AT&T Watson Speech Recognition
- Platform: Windows 95/NT on a Pentium 75 Mhz or higher
- Description: Watson is a software implementation of AT&T
Bell Laboratories voice processing technology. Watson includes
BLASR Speech Recognition and FlexTalk
speech synthesis (see Q5.5).
It requires no special hardware to run other than a standard sound card
and/or phone card. Technical details for BLASR Speech Recognition include:
- Compliant with Microsoft
Speech API and
Telephone API
- Speaker independent, continuous speech recognition
- Fast, run-time vocabulary change
- Open mic and telephone line environments
- SoundBlaster compatible sound card and drivers required
- Subword models and whole-word digit models
- Background, silence, and filler/garbage models
- 50 word name vocabulary or 100 word phrase real-time recognition with 95%
accuracy
- Rejection of out-of-vocabulary words
- American English only - other languages in development
- Barge-in speech begin/end notification - requires hardware echo
cancellation
The AT&T Advanced Speech Products
Group home page provides more detailed information including a
Frequently Asked Questions list,
information for application developers on the
Independent Software
Vendor (ISV) Program (including info on the
SDK,
licensing, and the
training program).
- Requirements: Uses 2 MB RAM, 10 MB Disk. Requires
a Pentium 75 MHz or higher CPU (uses < 50% CPU).
- Cost and Availability: WATSON is a software-based speech
platform with a Software Developers Kit (SDK) that allows application
developers to use voice processing in their applications.
It is not available as a stand-alone product.
Licensing information (inc. price) is provided in the
AT&T Advanced Speech Products
Group home page
- See also:
Watson FlexTalk
speech synthesis in Q5.5,
Microsoft Speech API, and Advanced Speech API.
- Contact: AT&T Advanced Speech Products Group
Suite 700, 44 East Mifflin Street, Madison, WI 53703, USA
Ph: 1-800-5-WATSON, Fax: 1-608-259-2269
Email: aspg@attmail.com
WWW: http://www.att.com/aspg/
Back to
Q6.5 of
Section 6 of the
comp.speech FAQ Home Page.
Administrivia,
Copyright,
Submit Information :
Last Revision: 13:49 31-May-1996