Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Training Through an Automatic Speech Recognizer Comprehension Ability of Hearing Impaired

IP.com Disclosure Number: IPCOM000121126D
Original Publication Date: 1991-Jul-01
Included in the Prior Art Database: 2005-Apr-03
Document File: 2 page(s) / 89K

Publishing Venue

IBM

Related People

Gopalakrishnan, PS: AUTHOR [+3]

Abstract

This article describes a method of using an Automatic Speech Recognizer (ASR) to train hearing impaired individuals to better perceive and comprehend spoken words. The standard ways of training had so far only limited success. The new method is based, first, on using ASR output together with other natural sources of information that are provided by impaired hearing, lipreading and tactile sensors; and it eventually leads to the elimination of the use of ASR by the hearing impaired individual.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Training Through an Automatic Speech Recognizer Comprehension Ability
of Hearing Impaired

      This article describes a method of using an Automatic
Speech Recognizer (ASR) to train hearing impaired individuals to
better perceive and comprehend spoken words.  The standard ways of
training had so far only limited success.  The new method is based,
first, on using ASR output together with other natural sources of
information that are provided by impaired hearing, lipreading and
tactile sensors; and it eventually leads to the elimination of the
use of ASR by the hearing impaired individual.

      There are various methods aimed at improving the ability of
hearing impaired individuals to comprehend speech.  In these
procedures the hearing impaired learn typical speech patterns either
using hearing/tactile devices, or trying to lipread.  For profoundly
impaired subjects these procedures often give only marginal results.
This can be explained by the fact that certain special inner brain
structures that provide recognition of speech patterns are
underdeveloped.  In the process of training they do not receive
enough information to cause changes in these underdeveloped inner
structures [1].  In this article we suggest a method for increasing
the information stream to underdeveloped inner brain structures,
thereby providing a more efficient training procedure for learning
typical speech patterns by the hearing impaired.  The new procedure
is based on the observation that if a person uses a few available
sources of information (say, via sight, tactile and hearing sensors)
and is able to develop a good comprehension of speech, then he still
can have good comprehension with a smaller number of sources of
information (say, dropping tactile devices (see [2]).  The reasonable
explanation for this followin...