Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!pipex!uunet!gatech!mailer.cc.fsu.edu!mailer.cc.fsu.edu!lilly
From: lilly@nhmfl.fsu.edu (Leonhard Lilly)
Subject: Re: The Klatt Synthesizer for Sparcs 
In-Reply-To: gordon@spot.Colorado.EDU's message of 5 Jan 93 16:21:24 GMT
Message-ID: <LILLY.93Jan5231915@prost.nhmfl.fsu.edu>
Sender: news@mailer.cc.fsu.edu (Usenet News File Owner)
Nntp-Posting-Host: prost.magnet.fsu.edu
Organization: Department of Physics, Florida State University, Tallahassee
	32306, USA
References: <1993Jan5.133424.8264@hubcap.clemson.edu> <gordon.726250884@spot.Colorado.EDU>
Date: 5 Jan 93 23:19:15
Lines: 47

In article <gordon.726250884@spot.Colorado.EDU> gordon@spot.Colorado.EDU (GORDON ALLEN R) writes:

>   I've also compiled klatt.  I presume the hello.par file should produce some
>   sound.  However, reading it in, calculating the waveform, and playing it back
>   does not seem to produce sound via the sparc d/a.  Also, attempting to save 
>   the waveform to a file produces a .au file that seems to have data in it and
>   a .dat and a .xgr file that are empty.
>
>   Does anyone know whats happening here?
>
>   Thanks
>
>
>   -- 
>   Allen Gordon   	         *If the folly of but one of us was changed to*
>   Research Associate         	 *intelligence, and divided amongst a thousand*
>   gordon@tramp.colorado.edu    *toads, each would be more intelligent than  *
>			         *Aristotle				      *

I tried to compile it as well. moving the *.c.00 files to *.c helps a
lot. Then you have to make sure that cc finds the audio library
libaudio.a and some of the SUN audio include files are in the right
place (either change the code or move the includes), check for al the
multimedia/*.h includes.

klatt itself doesn't produce any output for me either, but I didn't
check that too carefully. But uklatt hello.par mumbles "hello, my name
is". With  gen you can compose phonemes into longer structures (with a
lot of pauses), however not all phonemes are listed. Check lex.yy.c for
a complete listing. 
I didn't find any part doing the text to phoneme translation, and I
wasn't really impressed with the quality, but maybe somebody can
enlighten us as to how to tune the parameters etc.

Hope that helps

				Leo



--
-------------------------------------------------------------------------
Leonhard Lilly                     Internet:  lilly@theory.nhmfl.fsu.edu
Department of Physics                 Phone:  (904) 644 - 5075
FSU - NHMFL          
Tallahassee, FL 32306
-------------------------------------------------------------------------
