Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!warwick!pipex!uunet!zaphod.mps.ohio-state.edu!caen!batcomputer!ghost.dsi.unimi.it!lucca
From: lucca@ghost.dsi.unimi.it (antonio lucca)
Subject: intelligent text-to-phonemes converter?
Organization: Computer Science Dep. - Milan University
Date: Mon, 24 May 1993 11:44:28 GMT
Message-ID: <1993May24.114428.1898@ghost.dsi.unimi.it>
Lines: 22

I'd like some references about intelligent text-to-phonemes
works, that is  algorithms that doesn't translate 
text to phonetic strings word by word, but entire
propositions as well, taking into account too much of the 
semantic informations in context. 

For example the Amiga phonemes system more than phonetic 
simbols, allows also pich-stress variation symbols (numbers)
to be optionally used in phonetic strings, that let the system 
speech to be less robot-like. But the built-in translator function 
(text to phonetic-strings) doesn't fully use these capabilities 
(I claim, it's because of the hardness of the problem).

Thank's in advance.



Antonio Lucca

lucca@ghost.dsi.unimi.it
Computer Science Department
Universita' di Milano
