Browse Prior Art Database

Steno Captioning as a Speech Aid for the Speaking-Impaired

IP.com Disclosure Number: IPCOM000108544D
Original Publication Date: 1992-Jun-01
Included in the Prior Art Database: 2005-Mar-22
Document File: 4 page(s) / 178K

Publishing Venue

IBM

Related People

Kanevsky, D: AUTHOR [+4]

Abstract

There are about 2.4 million speaking-impaired in the USA. Currently their practical methods of communication are speech synthesizers, vocal synthesizers and sign-language. All these methods as they are used now suffer from several deficiencies. Sign-language is not known by most people, vocal synthesizers require expensive surgery, and the speech is incoherent. Speech synthesizers appear to be the most among counted above methods. Speech synthesizers are connected to computers and convert to voice messages what users type in on keyboards attached to these computers. With technical progress it is expected that this setup (speech synthesizer-laptop computer) will be implemented in a portable communication device suitable for everyday use (1).

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 37% of the total text.

Steno Captioning as a Speech Aid for the Speaking-Impaired

       There are about 2.4 million speaking-impaired in the USA.
Currently their practical methods of communication are speech
synthesizers, vocal synthesizers and sign-language. All these methods
as they are used now suffer from several deficiencies.  Sign-language
is not known by most people, vocal synthesizers require expensive
surgery, and the speech is incoherent.  Speech synthesizers appear to
be the most among counted above methods.  Speech synthesizers are
connected to computers and convert to voice messages what users type
in on keyboards attached to these computers.  With technical progress
it is expected that this setup (speech synthesizer-laptop computer)
will be implemented in a portable communication device suitable for
everyday use (1). Also, there are continuous efforts on improving the
quality of the text-to-voice conversion and implementing some
personal characteristics of a user (man vs. woman, age, etc.).  But
the following problem of the speech synthesizer approach has not yet
seriously addressed: conventional typing is slow.  Even experienced
users usually produce, via keyboard typing, not more than 60 words
per minute. Here, this problem is addressed by considering advanced
methods for producing typed texts via steno technology and automatic
error correction methods.

      Skillful stenographers are able to type 200 words per minute.
The setup that permits such high speed consists of a steno-machine
connected to the computer with a special program that translates the
stenographer's typed phonetic utterance into words. Our basic
suggestion is to permit the speaking-impaired to produce voice output
in real time. In order to do this, they will have to learn
stenography and type their messages on steno-machines connected to a
speech synthesizer.  Throughout this article we refer to this
stenotyping-to-speech system as steno-talk.  Actual implementation of
this solution requires solving the following problems.
1.  Character of the connection of a speech synthesizer to a
steno-machine
      The current steno-machines produce two levels of outputs.
First, they produce a phonetic-like utterance. Second, this phonetic-
script is translated into words.  This translation is based on a
lookup table and on sophisticated algorithms that use some
statistical, semantic and syntactic language features. This word
sequence can contain unrecognized utterances that happen if the typed
words are not in an attached vocabulary or if the stenographer
introduced errors in typing the phonetic-like utterance.  In turn the
text-to-speech converter is working by attaching phonetic labels to
the string of ASCII characters. This labeling procedure also is based
on a lookup table, and on sophisticated statistical, semantical and
syntactic algorithms (2).  Since the task of the suggested setup is
converting steno-typing utterance into voice, one can suggest more
strai...