Browse Prior Art Database

Using Speech Recognition Voice Signature Input to Take Specific Actions Based on Specific Inbound Callers on Voice Channel

IP.com Disclosure Number: IPCOM000030339D
Original Publication Date: 2004-Aug-06
Included in the Prior Art Database: 2004-Aug-06
Document File: 1 page(s) / 32K

Publishing Venue

IBM

Abstract

By using an adjunct speech recognition system that is always listening to the conversations, when a caller is recognized by their voice-signature, specialized actions can be taken. The system keeps all calls in cached memory during the duration of the call and speech recognition analysis, discarded when the call is completed (although other systems may be monitoring / recording these calls as well for background, detailed analysis or other purposes). Today, special handling in call centers is typically performed by asking the "special" group to call a different number or identify themselves through input of some sort of PIN or other data on the interface - these take time, reduce customer satisfaction and create unnecessary complexity. In addition, prank callers or other types of events can only be managed in a manual, responsive way. In other words, an agent recognizes the specific individual's voice and after the call tells the administrator they received another prank call from a different source which the administrator must block - the automated system would perform these function behind the scenes quickly and unobtrusively.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 65% of the total text.

Page 1 of 1

Using Speech Recognition Voice Signature Input to Take Specific Actions Based on Specific Inbound Callers on Voice Channel

This idea proposes an attached speech recognition system that caches and records in temporary memory 100% of the calls. The system evaluates the speech recognition data against known voice signatures and attempts to unobtrusively identify the end user. If a specified match occurs to a known identity, the system executes a specified action or attaches relevant data (the identify of the caller) to the call data. This allows for real-time actions to be taken, automatically adjusting the call-handling parameters based on the identity of the caller.

This concept has wide-spread potential use: inside call centers, office environments, law enforcement, etc. and can be provided as an interface to any voice/audio communication channel (telephony, voice over IP, standard PC microphone / chat)

Examples / specifics:

When we recognize that it's not just anybody that says "I have a software problem" but it is the CEO of the company that says "I have a software problem" we want to handle him differently and may let the agent know (or send to a different agent group). If a high level execs is recognized, we set the call or priority based on the employee's "job importance" or other HR data in the back-end system. It obviously costs more in terms of lost time/revenue for a high level manager, director, or other exec to wait in a queue to have a problem fixed t...