def forward(self, waveform: Tensor) -> Tensor:
r"""
Args:
waveform (Tensor): Tensor of audio of dimension (..., time).
Returns:
Tensor: Tensor of audio of dimension (..., time).
"""
if self.gain_type == "amplitude":
waveform = waveform * self.gain
if self.gain_type == "db":
waveform = F.gain(waveform, self.gain)
if self.gain_type == "power":
waveform = F.gain(waveform, 10 * math.log10(self.gain))
return torch.clamp(waveform, -1, 1)
def test_gain(self):
waveform_gain = F.gain(self.waveform_train, 3)
self.assertTrue(waveform_gain.abs().max().item(), 1.)
E = torchaudio.sox_effects.SoxEffectsChain()
E.set_input_file(self.test_filepath)
E.append_effect_to_chain("gain", [3])
sox_gain_waveform = E.sox_build_flow_effects()[0]
self.assertTrue(torch.allclose(waveform_gain, sox_gain_waveform, atol=1e-04))
def test_gain(self):
test_filepath = common_utils.get_asset_path('steam-train-whistle-daniel_simon.wav')
waveform, _ = torchaudio.load(test_filepath)
waveform_gain = F.gain(waveform, 3)
self.assertTrue(waveform_gain.abs().max().item(), 1.)
E = torchaudio.sox_effects.SoxEffectsChain()
E.set_input_file(test_filepath)
E.append_effect_to_chain("gain", [3])
sox_gain_waveform = E.sox_build_flow_effects()[0]
self.assertEqual(waveform_gain, sox_gain_waveform, atol=1e-04, rtol=1e-5)
def test_gain(self):
test_filepath = os.path.join(self.test_dirpath, "assets",
"steam-train-whistle-daniel_simon.wav")
waveform, _ = torchaudio.load(test_filepath)
waveform_gain = F.gain(waveform, 3)
self.assertTrue(waveform_gain.abs().max().item(), 1.)
E = torchaudio.sox_effects.SoxEffectsChain()
E.set_input_file(test_filepath)
E.append_effect_to_chain("gain", [3])
sox_gain_waveform = E.sox_build_flow_effects()[0]
torch.testing.assert_allclose(waveform_gain,
sox_gain_waveform,
atol=1e-04,
rtol=1e-5)
def func(tensor):
gainDB = 2.0
return F.gain(tensor, gainDB)
def test_gain(self):
path = get_asset_path('steam-train-whistle-daniel_simon.wav')
data, _ = load_wav(path)
result = F.gain(data, 3)
self.assert_sox_effect(result, path, ['gain', 3])
#Functional spectrogram deltas
specgram = torchaudio.transforms.Spectrogram()(waveform)
computed = torchaudio.functional.compute_deltas(specgram.contiguous(), win_length=3)
print(f"Shape of computed deltas: {computed.shape}")
plt.figure()
#The image is blank?? Probably matplotlib issue
plt.imshow(computed.log2()[0,:,:].detach().numpy(), cmap='gray')
# %%
#Applying gain and dithering
gain_waveform = F.gain(waveform, gain_db=5.0)
print(f'''
Min of gain_waveform: {gain_waveform.min()}
Max of gain_waveform: {gain_waveform.max()}
Mean of gain_waveform: {gain_waveform.mean()}
''')
dither_waveform = F.dither(waveform)
print(f'''
Min of dither_waveform: {dither_waveform.min()}
Max of dither_waveform: {dither_waveform.max()}
Mean of dither_waveform: {dither_waveform.mean()}
''')
# %%
#Applying filters