Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!rutgers!njin!princeton!phoenix!lseltzer
From: lseltzer@phoenix.Princeton.EDU (Linda Ann Seltzer)
Newsgroups: comp.music
Subject: Re: Digital Signal Processing spoken here?
Message-ID: <2026@idunno.Princeton.EDU>
Date: 22 Aug 90 17:56:31 GMT
References: <812@dg.dg.com>
Sender: news@idunno.Princeton.EDU
Organization: Princeton University, Princeton, New Jersey
Lines: 33

In article <812@dg.dg.com> smcgarry@dg-rtp.dg.com (Steve McGarry) writes:
> 
>The project I'm working on is a spectrum analysis of a digitized input
>signal.  Is there any
>source code available (with some explanations) for doing FFTs or Digital
>Filtering?

There are spectrum analyzers available which are stand alone units
independent of any computer.  Usually they take an analog input.
B&K manufactures one.  you would have to find some institution that
owns one.

Also, depending on your needs making spectrograms from an analog
signal may be sufficient, unless you really need to know the exact
numerical values of the locations of the peaks and valleys.

So there may be ways you can look at the spectrum of a signal without
worrying about FFT software.

As for software, probably the best tool for you is the ILS signal
processing package from Signal Technology in Santa Barbara.  There
are also other signal processing packages and they all include
graphics.  The problem is interpreting the output you get from
these packages, because I guarantee you that you will not always
get clean, understandable output.  If what you want to see is a
few peaks, doing linear prediction and then using the coefficients for
spectral estimation would show you the peaks without all the harmonics
and other garbage.

So the first question to ask is - what kind of signal do you have
and what are you interested in finding out.  You will probably
need to consult someone in electrical engineering at your institution
to obtain help in using whatever software package they have.