Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!doc.ic.ac.uk!pipex!howland.reston.ans.net!newsserver.jvnc.net!newsserver.technet.sg!ntuix!ntuvax.ntu.ac.sg!sd2197793
From: sd2197793@ntuvax.ntu.ac.sg
Subject: Pitch Detection Question
Message-ID: <1993Oct14.193840.1@ntuvax.ntu.ac.sg>
Lines: 29
Sender: news@ntuix.ntu.ac.sg (USENET News System)
Nntp-Posting-Host: v9000.ntu.ac.sg
Organization: Nanyang Technological University - Singapore
Date: Thu, 14 Oct 1993 11:38:40 GMT

Dear comp.speech readers,

As I am implementing LPC speech compression, I have been reading Digital Signal
Processing Applications Using The ADSP-2100 Family, Analog Devices, Prentice
Hall.

I refer to page 366, where pitch detection is discussed.  There, an equation
is given as
		
r sub e (k) = sum from j=1 to P { r sub a (j) times r sub s (j-k) }

where k=0...windowlength.

On page 372, under references, pitch detection is attributed to Markel
J D and Gray in 1980 Linear Prediction of Speech (New York: Springer-Verlag).

I do not have this book.

I would like to know how the above equation is derived.  It certainly worked
in extracting the pitch, by searching for the peak.


I would be most grateful if someone could derive the equation, or if it is
a simplification (to save computing time), to explain how that came about.

Tsay Mien

SD2197793@ntuvax.ntu.ac.sg

