Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!pipex!uunet!mcsun!sunic!kth.se!alv.nada.kth.se!d88-jwa
From: d88-jwa@alv.nada.kth.se (Jon Wtte)
Subject: Re: Mac/Amiga speech, non-english languages
In-Reply-To: blocker@dominic.ssc.gov's message of Fri, 2 Oct 1992 22:20:25 GMT
Message-ID: <D88-JWA.92Oct3124528@alv.nada.kth.se>
Originator: d88-jwa@alv.nada.kth.se
Sender: usenet@kth.se (Usenet)
Nntp-Posting-Host: alv.nada.kth.se
Organization: Royal Institute of Technology, Stockholm, Sweden
References: <1992Oct2.164714.3268@fel.tno.nl>
	<1992Oct2.222025.27565@sunova.ssc.gov>
Date: Sat, 3 Oct 1992 11:45:28 GMT
Lines: 62

> blocker@dominic.ssc.gov (Rich Hall) writes:

   There were several requests for Mac speech synthesizers. MacInTalk is a
   Borland utility that came with my version of Turbo Pascal for the Mac. 
   If I remember correctly, you could write to it just as easily at writing
   to standard output or a data file.
   The default mode was full text-to-speech conversion, but there was also a
   phoneme mode. 

No,

MacinTalk was a library from Apple, using the original sound driver
for the Mac 128k. This is '83 technology for PCs, not very hot...
Borland bundled the library with its pascal compiler, but since
Borland didn't have the drive to support that product, the compiler
is not available since several years.

The original Sound Driver calls are somewhat out of mode these days :-)
and MacinTalk, being a library and not trap based, is not compatible
with the enwest machines or system software version. HOWEVER - Apple
has some quite exciting voice activities going on, both in recognition
and text-to-speech. Text-to-speech appears to be in beta, though
seeding is limited, and they have one high-end engine (doing both
female and male voices really well) and one low-end engine (targetted
as generally available, so everyone can have it and developers get
the support for using SpeakString()...)

For those familiar with DecTalk, the new high-end package is much
better (but uses more memory...) Didn't Apple buy DecTalk some time
ago?

   characters of all fonts, styles, and sizes with a very high degree of
   accuracy. Then, these OCR algothms were adapted, with minor modifications,
   to reading handwriting. Are these same pattern-matching routines being
   used to match vocal sound waves to those of standard phonemes?

No. First, you have more information in handwriting (you know which
way the character was drawn) but you have more variety, so handwriting
is NOT (or should not be) done with traditional OCR techniques. Also,
recognizing phonemes is probably best done in the fourier domain,
whereas OCR pattern-matching is done in the sample domain.

   QuickTime on the Mac and MicroChannel IBMs have made significant advances 
   toward competing with the Amiga's separate Audio and Video coprocessors, 
   and have the advantage of being more widely available. Multimedia packages 

And have the advantage of being an emerging standard, and having
well-defined APIs and file exchange formats. I think that is much
more important than whatever the current hardware "can do." Also,
QuickTime is not only sound & video, it's a way of referring to
information with a time code.

Now, is there an online place I can find more information on
distinguishing phonemes in written english (or other) text?

Cheers,

-- 
Jon W{tte, h+@nada.kth.se, Sweden, Phone +46-8-107069

"From now on, I think I'll re-label the EQ on the
 desk as Fizz and Wobble, rather than HF and LF."
