Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!pipex!uunet!nntp.telebit.com!phr
From: phr@telebit.com (Paul Rubin)
Subject: Re: Algorithm for pitch detection
Message-ID: <C79o3n.LL9@telebit.com>
Sender: news@telebit.com
Nntp-Posting-Host: napa
Organization: Telebit Corporation; Sunnyvale, CA, USA
References: <1993May17.200530.16738@en.ecn.purdue.edu> <1993May18.162011.3214@Princeton.EDU> <1993May18.184742@IASTATE.EDU> <C79F11.C39@rahul.net> <C79n2p.LEu@telebit.com>
Date: Wed, 19 May 1993 09:09:22 GMT
Lines: 16

>davidj@rahul.net (David Josephson) writes:

>Hello netters,

>I posed a question on this a few weeks ago and there were no takers,
>so here goes again. What's a good strategy for pitch detection when
>the input signal is complex and not harmonically related? Is there
>anything less intensive than a full FFT per frame to go on? Problem
>is, both in speech and ethnomusicology (and everywhere in between)
>there are often interesting signals that are not harmonic series, so
>autocorrelation and wavelet techniques won't work.

There are all kinds of methods, and doing an FFT isn't such a good
one.  See F. J. Owens, _Signal Processing of Speech_
(McGraw-Hill 1993) for an overview.  The references will point
you to more detailed treatments.
