Newsgroups: comp.speech
Path: lyra.csx.cam.ac.uk!warwick!zaphod.crihan.fr!jussieu.fr!math.ohio-state.edu!usc!nic-nac.CSU.net!charnel.ecst.csuchico.edu!olivea!decwrl!murky.apple.com!mumbo.apple.com!gallant.apple.com!NewsWatcher!user
From: minow@apple.com (Martin Minow)
Subject: Re: Speech synthesis and phonemes - help!
Sender: news@gallant.apple.com
Message-ID: <minow-120594120316@17.203.48.21>
Date: Thu, 12 May 1994 19:09:37 GMT
References: <1994Apr29.110924.14879@unix.brighton.ac.uk>
Organization: Macintosh Developer Services
Followup-To: comp.speech
Lines: 23

In article <1994Apr29.110924.14879@unix.brighton.ac.uk>,
jwr@unix.brighton.ac.uk (reed) asked for help developing a speech
synthesizer.

This is a hard problem to solve, as many of us have discovered.

If you can find it, there is a Basic implementation of the NRL
text-to-phoneme rules in the Decus USA proceedings, Spring 1977.
"A RSTS/E Audio Response Application" J. C. Broihier & M. J.
Crowley, pp 1355-1376. (Try a good university library, or
beg a favor from your local Dec office.)

This will get you about 50-60% accuracy in converting running
English text. The rest requires a dictionary. This is, by the
way, not much worse than the rules from "high-end" synthesizers.

The rest will require digging through the literature, both books
and journal (primarily JASA) articles.

Good luck.

Martin Minow
minow@apple.com
