如何将TensorFlow代码的这一小部分翻译成pyTorch？

xxhby3vn 于 2022-11-23 发布在其他

关注(0)|答案(1)|浏览(137)

def transforms(x):
        # stft returns spectogram for each sample and each eeg
        # input X contains 3 signals, apply stft for each 
        # and get array with shape [samples, num_of_eeg, time_stamps, freq]
        # change dims and return [samples, time_stamps, freq, num_of_eeg]
        spectrograms = tf.signal.stft(x, frame_length=32, frame_step=4, fft_length=64)
        spectrograms = tf.abs(spectrograms)
        return tf.einsum("...ijk->...jki", spectrograms)

pytorch

来源：https://stackoverflow.com/questions/74523337/how-to-translate-this-small-part-of-tensorflow-code-into-pytorch

1条答案

按热度按时间

piv4azn71#

您可以找到STFT pytorch实现here的文档。剩下的是快进。它应该是：

def transforms(x: torch.Tensor) -> torch.Tensor:
    """Return Fourrier spectrogram."""
    spectrograms = torch.stft(x, win_length=32, n_fft=4, hop_length=64)
    spectrograms = torch.abs(spectrograms)
    return torch.einsum("...ijk->...jki", spectrograms)

赞(0）回复(0）举报 2022-11-23

我来回答

如何将TensorFlow代码的这一小部分翻译成pyTorch？

1条答案

相关问题

热门标签

最新问答