Newsgroups: comp.speech
Path: cantaloupe.srv.cs.cmu.edu!europa.chnt.gtegsc.com!howland.reston.ans.net!math.ohio-state.edu!newsfeed.acns.nwu.edu!ftpbox!mothost!schbbs!salaam!assaleh
From: assaleh@mot.com (Khaled Assaleh)
Subject: Re: Help...
Reply-To: assaleh@mot.com
Organization: Motorola, Inc., GSTG
Date: Tue, 27 Jun 1995 20:44:06 GMT
Message-ID: <1995Jun27.204406.5855@schbbs.mot.com>
References: <3so599$c66@senator-bedfellow.MIT.EDU>
Sender: news@schbbs.mot.com (SCHBBS News Account)
Nntp-Posting-Host: 137.144.1.19
Lines: 27

In article c66@senator-bedfellow.MIT.EDU, dalamar@ATHENA.MIT.EDU (Rand Al'Thor) writes:
> OK, I can't figure out if I'm too stupid to understand this, or
> there's a typo in the book, but... 
> 
> On page 115 of Rabiner & Juang, they present a method for obtaining N
> cepstral coefficients from M LPC coefficients. It says that generally
> N > M, and gives a method for finding cepstral coefficients M+1,...,N.
> However, the recursive equations that give the cepstral coefficients
> seem to depend on LPC coefficients M+1,...,N-1. My brain tells me
> these are not defined. Can anyone help? 
> 
> Thanks. 
> 
>   --Craig
> dalamar@athena.mit.edu
> -- 
>   --Craig
> 
> dalamar@athena.mit.edu

Actually, N can go up to infinity regardless of M, just consider a_i = 0.0 
for (i>M+1).

Hope this helps!

-Khaled
 
