signals#

diffsptk.impulse(order: int, **kwargs) → Tensor[source]#

Generate impulse sequence.

See impulse for details.

Parameters:

orderint >= 0: The order of the sequence, \(M\).
**kwargsadditional keyword arguments: See torch.eye.

Returns:

outTensor [shape=(M+1,)]: The impulse sequence.

Examples

>>> x = diffsptk.impulse(4)
>>> x
tensor([1., 0., 0., 0., 0.])

diffsptk.step(order: int, value: float = 1, **kwargs) → Tensor[source]#

Generate step sequence.

See step for details.

Parameters:

orderint >= 0: The order of the sequence, \(M\).
valuefloat: Step value.
**kwargsadditional keyword arguments: See torch.full.

Returns:

outTensor [shape=(M+1,)]: The step sequence.

Examples

>>> x = diffsptk.step(4, 2)
>>> x
tensor([2., 2., 2., 2., 2.])

diffsptk.ramp(arg: float, end: float | None = None, step: float = 1, eps: float = 1e-08, **kwargs) → Tensor[source]#

Generate ramp sequence.

See ramp for details.

Parameters:

argfloat: If end is None, this is the end value otherwise start value.
endfloat or None: The end value.
stepfloat != 0: The slope.
epsfloat: A correction value.
**kwargsadditional keyword arguments: See torch.arange.

Returns:

outTensor [shape=(?,)]: The ramp sequence.

Examples

>>> x = diffsptk.ramp(4)
>>> x
tensor([0., 1., 2., 3., 4.])

diffsptk.sin(order: int, period: float | None = None, magnitude: float = 1, **kwargs) → Tensor[source]#

Generate sinusoidal sequence.

See sin for details.

Parameters:

orderint >= 0: The order of the sequence, \(M\).
periodfloat > 0: The period.
magnitudefloat: The magnitude.
**kwargsadditional keyword arguments: See torch.arange.

Returns:

outTensor [shape=(M+1,)]: The sinusoidal sequence.

Examples

>>> x = diffsptk.sin(4)
>>> x
tensor([ 0.0000,  0.9511,  0.5878, -0.5878, -0.9511])

diffsptk.train(order: int, frame_period: float, norm: str | int = 'power', **kwargs) → Tensor[source]#

Generate pulse sequence.

See train for details.

Parameters:

orderint >= 0: The order of the sequence, \(M\).
frame_periodfloat >= 1: The frame period.
norm[‘none’, ‘power’, ‘magnitude’]: The normalization type.
**kwargsadditional keyword arguments: See torch.zeros.

Returns:

outTensor [shape=(M+1,)]: The pulse sequence.

Examples

>>> x = diffsptk.train(5, 2.3)
>>> x
tensor([1.5166, 0.0000, 0.0000, 1.5166, 0.0000, 1.5166])

diffsptk.nrand(*order: int, mean: float = 0, stdv: float = 1, var: float | None = None, **kwargs) → Tensor[source]#

Generate random number sequence.

See nrand for details.

Parameters:

orderint >= 0: The order of the sequence, \(M\).
meanfloat: The mean.
stdvfloat >= 0: The standard deviation.
varfloat >= 0: The variance. This overrides stdv.
**kwargsadditional keyword arguments: See torch.randn.

Returns:

outTensor [shape=(M+1,)]: The random value sequence.

Examples

>>> x = diffsptk.nrand(4)
>>> x
tensor([-0.8603,  0.6743, -0.9178,  1.5382, -0.2574])
>>> x = diffsptk.nrand(2, 4)
>>> x
tensor([[-0.2385, -0.0778, -0.0418, -1.6217,  0.1560],
        [ 1.6646,  0.8429,  0.9357, -0.5123,  0.9571]])