Undescribed Horrific Abuse, One Victim & Survivor of Many gmkarl at gmail.com
Thu Nov 10 13:20:55 PST 2022

1606 i got irft to match numpy [i partly cheated and trimmed the
number of frequencies to match numpy's, skipping more thoroughly
understanding the fourier transform]

now to scale its frequencies to match the waveform's frequencies

freq2period : frequencies are fractions of the total number of
samples, so a frequency of 0.5 is a period of N.

max_period is already set using the idx_wav2rec function, so it should
represent what's right

okay ... um ... so the question is, what is the waveform frequency, in
the simulated recording?
it has max_period, so the frequency would be max_period / N; its first
fourier component would be at 0.5 * max_period / N, right?
no, not quite: 0.5 is the nyquist frequency ... not sure ihave these
frequencies right.
0 is infinite period
1/N is a period of N.
freq = 1/period .
so the first non-DC fourier component would be 1/max_period, I think
... [and the phase of this could be used to align with it]

looks like that's what it already is
maybe it'll work better at this time.

it doesn't reconstruct the subwave yet, even when i set it to exactly
half period.
it looks to me like the reason it doesn't reconstruct it would be
simliar to the reason that the fft/ifft doesn't invert for me, if i
add extra frequencies.

More information about the cypherpunks mailing list