Re: Spectral Analysis & Resynthesis of Audio Signals.. Anyone else into this? Applications/Tools that reliably produce good results?

On May 31, 1:08 am, dbd <d...@xxxxxxxx> wrote:
On May 31, 12:07 am, Ron N <ron.nichol...@xxxxxxxxx> wrote:

On May 30, 8:28 pm, dbd <d...@xxxxxxxx> wrote:
...
Reassignment doesn't have two halves.

Some reassignment kernels use symmetric and antisymmetric
weighted windows which is equivalent to splitting the data
into two halves about the window symmetry point in how
the weightings can affect the final composite result.
...
--
rhn A.T nicholson d.0.t C-o-M
http://www.nicholson.com/rhn/dsp.html

If reassignment constrained the symmetric and anti-symmetric kernels
used to have equal magnitudes and then performed sum and differencing
between them it would separate 'halves', but since the magnitudes can
differ and the results are ratioed, there are no 'halves' produced.

The method doesn't directly produce "halves", but the result
is still strongly affected by how the halves differ, given
the significant overlap in the two weighting magnitudes. That
may, in fact, be the primary contribution.

I look at reassignment, derivative and phase vocoder methods
as producing 2,3 or maybe 4 element vectors per frequency "bin",
thus retaining more signal information in the result computation.
The gain in "resolution" is related to the the gain in information
retained per bin (or bins) used in the analysis, before throwing
away the rest.

..


IMHO. YMMV.
--
rhn A.T nicholson d.0.t C-o-M
http://www.nicholson.com/rhn/dsp.html
.

Relevant Pages