比较与torchaudio.transforms.AmplitudeToDB的差异

class torchaudio.transforms.AmplitudeToDB(stype: str = 'power', top_db: Optional[float] = None)

class mindspore.dataset.audio.AmplitudeToDB(stype=ScaleType.POWER, ref_value=1.0, amin=1e-10, top_db=80.0)

import numpy as np

fake_input = np.array([[[[-0.2197528, 0.3821656]]],
                       [[[0.57418776, 0.46741104]]],
                       [[[0.76986176, -0.5793846]]]]).astype(np.float32)

# PyTorch
import torch
import torchaudio.transforms as T

transformer = T.AmplitudeToDB(stype="power", top_db=80.)
torch_result = transformer(torch.from_numpy(fake_input))
print(torch_result)
# Out: tensor([[[[-84.1775,  -4.1775]]],
#        [[[ -2.4095,  -3.3030]]],
#        [[[ -1.1359, -81.1359]]]])

# MindSpore
import mindspore.dataset.audio as audio

transformer = audio.AmplitudeToDB(stype=audio.ScaleType.POWER, top_db=80., ref_value=1.0, amin=1e-10)
ms_result = transformer(fake_input)
print(ms_result)
# Out: [[[[-84.17748    -4.177484 ]]]
#      [[[ -2.4094608  -3.3030105]]]
#      [[[ -1.1358725 -81.13587  ]]]]