Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!pipex!uunet!utcsri!torn!watserv2.uwaterloo.ca!watserv1!sunee.uwaterloo.ca!broehl
From: broehl@sunee.uwaterloo.ca (Bernie Roehl)
Subject: Re: TMS5220 speech synthesizer
Message-ID: <BvL0nC.DsF@watserv1.uwaterloo.ca>
Sender: news@watserv1.uwaterloo.ca
Organization: University of Waterloo
References: <BvI2wu.A7t@usenet.ucs.indiana.edu>
Date: Sun, 4 Oct 1992 05:30:47 GMT
Lines: 23

In article <BvI2wu.A7t@usenet.ucs.indiana.edu> mdchaney@fractal.ucs.indiana.edu (M Darrin Chaney) writes:
>Does anyone have a programming manual for the TMS5220 chip from TI?  This
>was used in their home computer 12 years ago, and produced excellent speech
>(very near human quality) at about 200bytes/second.  All I know now is that
>it used a mathematical model of the human vocal tract.

This is what's known as an LPC algorithm.  It uses a set of parameters
(typically 10 or 12) to model the resonances of the vocal tract.

(An aside: Does anyone have any code for LPC compression/decompression?  I
came across one on wuarchive; it looks quite good, put it only does
compression, not playback).
 
>I've not been able to ever find programming help for it, not even from TI.

I believe it's been replaced by the TMS50c20 and its relatives.  I know the
50c40 (a preprogrammed LPC player) will support 5220 format speech roms.

-- 
	Bernie Roehl, University of Waterloo Electrical Engineering Dept
	Mail: broehl@sunee.UWaterloo.ca
	BangPath: uunet!watmath!sunee!broehl
	Voice:  (519) 885-1211 x 2607 [work]
