Voice Authentication System and Method

• US 20150249664A1
• Filed: 09/10/2013
• Published: 09/03/2015
• Est. Priority Date: 09/11/2012
• Status: Active Grant

• Current Assignee: Auraya Pty Limited
• Original Assignee: Auraya Pty Limited
• Inventors: Talhami, Habib Emile, Malegaonkar, Amit Sadanand, Malegaonkar, Renuka Amit, Summerfield, Clive David
• Granted Patent: US 9,491,167 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/683 using metadata automatically derived from the content
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 21/32 using biometric data, e.g. fingerprints, iris scans or voiceprints
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G10L 17/04 Training, enrolment or model building
• CPC Class Codes: H04L 63/0861 using biometrical features, e.g. fingerprint, retina-scan cryptographic mechanisms or cryptographic arrangements for entity authentication using biological data H04L9/3231

Abstract

A method for maintaining a voiceprint utilisable to authenticate a user comprises associating a trainable universal background model with the user. A voiceprint for the user is generated utilising the trainable universal background model and at least one enrolment voice sample provided by the user. The trainable universal background model is subsequently trained with additional voice samples provided by the user post enrolment that satisfy predefined adaptation criteria. The voiceprint is regenerated responsive to a change to the trainable universal background model.

Claims

1. A method for maintaining a voiceprint utilisable to authenticate a user, the method comprising:

associating a trainable universal background model with the user;

generating a voiceprint for the user utilising the trainable universal background model and at least one enrolment voice sample provided by the user;

training the trainable universal background model with additional voice samples provided by the user post enrolment that satisfy predefined adaptation criteria; and

re-generating the voiceprint responsive to a change to the trainable universal background model.

2. A method in accordance with claim 1, wherein the trainable universal background model initially associated with the user corresponds to a seed universal background model which is built from voice samples provided by a cohort of speakers with one or more shared voice characteristics.

3. A method in accordance with any one of the preceding claims, wherein the trainable universal background model is only trained with additional voice samples provided by the user.

4. A method in accordance with claim 3, wherein the additional voice samples are provided by the user as part of one or more authentication determinations which reference the voiceprint.

5. A method in accordance with claim 4, wherein the training step is carried out dynamically as new voice samples are provided for authentication.

6. A method in accordance with claim 4 or 5, wherein one of the adaptation criteria requires that the additional voice samples resulted in a successful authentication of the user.

7. A method in accordance with claim 6, wherein a further adaptation criterion is that the additional voice samples satisfy an audio quality criterion.

8. A method in accordance with claim 7, wherein the audio quality criterion is that an individual equal error rate for the additional voice samples is less than a predefined threshold.

9. A method in accordance with any one of the preceding claims, further comprising continuing to train the trainable universal background model with additional voice samples until a quality measure for at least one of the trainable universal background model and voiceprint has been met.

20. Computer program code comprising at least one instruction which, when executed by a computer, is arranged to implement the method in accordance with any one of claims 1 to 9.

10. A voice authentication system comprising:

a processing module operable to generate a voiceprint for a user, utilising a trainable universal background model stored in association with the user and at least one enrolment voice sample provided by the user;

a biometric trainer operable to train the trainable universal background model with an additional voice sample provided by the user post enrolment that satisfies predefined adaptation criteria, such that the processing module is further operable to regenerate the voiceprint responsive to a change to the trainable universal background model.

11. A voice authentication system in accordance with claim 10, wherein the voiceprint is regenerated utilising the additional voice sample.

12. A voice authentication system in accordance with claim 10 or 11, wherein the trainable universal background model initially associated with the user corresponds to a seed universal background model which is built from voice samples provided by a cohort of speakers with one or more shared voice characteristics.

13. A voice authentication system in accordance with any one of claims 10 to 12, wherein the trainable universal background model is dynamically trained with a plurality of voice samples provided by the user in one or more authentication sessions that reference the voiceprint.

14. A voice authentication system in accordance with claim 13, wherein one of the adaptation criteria requires that the additional voice samples resulted in a successful authentication of the user.

15. A voice authentication system in accordance with claim 14, wherein a further adaptation criterion is that the additional voice samples satisfy an audio quality criterion.

16. A voice authentication system in accordance with claim 15, wherein the audio quality criterion is that an individual equal error rate for the additional voice samples is less than a predefined threshold.

17. A voice authentication system in accordance with any one of claims 10 to 16, wherein the training module is operable to continue to train the trainable universal background model with additional samples until a quality measure for at least one of the trainable universal background model and voiceprint has been met.

18. A voice authentication method, comprising utilising a voice sample provided by a user during an authentication session to dynamically train a universal background model associated with that user, such that the trained universal background model and voice sample provided for authentication are in turn utilised to regenerate a user voiceprint for use in subsequent authentication sessions.

19. A system substantially as hereinbefore described with reference to the accompanying drawings.