Re: How do YOU handle this disparity in fft?

[Macy]

just went to 
<http://www.octave.org>
to look for that package and could not find ANYTHING ?!

You describe the same function but is "...better documented ..." Well, that 
wouldn't take much. But the function is NOT the same, because the first index 
value is WRONG. Plus, my script does NOT return the Nyquist value, since 
Nyquist rate signals appear at the next bin just outside the range used. Thus, 
it is ignored with no change applied to the value.

The whole point of the exercise of fftreal.m was to produce a simple ENERGY 
accurate frequency plot, so I could compare apples to apples, including any DC 
offset that works its way into the signal st that that offset appears with a 
correct value.


just went to 
<http://www.octave.org>
to look for that package and could not find ANYTHING ?!

AND, (pkg install -forge ltfat) doesn't mean anything to me.

Since I evidently have no way of obtaining these  programs, would you send me 
the scripts, or such?

Regards,
Robert


--- address@hidden wrote:

From: address@hidden
To: "Macy" <address@hidden>
Cc: address@hidden
Subject: Re: How do YOU handle this disparity in fft?
Date: Wed, 12 Feb 2014 14:33:59 +0100 (CET)

>...snip....
This function is part of LTFAT (pkg install -forge ltfat) just better
documented and calling FFTW directly, but not including your sqrt(2)
normalization of the first and last sample (DC and Nyquest terms). There
is also a ifftreal function for getting back again.

Cheers,
Peter