Identifying contact center agents based upon biometric characteristics of an agent's speech

• US 9,247,056 B2
• Filed: 03/01/2007
• Issued: 01/26/2016
• Est. Priority Date: 02/28/2007
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Mandalia, Baiju D., Moore, Victor S., Nusbickel, Wendi L.
• Primary Examiner(s): Tsang, Fan
• Assistant Examiner(s): BEZUAYEHU, SOLOMON G
• Application Number:
  US11/680,908
• Publication Number:
  US 20080205624A1
• Time in Patent Office: 3,253 Days
• Field of Search: 37926501-26514, 37926601-26604, 379 8801- 8819
• US Class Current: 1/1
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• CPC Class Codes: H04M 2201/41 using speaker recognition speaker recognition per se G10L17/00
• CPC Class Codes: H04M 3/51 Centralised call answering arrangements requiring operator intervention , e.g. call or contact centers for telemarketing

Abstract

The present invention discloses a contact center with speaker identification and verification (SIV) capabilities. In the invention, a set of contact center components can provide automated interactive communications with callers, can provide queue management for callers waiting to communicate with live agents, and can provide skills based routing for assigning live agents to callers. The SIV component can analyze speech utterances to determine a speaker identify based upon biometric characteristics of the analyzed speech utterances. Additionally, the SIV component can process speech from contact center sessions. In one embodiment, the SIV component can prevent agent substitutions from occurring of which the call center is unaware. The SIV component can also be used to distinguish whether communication session content was spoken by a contact center agent or a caller.

Claims

1. A contact center comprising:

the contact center configured to provide automated interactive communications with callers, providing queue management for callers waiting to communicate with live human agents, and providing skills based routing for assigning live human agents to callers; and

a speech identification and verification (SIV) component configured to analyze speech utterances and configured to determine a speaker identity based upon biometric characteristics of the analyzed speech utterances, wherein said SIV component processes speech from contact center sessions that are each between a caller and a human agent to automatically identify the human agent of the contact center involved in each of the contact center sessions based on the biometric characteristics of the speech utterances of that human agent, wherein said SIV component determines whether the human agent that logged into the contact center with a user identifier uniquely assigned to the human agent is indeed the human agent involved in the contact center sessions.

2. The contact center of claim 1, wherein a use of the SIV component prevents agent substitutions from occurring of which the call center is unaware, wherein substitutions for which prevention occurs are each a substation where a human who is not the human agent substitute for the human agent by using the user identifier and password of the human agent.

3. The contact center of claim 1, further comprising:

speech processing components that automatically convert the speech utterances to text, which is further processed by software stored in a machine readable medium, wherein results of the further processing is used by the contact center, wherein the speech identification and verification (SIV) component is used to distinguish which of the human agent and the caller spoke particular ones of the analyzed speech utterances during each one of the contact center sessions, wherein the converted text is linked to a speaker of content of the converted text based upon results produced by the SIV component.

4. The contact center of claim 1, further comprising:

an agent desktop comprising an HTTP browser and a SIP based phone each based upon open standard, wherein the agent desktop is configured to operate as an interface between the contact center agents and the contact center.

5. The contact center of claim 1, further comprising:

a portal server configured to provide an agent portal, through which the contact center agents communicate, said agent portal including a plurality of agent portlets within which contact center information is presented; and

a standards based Web browser within which the agent portal is presented, wherein the Web browser is part of an agent desktop that is configured to operate as an interface between the contact center agents and the contact center.

6. The contact center of claim 1, further comprising:

a plurality of contact center components conforming to open standards that intercommunicate utilizing open standards, said components together forming the contact center.

7. The contact center of claim 6, wherein each of the contact center components execute within a middleware solution having IP Multimedia Subsystem (IMS) capabilities.

8. The contact center of claim 6, wherein each of the contact center components are service oriented architecture (S OA) components.

9. A method for contact centers to identify contact center agents based upon voice characteristics of the human agents comprising:

a human agent logging onto a contact center and providing authentication information that includes a user name unique to that human agent and a corresponding password for the user name;

the contact center authenticating the agent using the authentication information;

the call center transferring a caller to the human agent to initiate a contact center communication session between the human agent and the caller;

receiving speech content associated with the contact center communication session;

extracting biometric characteristics contained within the speech content of the contact center communication session;

comparing the extracted biometric characteristics against previously stored biometric characteristics associated with the human agent;

determining an identity of a speaker of the content based upon results of the comparing step;

comparing the identity of the speaker with an identity of a human associated with the user name, wherein the comparison is performed to verify that a human logged in as the human agent via the user name is in fact the speaker;

a contact center performing at least one programmatic action based upon results of the determined identity, wherein the programmatic action is determining whether inappropriate phrases spoken during the communication session were attributable to the human agent or to the caller and taking corrective or punishment actions against the human agent when the human agent is determined to have spoken the inappropriate phrases.

10. The method of claim 9, wherein said contact center communication session includes speech of the human agent and a contact center customer, and wherein said received speech content is a segment of speech provided by the human agent and the contact center customer, said determining step further comprising:

determining utilizing a software program stored in a machine readable memory which one of the human agent and the contact center customer provided said speech content, wherein said determined identity is one of the human agent and the contact center customer.

11. The method of claim 10, further comprising:

speech-to-text converting said received speech content;

matching converted text with a previously defined entry in a set of watched words; and

based upon said matching step, triggering the determining step, wherein said programmatic action is at least one of an agent monitoring action, a supervisor notification action, a supervisor conferencing action, and a programmatic action that changes a performance rating associated with the contact center agent.

12. The method of claim 10, further comprising:

automatically transcribing the speech of the contact center communication session, wherein different speakers are indicated based upon results of the determining step.

13. The method of claim 9, wherein the programmatic action is an agent verification action, said method further comprising:

verifying that the identity matches an identity of a contact agent to whom the contact center directed a caller for live interactions during the communication session.

14. The method of claim 9, further comprising:

automatically converting the speech content to text; and

adding speaker identifying text to the converted text that indicates which of the contact center agent and a caller provided the associated converted text, wherein the added speaker identifying text is based on results of the determining step.

15. The method of claim 9, wherein the programmatic action verifies whether or not the a human logged in as the human agent via the user name is in fact the human agent or is a different human substituting for the human agent.

16. The method of claim 9, wherein the steps of claim 9 are performed by a plurality of contact center components conforming to open standards that intercommunicate utilizing open standards, said components together forming the open contact center, said contact center providing automated interactive communications with callers, providing queue management for callers waiting to communicate with call center agents, providing skills based routing for assigning live agents to callers based upon skills of the live agents and skills needed by the callers, and providing tooling for provisioning and monitoring the contact center agents.

17. The method of claim 9, wherein the programmatic action determines whether inappropriate phrases spoken during the communication session were attributable to the human agent or where attributable to the caller, wherein the programmatic action takes corrective or punishment actions against the human agent when the human agent is determined to have spoken the inappropriate phrases, and wherein the program action does not take corrective or punishment actions against the human caller when the caller is determined to have spoken the inappropriate phrases.

18. A contact center comprising:

an agent desktop comprising a standard HTTP browser and a standard phone, wherein the agent desktop is configured to permit a human agent to handle call center communications;

a portal server configured to provide an agent portal, through which the human agent utilizing the agent desktop communicates, said agent portal including a plurality of agent portlets within which contact center information is presented, wherein a speech identification and verification (SIV) component determines whether the human agent that logged into the contact center with a user identifier uniquely assigned to the human agent is indeed the human agent involved in the contact center sessions;

an application server executing contact center applications configured to receive input for the contact center applications from the agent portlets and to present contact center output through the agent portlets; and

the SIV component configured to analyze speech utterances and configured to determine a speaker identity based upon biometric characteristics of the analyzed speech utterances, wherein said SIV component processes speech from contact center sessions to automatically identify whether a segment of speech content spoken during a contact center session between a caller and a human using the agent desktop, was spoken by the human agent of the contact center or not based upon comparing biometric characteristics of the segment of speech content with previously stored biometric characteristics of the human agent, wherein said SIV component determines whether the human agent that logged into the contact center is the human agent involved in the contact center sessions.

19. The contact center of claim 18, wherein communications between the agent desktop and that contact center conform to open standards, said open standards consisting of a hypertext transfer protocol (HTTP) based standard and a Session Initiation Protocol (SIP) based standard.

20. The contact center of claim 18, wherein the software and hardware of the contact center is based upon industry standards, wherein the interfaces and communication protocols for the contact center use open standards, and wherein the contact center has IP Multimedia Subsystem (IMS) capabilities.