From: Eric Durant <edurant@engin.umich.edu>
Subject: Re: fft for spectrum display
Date: 03 Oct 1999 00:00:00 GMT
Message-ID: <37F7E382.54F828F@engin.umich.edu>
Content-Transfer-Encoding: 7bit
References: <7t8kvl$7av$1@news1.xs4all.nl>
X-Accept-Language: en
Content-Type: text/plain; charset=us-ascii
Organization: University of Michigan EECS
Mime-Version: 1.0
Newsgroups: comp.dsp

Marco Nelissen wrote:
>Are there any shortcuts that could be taken to speed up an FFT, 
>given that it will be used for a spectrum analyzer display and 
>therefore doesn't need to be extremely accurate? All of the free 
>(*really* free, as in: non GPLd) fft code I found on the web is 
>relatively slow, but I'd like something to spice up an audio 
>application without increasing cpu usage too much.

It sounds like a power spectral density (PSD) may be more appropriate
for your use than a DFT.  MatLab has a variety of PSD implementations
and points to the following references:

:   [1] Petre Stoica and Randolph Moses, Introduction To 
:       Spectral Analysis, Prentice hall, 1997, pg, 15
:   [2] A.V. Oppenheim and R.W. Schafer, Discrete-Time Signal 
:       Processing, Prentice-Hall, 1989, pg. 731
:   [3] A.V. Oppenheim and R.W. Schafer, Digital Signal       
:       Processing, Prentice-Hall, 1975, pg. 556

-- Eric Durant
   http://www.edurant.com/