Differences with torchaudio.transforms.SpectralCentroid

torchaudio.transforms.SpectralCentroid

class torchaudio.transforms.SpectralCentroid(sample_rate: int, n_fft: int = 400, win_length: Optional[int] = None,
                                             hop_length: Optional[int] = None, pad: int = 0,
                                             window_fn: Callable[[...], torch.Tensor] = <built-in method hann_window of type object>,
                                             wkwargs: Optional[dict] = None)

For more information, see torchaudio.transforms.SpectralCentroid.

mindspore.dataset.audio.SpectralCentroid

class mindspore.dataset.audio.SpectralCentroid(sample_rate, n_fft=400, win_length=None, hop_length=None,
                                               pad=0, window=WindowType.HANN)

For more information, see mindspore.dataset.audio.SpectralCentroid.

Differences

PyTorch：Compute the spectral centroid for each channel along the time axis. Customized window function and different parameter configs for window function are both supported.

MindSpore：Compute the spectral centroid for each channel along the time axis.

Categories	Subcategories	PyTorch	MindSpore	Difference
Parameter	Parameter1	sample_rate	sample_rate	-
	Parameter2	n_fft	n_fft	-
	Parameter3	win_length	win_length	-
	Parameter4	hop_length	hop_length	-
	Parameter5	pad	pad
	Parameter6	window_fn	window	MindSpore only support 5 window functions
	Parameter7	wkwargs	-	Arguments for window function, not supported by MindSpore

Code Example

import numpy as np

fake_input = np.array([[[1, 1, 2, 2, 3, 3, 4]]]).astype(np.float32)

# PyTorch
import torch
import torchaudio.transforms as T

transformer = T.SpectralCentroid(sample_rate=44100, n_fft=8, window_fn=torch.hann_window)
torch_result = transformer(torch.from_numpy(fake_input))
print(torch_result)
# Out: tensor([[[4436.1182, 3768.7986]]])

# MindSpore
import mindspore.dataset.audio as audio

transformer = audio.SpectralCentroid(sample_rate=44100, n_fft=8, window=audio.WindowType.HANN)
ms_result = transformer(fake_input)
print(ms_result)
# Out: [[[[4436.117  3768.7979]]]]