System and method for speaker recognition on mobile devices

• US 10,749,864 B2
• Filed: 01/25/2018
• Issued: 08/18/2020
• Est. Priority Date: 02/24/2012
• Status: Active Grant

• Current Assignee: Cirrus Logic Incorporated
• Original Assignee: Cirrus Logic Incorporated
• Inventors: Garcia Gomar, Marta
• Primary Examiner(s): Goodchild, William J.
• Application Number:
  US15/879,797
• Publication Number:
  US 20180152446A1
• Time in Patent Office: 936 Days
• Field of Search: None
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• CPC Class Codes: G10L 17/22 Interactive procedures; Man-machine interfaces
• 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
• CPC Class Codes: H04W 12/06 Authentication

Abstract

A speaker recognition system for authenticating a mobile device user includes an enrollment and learning software module, a voice biometric authentication software module, and a secure software application. Upon request by a user of the mobile device, the enrollment and learning software module displays text prompts to the user, receives speech utterances from the user, and produces a voice biometric print. The enrollment and training software module determines when a voice biometric print has met at least a quality threshold before storing it on the mobile device. The secure software application prompts a user requiring authentication to repeat an utterance based at least on an attribute of a selected voice biometric print, receives a corresponding utterance, requests the voice biometric authentication software module to verify the identity of the second user using the utterance, and, if the user is authenticated, imports the voice biometric print.

Claims

1. A method of operation of a speaker recognition system, comprising:

receiving, at an input of the speaker recognition system, at least one new voice sample from a user for speaker authentication;

determining a first offset value from scores assigned to previously received voice samples from said user, wherein the first offset value comprises an average of respective offsets of said scores from a mean score value;

determining a second offset value from at least one score assigned to the at least one new voice sample from said user;

determining whether a difference between said second offset value and said first offset value exceeds a proximity threshold; and

if the difference between said second offset value and said first offset value exceeds the proximity threshold, determining that further system training is required, and indicating to the user that said further system training is required.

2. A method as claimed in claim 1, wherein the at least one new voice sample is a most recently received voice sample.

3. A method as claimed in claim 1, wherein the at least one new voice sample comprises a plurality of recently received voice samples including a most recently received voice sample.

4. A method as claimed in claim 3, wherein a number of samples in the plurality of recently received voice samples is configurable.

5. A method as claimed in claim 1, further comprising:

determining whether a number of false rejections in a recent set of received voice samples exceeds a threshold value; and

if the number of false rejections in the recent set of received voice samples exceeds the threshold value, determining that further system training is required.

6. A method as claimed in claim 5, wherein the threshold value is configurable.

7. A method as claimed in claim 5, wherein the number of received voice samples in said recent set of received voice samples is configurable.

8. A speaker recognition system, comprising:

an input for receiving at least one new voice sample from a user for speaker authentication,

wherein the speaker recognition system is configured for performing a method comprising;

receiving at least one new voice sample from the user for speaker authentication;

determining a first offset value from scores assigned to previously received voice samples from said user, wherein the first offset value comprises an average of respective offsets of said scores from a mean score value;

determining a second offset value from at least one score assigned to the at least one new voice sample from said user;

determining whether a difference between said second offset value and said first offset value exceeds a proximity threshold; and

if the difference between said second offset value and said first offset value exceeds the proximity threshold, determining that further system training is required, and indicating to the user that said further system training is required.

9. A computer program product, comprising a non-transitory computer-readable medium, having stored thereon instructions for causing a processor to perform a method comprising:

receiving at least one new voice sample from a user for speaker authentication;

determining a first offset value from scores assigned to previously received voice samples from said user, wherein the first offset value comprises an average of respective offsets of said scores from a mean score value;

determining a second offset value from at least one score assigned to the at least one new voice sample from said user;

determining whether a difference between said second offset value and said first offset value exceeds a proximity threshold; and

if the difference between said second offset value and said first offset value exceeds the proximity threshold, determining that further system training is required, and indicating to the user that said further system training is required.