Hi,

I'm currently a 17-year-old student trying to do a project, involving
embedding data within an audio file. Hopefully, it will survive MP3
compression... so I've come to the conclusion I will have to convert all the
samples to the freq. domain to do any analysis.

I downloaded FFTW and I have it working, but it assumes that the user knows
what he's doing (which I do not really).

I found this software kit at http://www.eobj.com/eofft-dll.html that would
be perfect, except it costs $60... Basically, it will calculate forward and
inverse FFT and can add extra output tasks such as different windowing
functions, converting output of forward FFT to magnitude (dB) or power.

Is there any software kit (preferably in C/C++ or Java) for free that can
provide similar features? Actually, all I really need are examples showing
me how to do the above tasks I mentioned (mainly converting output of
forward FFT to magnitude, after using a windowing function).


Thanks.


David Chen

David Chen wrote:

This might do what you want:

http://www.iowegian.com/scopedsp.htm

Also, there's a lot of tutorial info at www.dspguru.com that you might
find interesting.

--
Jim Thomas Principal Applications Engineer Bittware, Inc
jthomas@bittware.com http://www.bittware.com (703) 779-7770
There are 10 kinds of people: the ones who understand binary code,
and the ones who don't.

I'm not sure what you are trying to do though it sounds like you are
attempting to fingerprint recordings. Designing signals that survive
conversion to MP3 requires an expert understanding of the conversion process
and only indirectly would be helped by understanding frequency analysis.

If you have a real signal organized as single or double precision real data,
you can fill a buffer of length N and compute the FFT using the rfftw series
of functions. The output of rfftw is complex organized as follows:

Real (0) DC
Real frequency (1)
Real frequency (2)
..
Real frequency (N) Nyquist frequency
Imaginary frequency (N-1)
Imaginary frequency (N-2)
...
Imaginary frequency (1)

The magnitude at a frequency I is

mangitude(I) = sqrt(Real(I) * Real(I) + Imaginary(I) * Imaginary(I))

QED

Windowing may be done in the time domain by multiplying the input series by
the window function on a point by point basis or in the frequency domain by
convolution. It is generally easier to window in the time domain. The Harris
paper on windowing mentioned in the comp.dsp FAQ is a good starting point on
the subject of windowing.

In fact, the comp.dsp FAQ is a good starting point for any question you might
want to ask in this group.

In article <qvLKa.4$BS4.2700474@newssvr21.news.prodigy.com> , "David Chen"
<david_c11@yahoo.com> wrote:

"David Chen" <david_c11@yahoo.com> wrote in message news:<qvLKa.4$BS4.2700474@newssvr21.news.prodigy.c om>...
Hi David
u may want to try the software called Autosignal. U can download the
trial version for free from the web. just search for it in google.
-Nithin