imdct#

diffsptk.IMDCT#: alias of InverseModifiedDiscreteCosineTransform

class diffsptk.InverseModifiedDiscreteCosineTransform(frame_length: int, window: str = 'sine', learnable: bool | list[str] = False)[source]#

This is the opposite module to ModifiedDiscreteCosineTransform().

Parameters:

frame_lengthint >= 2: The frame length, \(L\).
window[‘sine’, ‘vorbis’, ‘kbd’, ‘rectangular’]: The window type.
learnablebool or list[str]: Indicates whether the parameters are learnable. If a boolean, it specifies whether all parameters are learnable. If a list, it contains the keys of the learnable parameters, which can only be “basis” and “window”.

forward(y: Tensor, out_length: int | None = None) → Tensor[source]#

Compute inverse modified discrete cosine transform.

Parameters:

yTensor [shape=(…, 2T/L, L/2)]: The spectrum.
out_lengthint or None: The length of the output waveform.

Returns:

outTensor [shape=(…, T)]: The reconstructed waveform.

Examples

>>> x = diffsptk.ramp(3)
>>> x
tensor([0., 1., 2., 3.])
>>> mdct_params = {"frame_length": 4, "window": "vorbis"}
>>> mdct = diffsptk.MDCT(**mdct_params)
>>> imdct = diffsptk.IMDCT(**mdct_params)
>>> y = imdct(mdct(x))
>>> y
tensor([1.0431e-07, 1.0000e+00, 2.0000e+00, 3.0000e+00])

diffsptk.functional.imdct(y: Tensor, out_length: int | None = None, frame_length: int = 400, window: str = 'sine') → Tensor[source]#

Compute inverse modified discrete cosine transform.

Parameters:

yTensor [shape=(…, 2T/L, L/2)]: The spectrum.
out_lengthint or None: The length of the output waveform.
frame_lengthint >= 2: The frame length, \(L\).
window[‘sine’, ‘vorbis’, ‘kbd’, ‘rectangular’]: The window type.

Returns:

outTensor [shape=(…, T)]: The reconstructed waveform.