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

Using Spoken Telephone Numbers to Reduce Perplexity in Data Input

IP.com Disclosure Number: IPCOM000115440D
Original Publication Date: 1995-May-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 2 page(s) / 81K

Publishing Venue

IBM

Related People

Sherwin Jr, EB: AUTHOR

Abstract

Disclosed is the use of a spoken telephone number, together with a database relating name and address information to telephone numbers, for providing data input to a system having speech recognition capabilities. This method is particularly useful when applied to the processing of credit card application forms, since it is very difficult, or presently even impossible, for a speech recognition system to process spoken name and address information reliably. The limitation of each telephone number digit to ten possibilities, compared to the large number of possibilities for name and address data, greatly reduces the perplexity of the spoken data to be recognized by the system.

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

Using Spoken Telephone Numbers to Reduce Perplexity in Data Input

      Disclosed is the use of a spoken telephone number, together
with a database relating name and address information to telephone
numbers, for providing data input to a system having speech
recognition capabilities.  This method is particularly useful when
applied to the processing of credit card application forms, since it
is very difficult, or presently even impossible, for a speech
recognition system to process spoken name and address information
reliably.  The limitation of each telephone number digit to ten
possibilities, compared to the large number of possibilities for name
and address data, greatly reduces the perplexity of the spoken data
to be recognized by the system.

      To use this method, a user, such as a data entry clerk, reads
the full telephone number, from a form, such as a credit card
application form, into a system having speech recognition capability.
This spoken information is converted into a machine-readable format
through the use of a continuous speech subsystem, such as ICSS (IBM*
Continuous Speech Series).  A request is made to the database for the
address associated with this phone number.  This method is preferably
chosen because a database containing the nationwide reverse directory
exceeds the storage capacity of most workstations and many Local Area
Network (LAN-based) systems.

      Since the database call typically has to go back to a
mainframe, several seconds are generally required for processing.
During this time, in a preferred implementation of this method, the
user speaks the zip-code from the document into the system.  This
data is converted into a machine-readable format, and the city and
state of the address are retrieved from a zip-code database.  This
database can typically reside on the workstation or LAN, since a
database of five-digit zip-codes can fit on a single (CD) Compact
Disk.  By this point, the reverse directory database has either
returned an address matching the telephone number or a code
indicating that no address can be found.

      In a preferre...