[935] in linux-announce channel archive
AbbotDemo 0.6: A large vocabulary continuous speech recognition system
daemon@ATHENA.MIT.EDU (Lars Wirzenius)
Sat Aug 19 09:11:56 1995
Date: Fri, 18 Aug 1995 19:09:48 +0300
From: Lars Wirzenius <wirzeniu@cc.helsinki.fi>
To: linux-announce@vger.rutgers.edu
From: ajr@eng.cam.ac.uk (Tony Robinson)
Newsgroups: comp.os.linux.announce
Subject: AbbotDemo 0.6: A large vocabulary continuous speech recognition system
Organization: Engineering Department, Cambridge University, England.
Reply-To: AbbotDemo@compute.demon.co.uk
Approved: linux-announce@news.ornl.gov (Lars Wirzenius)
Followup-to: comp.os.linux.development.apps
AbbotDemo is a 5,000 word near-real-time speaker-independent continuous
speech recognition system for American and British accented English.
It is available by FTP from sunsite.unc.edu (and mirrors) in directory
/pub/Linux/apps/sound/speech. The file AbbotDemo.README gives more
information on this system and AbbotDemo-0.6-Linux.tar.gz provides
executables (2.3 Mbyte). Sorry but at this stage no sources and only
limited documentation is provided.
You'll need X, a 16 bit soundcard and a reasonable quality microphone to
get it to work. I run it on a 486/66 with 12 Mbyte RAM.
This package is meant to provide a fair representation of the state of
the art in large vocabulary speech recognition. It will make mistakes,
something more than one word wrong in every ten words uttered (depending
on how quiet your acoustic environment is, the quality of your audio
kit, how clearly you speak, how well your accent is matched to those the
system was trained on, whether you use just the words in the dictionary
and how well the words fit the language model which was derived from the
Wall Street Journal).
We hope that the package will provide a bit of fun, entertainment and
education in speech recognition. If you think of some cool new uses
for this technology please let us know!
Tony Robinson (Cambridge University)
Mike Hochberg (Cambridge University)
Steve Renals (Sheffield University)
and many many more.
--
Send comp.os.linux.announce submissions to: linux-announce@news.ornl.gov
PLEASE remember a short description of the software.
