1 Nov
2002
1 Nov
'02
1:30 p.m.
On Fri, Nov 01, 2002 at 05:47:04 +0100, Tim Goetze wrote:
the difference between the two is next to nothing,
mostly a DC
offset. i'll use yours for the next FTs. is that 'Harris' as in
'S.W.Harris'? ;)
Er, no ;) Harris is a very common surname.
after some NR 5-10 reading, it looks to me like
harmonic_gen is
missing the work pcshft() from the book text is performing, yet
its output seems well-scaled.
I think thats for rescaling the domain from [-1,1] to [a,b], but our
domain is [-1,1], more-or-less, so we are OK. To make sure we should
probably clamp the input to [-1,1]. I could also have remembered it wrong
though.
it just occurred to me that it may be feasible to skip
the
coefficient calculation in realtime by simply storing these
instead of harmonic amplitude tables. makes lerping difficult
though.
IIRC that is what we discussed before. It might well work, but its just an
optimisation, so best leave it out for now.
- Steve