Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!warwick!marble.uknet.ac.uk!mcsun!uunet!elroy.jpl.nasa.gov!swrinde!network.ucsd.edu!munnari.oz.au!titan!dion.trl.oz.au!duke
From: duke@dion.trl.oz.au
Subject: comp.speech
Message-ID: <1993Mar29.135006.1@dion.trl.oz.au>
Lines: 18
Sender: root@trl.oz.au (System PRIVILEGED Account)
Organization: Telecom Research Laboratories, Melbourne, Australia
References: <3057@irit.irit.fr>
Date: Mon, 29 Mar 1993 18:50:06 GMT

In article <3057@irit.irit.fr>, joly@cerise.irit.fr (Philippe JOLY) writes:
> Could someone direct me to an ftp site where i can find
> something about SPEAKER RECOGNITION .
> (I'm trying to find automatically when actors are playing in 
> a part of a movie .)
> 
Speaker recognition has two main branches: speaker verification and speaker
identification.  The former uses cooperative speakers who wish to gain access
to a service, and would not be suitable for your task. The latter is much more
difficult and is normally the domain of forensic science (proving who made
some threat or conspiracy over the phone, for example.)  I suggest you look
in that direction.  My feeling is that it would be very labour intensive in 
selecting the appropriate speech tokens on which to base the analysis, but
this is not my field.

Peter Duke
p.duke@trl.oz.au

