Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!doc.ic.ac.uk!agate!ames!haven.umd.edu!uunet!munnari.oz.au!hp9000.csc.cuhk.hk!cucs5.cs.cuhk.hk!yu030
From: yu030@cs.cuhk.hk (Yu Wai Leung)
Subject: Re: Klatt SYnthesizer (How to work)
Message-ID: <1993Jan11.013217.8139@cucs5.cs.cuhk.hk>
Sender: news@cucs5.cs.cuhk.hk
Organization: Faculty of Engineering, The Chinese U. of Hong Kong
References: <1993Jan8.024840.12894@cucs5.cs.cuhk.hk> <gordon.726507433@spot.Colorado.EDU>
Date: Mon, 11 Jan 1993 01:32:17 GMT
Lines: 35

gordon@spot.Colorado.EDU (GORDON ALLEN R) writes:

>I got this to work, sort of.  With the file hello.par, klatt produces a .dat
>file, which I presume is the waveform and a .xgr file, which can be used with

>	graph ... | xplot

>to produce a plot of the waveform.  When I attempt to play the signal, it does
>produce sound on the sparc, but it is not recognizable.  The waveform does not
>really look like speech, also.  Is this some sort of logarithmic compression?

>Finally, when klatt is exited, I must run stty sane to get the keyboard/monitor
>to respond normally.

>Any suggestions?

>Allen

According to my interpretation, "klatt" will produce files with the 
following extensions: .xgr, .dat and .au. Only files of extension
.au are playable on sparc stations. And the .xgr and .dat files
are binary files which contain the entire waveform produced by the
synthesizer. If you found that the .xgr's and .dat's non-readable,
you can try to extract the waveform by yourself by looking into
the content of the short integer array "iwave" in the source file
"klatt.c". In "klatt.c", the content of "iwave" are compressed
(using mu-law) and the results will be placed in a character array
called "plabuf". 

Cheers!

Yu Wai-Leung
Dept of Computer Science, 
Chinese University of Hong Kong
e-mail: yu030@cs.cuhk.hk
