ifreqt2#

class diffsptk.SecondOrderAllPassInverseFrequencyTransform(in_order, out_order, alpha=0, theta=0, n_fft=512)[source]#

Second-order all-pass inverse frequecy transform module.

Parameters:

in_orderint >= 0 [scalar]: Order of input sequence, \(M_1\).
out_orderint >= 0 [scalar]: Order of output sequence, \(M_2\).
alphafloat [-1 < alpha < 1]: Frequency warping factor, \(\alpha\).
thetafloat [0 <= theta <= 1]: Emphasis frequency, \(\theta\).
n_fftint >> math:M_1 [scalar]: Number of FFT bins. Accurate conversion requires the large value.

forward(c1)[source]#

Perform second-order all-pass inverse frequency transform.

Parameters:

c1Tensor [shape=(…, M1+1)]: Warped sequence.

Returns:

c2Tensor [shape=(…, M2+1)]: Output sequence.

Examples

>>> c1 = diffsptk.nrand(3)
>>> c1
tensor([ 0.0304,  0.5849, -0.8668, -0.7278])
>>> freqt2 = diffsptk.SecondOrderAllPassFrequencyTransform(3, 4, .1, .3)
>>> c2 = freqt2(c1)
>>> c2
tensor([ 0.0682,  0.4790, -1.0168, -0.6026,  0.1094])
>>> ifreqt2 = diffsptk.SecondOrderAllPassInverseFrequencyTransform(4, 3, .1, .3)
>>> c3 = ifreqt2(c2)
>>> c3
tensor([ 0.0682,  0.4790, -1.0168, -0.6026,  0.1094])