VERIFICATION OF USER PRESENCE DURING AN INTERACTIVE VOICE RESPONSE SYSTEM SESSION

• US 20110051907A1
• Filed: 08/26/2009
• Published: 03/03/2011
• Est. Priority Date: 08/26/2009
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Jaiswal, Peeyush, Daye, Sheri G., Wang, Fang
• Granted Patent: US 8,130,915 B2
• Time in Patent Office: Days
• US Class Current: 379/88.02
• CPC Class Codes: G10L 17/04 Training, enrolment or model building
• CPC Class Codes: G10L 17/22 Interactive procedures; Man-machine interfaces
• CPC Class Codes: H04M 2201/41 using speaker recognition speaker recognition per se G10L17/00
• CPC Class Codes: H04M 3/4936 Speech interaction details speech recognition per se G10L15/00

Abstract

In those transactions requiring the approval of two or more users, users can perform verification of identity and then completion of desired transaction using an IVR system. The combination of two or more user voice prints previously collected during an enrollment phase can be used for verification of parties to the transaction. During a subsequent verification phase of the IVR system, the users can speak their own password phase when prompted to do so during a defined period of time and the IVR system captures them in a captured mixed voice print. Verification can then be provided by comparing the captured mixed voice print with a joined voice reference model built from stored voice reference models.

Claims

1. A method of verification of user presence during an interactive voice response (IVR) system session, comprising:

during an enrollment phase of an interactive voice response (IVR) system;

a first user and a second user providing a first voice print of the first user and a second voice print of the second user to the IVR system, wherein the first user and a second user are authorized users of a joint account of the IVR system;

the IVR system capturing the first and second voice prints and creating a first voice reference model of the first user from the first voice print and a second voice reference model of the second user from the second voice print;

the IVR system storing the first and second voice reference models;

the IVR system receiving a request to conduct a transaction of the joint account that requires presence of both the first and second users during the transaction to complete the transaction;

in response to the IVR system receiving a request to conduct a transaction of the joint account, during a verification phase;

the IVR system requesting that the first user speak a first password of the first user during a defined period of time and that the second user speak a second password of the second user also during the defined period of time;

during the defined period of time, the IVR system receiving the first password from the first user and the second password from the second user and creating a mixed voice print of the first and second passwords received from the first and second user, respectively;

the IVR system determining which of the first and second users spoke first at a first time stamp during the mixed voice print and when the other user of the first and second users spoke second at a second time stamp during the mixed voice print;

the IVR system creating a joined voice reference model of the first and second users from the first and second voice reference models created during the enrollment phase, comprised at a first time stamp of the joined voice reference model the voice reference model of the user determined to have spoken first and at a second time stamp of the voice reference model of the user determined to have spoken second and wherein the first and second time stamps of the joined voice reference model correspond in time of the joined voice reference model to the first and second time stamps of the mixed voice print;

the IVR system comparing the mixed voice print to the joined voice reference model;

the IVR system verifying presence of the first and second users and permitting the transaction to proceed when the mixed voice print matches the joined voice reference model.

2. The method of claim 1, wherein the first user provides the first voice print by speaking the first password of the first user and wherein the second user provides the second voice print by speaking the second password of the second user.

3. The method of claim 1, wherein the request to conduct a transaction of the joint account is initiated by one or more of the first and second users calling to the IVR system and referencing the joint account.

4. The method of claim 1, wherein the first and second calls received by the IVR system overlap in time.

5. The method of claim 1, wherein the IVR system requests that the first user speak the first password and that the second user speak the second password of the second user during the defined period of time by prompting the first and second users with a prompt indicator.

6. The method of claim 5, wherein the prompt indicator is a beep.

7. The method of claim 1, wherein during the defined period of time, the first and second calls of the first and second users, respectively, initiate the verification phase by a conference call to the IVR system.

8. The method of claim 1, wherein the IVR system uses the first and second voice reference models created during the enrollment phase to determine which of the first and second users spoke first during the mixed voice print and when the other user of the first and second users spoke second during the mixed voice print.

9. The method of claim 1, wherein the joined voice reference model is a dynamic model specific to the first and second time stamps of the spoken first and second passwords supplied by the first and second users, respectively, during the defined period of time of the mixed voice print.

10. The method of claim 1, wherein during the enrollment phase the first and second users provide the first and second voice prints to the IVR system independently.

11. An interactive voice response (IVR) system, comprising:

a feature extraction element that receives a first voice print of a first user and a second voice print of a second user, wherein the feature extraction element receives the first and second voice prints during an enrollment phase of the IVR system, wherein the feature extraction element receives during a verification phase of the IVR system a call from the first and second users during which a joint account of the first and second users is identified and the first and second users request to conduct a transaction of the joint account that requires the presence of the first and second users to complete the transaction, and wherein during the verification phase the feature extraction element receives a first password from the first user during a defined period of time and a second password from the second user during the defined period of time;

a model building element that receives from the feature extraction element the first and second voice prints of the first and second users and creates first and second voice reference models of the first and second users, respectively;

a storage element that stores the first and second voice reference models;

a joint model building element that receives during the verification phase the first and second passwords from the first and second users, creates a mixed voice print, determines which of the first and second users spoke first at a first time stamp during the mixed voice print and when the other user of the first and second users spoke second at a second time stamp during the mixed voice print, and creates a joined voice reference model of the first and second users from the first and second voice reference models created during the enrollment phase, wherein at a first time stamp of the joined voice reference model has the voice reference model of the user determined to have spoken first and at a second time stamp of the voice reference model of the user determined to have spoken second and wherein the first and second time stamps of the joined voice reference model correspond in time of the joined voice reference model to the first and second time stamps of the mixed voice print;

a model comparison element that during the verification phase compares the mixed voice print with the joined voice reference model,

wherein when the mixed voice print matches the joined voice reference model, the IVR system verifies the presence of the first and second users and permits the transaction to proceed.

12. The system of claim 11, wherein the first and second users are authorized users of a joint account of the IVR system.

13. The system of claim 11, wherein the first user provides the first voice print by speaking the first password of the first user and wherein the second user provides the second voice print by speaking the second password of the second user.

14. The system of claim 11, wherein the first and second users request to conduct a transaction of the joint account that requires the presence of the first and second users to complete the transaction is initiated by one or more of the first and second users calling to the IVR system and referencing the joint account during the verification phase.

15. The system of claim 14, wherein the feature extraction element requests that the first user speak the first password and that the second user speak the second password of the second user during the defined period of time by prompting the first and second users with a prompt indicator.

16. The system of claim 15, wherein the prompt indicator is a beep.

17. The system of claim 11, wherein the joined voice reference model is a dynamic model specific to the first and second time stamps of the spoken first and second passwords supplied by the first and second users, respectively, during the defined period of time of the mixed voice print.