Merge branch 'installation' into codec_use

Author: Sam Anklesaria

.github/scripts/unittest-linux/install.sh

@@ -85,6 +85,9 @@ export BUILD_CPP_TEST=1
 python setup.py install
 
 # 3. Install Test tools
+conda install -y "ffmpeg<5"
+python -c "import torch; import torchaudio; import torchcodec; print(torch.__version__, torchaudio.__version__, torchcodec.__version__)"
+
 printf "* Installing test tools\n"
 NUMBA_DEV_CHANNEL=""
 if [[ "$(python --version)" = *3.9* || "$(python --version)" = *3.10* ]]; then
@@ -94,7 +97,7 @@ if [[ "$(python --version)" = *3.9* || "$(python --version)" = *3.10* ]]; then
 fi
 (
     set -x
-    conda install -y -c conda-forge ${NUMBA_DEV_CHANNEL} sox libvorbis parameterized 'requests>=2.20' 'ffmpeg>=6,<7'
+    conda install -y -c conda-forge ${NUMBA_DEV_CHANNEL} sox libvorbis parameterized 'requests>=2.20' 
     pip install kaldi-io SoundFile librosa coverage pytest pytest-cov scipy expecttest unidecode inflect Pillow sentencepiece pytorch-lightning 'protobuf<4.21.0' demucs tinytag pyroomacoustics flashlight-text git+https://github.com/kpu/kenlm
 
     # TODO: might be better to fix the single call to `pip install` above

src/torchaudio/functional/filtering.py

@@ -6,7 +6,6 @@
 from torch import Tensor
 
 from torchaudio._extension import _IS_TORCHAUDIO_EXT_AVAILABLE
-from torchaudio._internal.module_utils import dropping_support
 
 
 def _dB2Linear(x: float) -> float:
@@ -325,7 +324,7 @@ def biquad(waveform: Tensor, b0: float, b1: float, b2: float, a0: float, a1: flo
     a1 = torch.as_tensor(a1, dtype=dtype, device=device).view(1)
     a2 = torch.as_tensor(a2, dtype=dtype, device=device).view(1)
 
-    output_waveform = _lfilter_deprecated(
+    output_waveform = lfilter(
         waveform,
         torch.cat([a0, a1, a2]),
         torch.cat([b0, b1, b2]),
@@ -699,8 +698,8 @@ def filtfilt(
         Tensor: Waveform with dimension of either `(..., num_filters, time)` if ``a_coeffs`` and ``b_coeffs``
         are 2D Tensors, or `(..., time)` otherwise.
     """
-    forward_filtered = _lfilter_deprecated(waveform, a_coeffs, b_coeffs, clamp=False, batching=True)
-    backward_filtered = _lfilter_deprecated(
+    forward_filtered = lfilter(waveform, a_coeffs, b_coeffs, clamp=False, batching=True)
+    backward_filtered = lfilter(
         forward_filtered.flip(-1),
         a_coeffs,
         b_coeffs,
@@ -998,7 +997,7 @@ def _lfilter_core(
     _lfilter = _lfilter_core
 
 
-def _lfilter_deprecated(waveform: Tensor, a_coeffs: Tensor, b_coeffs: Tensor, clamp: bool = True, batching: bool = True) -> Tensor:
+def lfilter(waveform: Tensor, a_coeffs: Tensor, b_coeffs: Tensor, clamp: bool = True, batching: bool = True) -> Tensor:
     r"""Perform an IIR filter by evaluating difference equation, using differentiable implementation
     developed separately by *Yu et al.* :cite:`ismir_YuF23` and *Forgione et al.* :cite:`forgione2021dynonet`.
     The gradients of ``a_coeffs`` are computed based on a faster algorithm from :cite:`ycy2024diffapf`.
@@ -1067,7 +1066,6 @@ def _lfilter_deprecated(waveform: Tensor, a_coeffs: Tensor, b_coeffs: Tensor, cl
 
     return output
 
-lfilter = dropping_support(_lfilter_deprecated)
 
 def lowpass_biquad(waveform: Tensor, sample_rate: int, cutoff_freq: float, Q: float = 0.707) -> Tensor:
     r"""Design biquad lowpass filter and perform filtering.  Similar to SoX implementation.
@@ -1117,7 +1115,7 @@ def _overdrive_core_loop_generic(
     _overdrive_core_loop_cpu = _overdrive_core_loop_generic
 
 
-def _overdrive_deprecated(waveform: Tensor, gain: float = 20, colour: float = 20) -> Tensor:
+def overdrive(waveform: Tensor, gain: float = 20, colour: float = 20) -> Tensor:
     r"""Apply a overdrive effect to the audio. Similar to SoX implementation.
 
     .. devices:: CPU CUDA
@@ -1172,7 +1170,6 @@ def _overdrive_deprecated(waveform: Tensor, gain: float = 20, colour: float = 20
 
     return output_waveform.clamp(min=-1, max=1).view(actual_shape)
 
-overdrive = dropping_support(_overdrive_deprecated)
 
 def phaser(
     waveform: Tensor,

src/torchaudio/functional/functional.py

@@ -2476,7 +2476,7 @@ def preemphasis(waveform, coeff: float = 0.97) -> torch.Tensor:
     return waveform
 
 
-def _deemphasis(waveform, coeff: float = 0.97) -> torch.Tensor:
+def deemphasis(waveform, coeff: float = 0.97) -> torch.Tensor:
     r"""De-emphasizes a waveform along its last dimension.
     Inverse of :meth:`preemphasis`. Concretely, for each signal
     :math:`x` in ``waveform``, computes output :math:`y` as
@@ -2498,9 +2498,8 @@ def _deemphasis(waveform, coeff: float = 0.97) -> torch.Tensor:
     """
     a_coeffs = torch.tensor([1.0, -coeff], dtype=waveform.dtype, device=waveform.device)
     b_coeffs = torch.tensor([1.0, 0.0], dtype=waveform.dtype, device=waveform.device)
-    return torchaudio.functional.filtering._lfilter_deprecated(waveform, a_coeffs=a_coeffs, b_coeffs=b_coeffs)
+    return torchaudio.functional.filtering.lfilter(waveform, a_coeffs=a_coeffs, b_coeffs=b_coeffs)
 
-deemphasis = dropping_support(_deemphasis)
 
 def frechet_distance(mu_x, sigma_x, mu_y, sigma_y):
     r"""Computes the Fréchet distance between two multivariate normal distributions :cite:`dowson1982frechet`.

src/torchaudio/transforms/_transforms.py

@@ -2158,4 +2158,4 @@ def forward(self, waveform: torch.Tensor) -> torch.Tensor:
         Returns:
             torch.Tensor: De-emphasized waveform, with shape `(..., N)`.
         """
-        return F.functional._deemphasis(waveform, coeff=self.coeff)
+        return F.functional.deemphasis(waveform, coeff=self.coeff)

test/torchaudio_unittest/functional/torchscript_consistency_impl.py

@@ -285,12 +285,7 @@ def test_lfilter(self):
             device=waveform.device,
             dtype=waveform.dtype,
         )
-        # This is hack for those functions which are deprecated with decorators
-        # like @deprecated or @dropping_support. Adding the decorators breaks
-        # TorchScript. So here we use the private function which make the tests
-        # pass, but that's a lie: the public (deprecated) function doesn't
-        # support torchscript anymore
-        self._assert_consistency(F.filtering._lfilter_deprecated, (waveform, a_coeffs, b_coeffs, True, True))
+        self._assert_consistency(F.lfilter, (waveform, a_coeffs, b_coeffs, True, True))
 
     def test_filtfilt(self):
         waveform = common_utils.get_whitenoise(sample_rate=8000)
@@ -490,7 +485,7 @@ def test_perf_biquad_filtering(self):
         def func(tensor):
             a = torch.tensor([0.7, 0.2, 0.6], device=tensor.device, dtype=tensor.dtype)
             b = torch.tensor([0.4, 0.2, 0.9], device=tensor.device, dtype=tensor.dtype)
-            return F.filtering._lfilter_deprecated(tensor, a, b)
+            return F.lfilter(tensor, a, b)
 
         self._assert_consistency(func, (waveform,))
 
@@ -535,12 +530,7 @@ def test_overdrive(self):
         def func(tensor):
             gain = 30.0
             colour = 50.0
-            # This is hack for those functions which are deprecated with decorators
-            # like @deprecated or @dropping_support. Adding the decorators breaks
-            # TorchScript. So here we use the private function which make the tests
-            # pass, but that's a lie: the public (deprecated) function doesn't
-            # support torchscript anymore
-            return F.filtering._overdrive_deprecated(tensor, gain, colour)
+            return F.overdrive(tensor, gain, colour)
 
         self._assert_consistency(func, (waveform,))