Re: I have modified firls.m to include all types of FIRs plus HT and diff

[je suis]

Accidentally, while looking over the q vector, I saw there's no
slope/intercept, so I mistakenly added them. When I saw the figures
improved, I also realized that the derivation has no slope/intercept
and, instead, all I did was multiply by a constant.

After some hammering, I ended up using 4 as a multiplier, but,
honestly, I don't understand why it is so. The numbers, though, come
close to the example in the previous replies (the one with the 44th
order differentiator).

The FFT also shows very close results. I thought maybe it's a matter
of rounding up, probably related to the F(1)=0, so I played a bit with
the tolerances there and the workarounds for division by zero, but I
only saw changes, not necessarily good ones. Maybe it's a matter of
algorithms? The a=Q\b, or expint()? I don't know. The result is that
the 1/f^2 weighting for the differentiators doesn't quite give the
same results as Matlab.

In the end, after reviewing the rest of the script, I left it as it is
now, in the attachment. Still no checks, I'll add those only after the
script is declared as usable. The numbers for the rest look spot on,
but it's safer to say this after some new testing.

Vlad