, sr2), tensor)
 
-    def test_ComplexNorm(self):
-        tensor = torch.rand((1, 2, 201, 2))
-        self._assert_consistency(T.ComplexNorm(), tensor)
-
     def test_MuLawEncoding(self):
         tensor = common_utils.get_whitenoise()
         self._assert_consistency(T.MuLawEncoding(), tensor)

torchaudio/functional/__init__.py

@@ -1,6 +1,5 @@
 from .functional import (
     amplitude_to_DB,
-    complex_norm,
     compute_deltas,
     compute_kaldi_pitch,
     create_dct,
@@ -52,7 +51,6 @@
 
 __all__ = [
     'amplitude_to_DB',
-    'complex_norm',
     'compute_deltas',
     'compute_kaldi_pitch',
     'create_dct',

torchaudio/functional/functional.py

@@ -28,7 +28,6 @@
     "DB_to_amplitude",
     "mu_law_encoding",
     "mu_law_decoding",
-    "complex_norm",
     "phase_vocoder",
     'mask_along_axis',
     'mask_along_axis_iid',
@@ -722,32 +721,6 @@ def mu_law_decoding(
     return x
 
 
-@_mod_utils.deprecated(
-    "Please convert the input Tensor to complex type with `torch.view_as_complex` then "
-    "use `torch.abs`. "
-    "Please refer to https://github.com/pytorch/audio/issues/1337 "
-    "for more details about torchaudio's plan to migrate to native complex type.",
-    version="0.11",
-)
-def complex_norm(
-        complex_tensor: Tensor,
-        power: float = 1.0
-) -> Tensor:
-    r"""Compute the norm of complex tensor input.
-
-    Args:
-        complex_tensor (Tensor): Tensor shape of `(..., complex=2)`
-        power (float, optional): Power of the norm. (Default: `1.0`).
-
-    Returns:
-        Tensor: Power of the normed input tensor. Shape of `(..., )`
-    """
-
-    # Replace by torch.norm once issue is fixed
-    # https://github.com/pytorch/pytorch/issues/34279
-    return complex_tensor.pow(2.).sum(-1).pow(0.5 * power)
-
-
 def phase_vocoder(
         complex_specgrams: Tensor,
         rate: float,

torchaudio/transforms.py

@@ -26,7 +26,6 @@
     'MuLawEncoding',
     'MuLawDecoding',
     'Resample',
-    'ComplexNorm',
     'TimeStretch',
     'Fade',
     'FrequencyMasking',
@@ -900,42 +899,6 @@ def forward(self, waveform: Tensor) -> Tensor:
             self.kernel, self.width)
 
 
-class ComplexNorm(torch.nn.Module):
-    r"""Compute the norm of complex tensor input.
-
-    Args:
-        power (float, optional): Power of the norm. (Default: to ``1.0``)
-
-    Example
-        >>> complex_tensor = ... #  Tensor shape of (…, complex=2)
-        >>> transform = transforms.ComplexNorm(power=2)
-        >>> complex_norm = transform(complex_tensor)
-    """
-    __constants__ = ['power']
-
-    def __init__(self, power: float = 1.0) -> None:
-        warnings.warn(
-            'torchaudio.transforms.ComplexNorm has been deprecated '
-            'and will be removed from future release.'
-            'Please convert the input Tensor to complex type with `torch.view_as_complex` then '
-            'use `torch.abs` and `torch.angle`. '
-            'Please refer to https://github.com/pytorch/audio/issues/1337 '
-            "for more details about torchaudio's plan to migrate to native complex type."
-        )
-        super(ComplexNorm, self).__init__()
-        self.power = power
-
-    def forward(self, complex_tensor: Tensor) -> Tensor:
-        r"""
-        Args:
-            complex_tensor (Tensor): Tensor shape of `(..., complex=2)`.
-
-        Returns:
-            Tensor: norm of the input tensor, shape of `(..., )`.
-        """
-        return F.complex_norm(complex_tensor, self.power)
-
-
 class ComputeDeltas(torch.nn.Module):
     r"""Compute delta coefficients of a tensor, usually a spectrogram.