Browse Prior Art Database

Simultaneous Use of Dual Tone Multi-Frequency and Voice Recognition in Voice Response Unit Applications

IP.com Disclosure Number: IPCOM000117473D
Original Publication Date: 1996-Mar-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 2 page(s) / 71K

Publishing Venue

IBM

Related People

Cobbett, M: AUTHOR [+3]

Abstract

Most Voice Response Unit (VRU) applications are derived from applications responding to digitally entered input (DTMF): a caller would be given a list of options and would make a choice based on a number 'press 1 for Balance Enquiry; 2 for Statement...' etc.. The first installations of voice recognition were, therefore, aimed at DTMF replacement, allowing the caller to use voice input to guide the application either because no DTMF was available, or as a more natural interface.

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

Simultaneous Use of Dual Tone Multi-Frequency and Voice Recognition
in Voice Response Unit Applications

      Most Voice Response Unit (VRU) applications are derived from
applications responding to digitally entered input (DTMF): a caller
would be given a list of options and would make a choice based on a
number 'press 1 for Balance Enquiry; 2 for Statement...' etc..  The
first installations of voice recognition were, therefore, aimed at
DTMF replacement, allowing the caller to use voice input to guide the
application either because no DTMF was available, or as a more
natural interface.

      For current users, and in noisy environments, this may
cause a problem: the voice recognition hitrate may decrease under
such conditions, and callers who have some familiarity with VRU
applications may wish to alternate DTMF and voice input.

      In many cases, the only solution is to prompt the caller in
advance to make a choice between DTMF and voice recognition function.
This results in duplication of applications -- requiring a separate
DTMF and voice recognition version -- and puts additional load on the
caller who is now obliged to make application choices in addition to
menu choices.  Further, having made a choice, the caller is obliged
to continue in that mode, or terminate the call and redial.

      The solution described here integrates the incoming data
streams and underlying application functions (DTMF detection and
voice recognition) in such a way that externally it is transparent to
the caller which mode is used, and there is no limitation on the
number of switches made between those modes.

      At the user level, both DTMF and voice input for recognition
are dealt with in the same way; at the application level, the
application can use either input mode seamlessly.

      For e...