Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!mit-eddie!uw-beaver!Teknowledge.COM!polya!davem
From: davem@polya.Stanford.EDU (David K. Mellinger)
Newsgroups: comp.dsp
Subject: Re: log frequency fft
Message-ID: <11961@polya.Stanford.EDU>
Date: 26 Sep 89 20:07:55 GMT
References: <89264.171306P85025@BARILVM.BITNET> <9520001@hpsad.HP.COM> 
Reply-To: davem@Polya.Stanford.EDU (David K. Mellinger)
Organization: Stanford University
Lines: 16

>In article <9520001@hpsad.HP.COM> toma@hpsad.HP.COM (Tom Anderson) writes:
>
>   It seems like you need a logarithmic frequency axis.  I have often
>   wished for an FFT type algorithm with logarithmic frequency spacings.
>   Does anyone know of one?

Try looking at "An Algorithm and Architecture for Constant-Q Spectrum
Analysis," by Gary W. Schwede.  "Q" refers to the ratio of bandwidth
to frequency, sometimes called the "quality" of a filter.
Constant Q <==> logarithmic frequency.

The paper is in the Proceedings of the IEEE International Conference
on Acoustics, Speech, and Signal Processing (ICASSP), April 1983, pp.
1384-1387.
				Dave Mellinger
				davem@cs.Stanford.EDU