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

INTERACTING WITH DUAL-TONE MULTI-FREQUENCY-BASED SYSTEMS USING A VOICE ARTIFICIAL INTELLIGENCE

IP.com Disclosure Number: IPCOM000249935D
Publication Date: 2017-May-04
Document File: 6 page(s) / 702K

Publishing Venue

The IP.com Prior Art Database

Related People

Vikas Vashisht: AUTHOR [+2]

Abstract

Techniques presented herein assist a user on a call with sending dual-tone multi-frequency (DTMF) signals without necessarily having to use a DTMF keypad to enter the digits. The system may determine user intent and respond accordingly instead of forcing the user to speak/announce the digits. The user may be guided to take additional actions to help the user enter digits when on a call via a device without a keypad.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Copyright 2017 Cisco Systems, Inc. 1

INTERACTING WITH DUAL-TONE MULTI-FREQUENCY-BASED SYSTEMS USING A VOICE ARTIFICIAL INTELLIGENCE

AUTHORS: Vikas Vashisht Keith Griffin

CISCO SYSTEMS, INC.

ABSTRACT

Techniques presented herein assist a user on a call with sending dual-tone multi-

frequency (DTMF) signals without necessarily having to use a DTMF keypad to enter the

digits. The system may determine user intent and respond accordingly instead of forcing

the user to speak/announce the digits. The user may be guided to take additional actions to

help the user enter digits when on a call via a device without a keypad.

DETAILED DESCRIPTION

When working with collaboration voice artificial intelligence (AI), a user is

commonly requested to input a digit via a keypad / dial pad. This is especially true for

existing collaboration systems. For example, during the join flow of a meeting, these

systems can request a user personal identification number (PIN), security code, meeting

identification (ID), or other input that is intended to be transmitted via dual-tone multi-

frequency (DTMF) signals. However, most voice AI systems are not equipped with a

keypad. As such, described herein are techniques that address the need for DTMF

interaction in order to help the user send the DTMF responses using voice AI.

Automated speech recognition (ASR) and natural language processing (NLP) are

used to understand when a keypad interaction is requested in a voice AI session. The voice

AI may use ASR, NLP, and/or Text-to-Speech (TTS) technology to recognize the need for

interaction with the keypad. The voice AI may also interact with the user to assist the user

with gathering the requested information, transcribing the digits to appropriate DTMF

signals/digits, and/or transmitting the DTMF signals. Figure 1 below is a block diagram

illustrating an example embodiment for implementing this solution.

Copyright 2017 Cisco Systems, Inc. 2

Figure 1

An example method is provided as follows.

STEP 1: The voice AI uses ASR and NLP to analyze a far end audio stream to determine

and identify whether a DTMF is being requested.

STEP 2: If the voice AI recognizes a user needs to enter DTMF digits, a message may be

played for the user using TTS explaining a possible set of responses/actions that the user

can take.

STEP 3: User speech (near end audio) is analyzed by the ASR/NLP module to determine

the user intent/response.

STEP 4: User intent/responses are recognized and mapped/transcribed to the DTMF digits

required to be sent to the far end.

STEP 5: The DTMF digits are sent to the far end using a DTMF generator.

Copyright 2017 Cisco Systems, Inc. 3

Four example use cases are provided as follows.

Use Case 1: Spoken Digits to DTMF Signals

In this example use case, the user is prompted by the far end to enter the meeting

ID followed by “#” (i.e., the pound/hashtag symbol). The voice AI recognizes that the user

is expected to enter the DTMF digits and waits for the use...