Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site rna.UUCP Path: utzoo!linus!philabs!cmcl2!rna!dan From: dan@rna.UUCP (Dan Ts'o) Newsgroups: net.periphs,net.micro.pc Subject: Re: wanted: text-to-speech synthesizer recommendations Message-ID: <369@rna.UUCP> Date: Wed, 20-Feb-85 20:44:54 EST Article-I.D.: rna.369 Posted: Wed Feb 20 20:44:54 1985 Date-Received: Sun, 24-Feb-85 04:45:24 EST References: Reply-To: dan@rna.UUCP ( Ts'o) Organization: Rockefeller Neurobiology, NYC Lines: 24 Xref: linus net.periphs:526 net.micro.pc:2977 Summary: x I have seen and heard a number of text-to-speech boxes including VOTRAX's and the original speak(6) program which came with early versions of UNIX. The only text-to-speech box which I consider "fair-to-good" in terms of speech clarity and quality is DECTalk. Unfortunately it is $4000. The others, like VOTRAX's Type-and-talk are sort-of-okay. It is only $250, but the quality is at very best "fair" and often "poor". I wrote a simple preprocessor to the VOTRAX box which recognizes words and "rewrites" the word to produce better speech output. Often the box gets fooled and the output is rather unintelligible. With the preprocessor, one can even give the box a bit of a French accent. But it begins to involve massive table lookup. In short DECTalk is impressive, though far from perfect. The others I've heard are pretty bad and should at least have some preprocessing help. Cheers, Dan Ts'o Dept. Neurobiology Rockefeller Univ. 1230 York Ave. NY, NY 10021 212-570-7671 ...cmcl2!rna!dan