From: Rune Allnor on
On 1 Feb, 12:41, "Afinko" <afi...(a)> wrote:
> Hi,
> I am trying to use Kaiser-Bessel window before FFT.
> However, I am a bit confused about what is the difference between Kaiser
> and Kaiser-Bessel window.

Unless somebody comes up with very compelling arguments to
do otherwise, I'd interpret 'Kaiser-Bessel' the same way
as 'Dolph-Chebychev': Kaiser used Bessel functions to design
his window just the same way Dolph used Chebychev polynomials
to design his.

From: Richard Dobson on
On 01/02/2010 13:36, Grant Griffin wrote:
> Afinko wrote:
>> I found a matlab code for generating "Kaiser-Bessel derived (KBD)
>> window":
>> kbdwin.m
>> With this function, such KBD windows can be generated:
>> And, these are not like the Kaiser windows (generated by kaiser.m) at
>> all:
>> Which means, that at least, "Kaiser" and "Kaiser-Bessel derived" windows
>> are not the same.
> Sorry about that. Can you tell us the advantage of the Kaiser window
> over the Kaiser-Bessel window?

The picture referenced above shows some (relatively) strange-looking
windows; one looks like a gothic arch, the others with increasing
amounts of flat-topness. That is definitely different from a "classic"
Kaiser window which with the right param can be made to closely
approximate Hann, Gaussian, etc.

Richard Dobson

From: Vladimir Vassilevsky on

Afinko wrote:

>>Sorry about that. Can you tell us the advantage of the Kaiser window
>>over the Kaiser-Bessel window?
> Well, I have no idea what is the difference between "Kaiser",
> "Kaiser-Bessel", or "Kaiser-Bessel derived" window. I am totally confused.
> But I found in few papers, that "Kaiser-Bessel derived" window is commonly
> used in Advanced Audio Coding (AAC) codec.
> Is there someone, who is clear in this topic?

KBD window is Kaiser window modified in the way so overlapped DCT would
be invertible.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant