System, method and computer program product for verifying an identity using voiced to unvoiced classifiers

• US 7,603,275 B2
• Filed: 10/31/2005
• Issued: 10/13/2009
• Est. Priority Date: 10/31/2005
• Status: Active Grant

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Tavares, Clifford
• Primary Examiner(s): Smits; Talivaldis I
• Assistant Examiner(s): GODBOLD, DOUGLAS
• Application Number:
  US11/264,280
• Publication Number:
  US 20070100620A1
• Time in Patent Office: 1,443 Days
• Field of Search: 704/208, 704/214, 704/232, 704245-250, 704/273
• US Class Current: 704/250
• CPC Class Codes: G10L 17/06 Decision making techniques; Pattern matching strategies

Abstract

Embodiments of a system, method and computer program product for verifying an identity claimed by a claimant using voiced to unvoiced classifiers are described. In accordance with one embodiment, a speech sample from a claimant claiming an identity may be captured. From the speech sample, a ratio of unvoiced frames to a total number of frames in the speech sample may be calculated. An equal error rate value corresponding to the speech sample can then be determined based on the calculated ratio. The determined equal error rate value corresponding to the speech sample may be compared to an equal error rate value associated with the claimed identity in order to select a decision threshold. A match score may be also be generated based on a comparison of the speech sample to a voice sample associated with the claimed identity. A decision whether to accept the identity claim of the claimant can then be made based on a comparison of the match score to the decision threshold.

Claims

1. A method, comprising:

using an audio pick up device to capture a speech sample from a claimant claiming an identity; and

using computer software embodied on a computer readable medium to perform the following steps;

calculating a ratio of unvoiced frames to a total number of frames in the speech sample,

determining an equal error rate value corresponding to the speech sample based on the calculated ratio,

comparing the determined equal error rate value corresponding to the speech sample to an equal error rate value associated with the claimed identity to select a decision threshold,

generating a match score based on a comparison of the speech sample to a voice sample associated with the claimed identity, and

deciding whether to accept the identity claim of the claimant based on a comparison of the match score to the decision threshold.

2. The method of claim 1, wherein the speech sample is processed at least to identify voiced and unvoiced frames in the speech sample before the calculating of the ratio.

3. The method of claim 1, wherein the determining of the equal error rate value corresponding to the speech sample includes using a function that describes a relationship between equal error rate values and ratios of unvoiced frames to a total number of frames in a population.

4. The method of claim 1, wherein the equal error rate value associated with the claimed identity is based on a ratio of unvoiced frames to a total number of frames in the voice sample associated with the claimed identity and the function that describes a relationship between equal error rate values and ratios of unvoiced frames to a total number of frames in a population.

5. The method of claim 1, wherein the match score is modified with a penalty prior to deciding whether to accept the identity claim.

6. The method of claim 5, wherein the penalty is based on a standard deviation associated with the voice sample of the claimed identity.

7. The method of claim 5, wherein the penalty is based on a difference between a standard deviation associated with the voice sample of the claimed identity and a ratio of unvoiced frames to the total number of frames in the voice sample associated with the claimed identity.

8. The method of claim 7, wherein the standard deviation associated with the voice sample of the claimed identity is based on the ratio of unvoiced frames to the total number of frames in the voice sample associated with the claimed identity and a function that describes a relationship between standard deviations and ratios of unvoiced frames to the total number of frames in a population.

9. The method of claim 7, wherein the standard deviation and the ratio of unvoiced frames to the total number of frames in the voice sample associated with the claimed identity are stored with the voice sample of the identity.

10. The method of claim 1, wherein if the equal error rate value corresponding to the speech sample is selected as the decision threshold, calculating a standard deviation of the determined ratio of unvoiced frames for the speech sample.

11. The method of claim 10, wherein the standard deviation associated with the speech sample is based on the ratio of unvoiced frames to the total number of frames in the speech sample and a function that describes a relationship between standard deviations and ratios of unvoiced frames to the total number of frames in a population.

12. A system, comprising:

an audio pick up device for capturing a speech sample from a claimant claiming a computer readable medium; and

a central processing unit interacting with the computer readable medium to perform the following steps;

a component, implemented in a computer, capable of;

calculating a ratio of unvoiced frames to a total number of frames in the speech sample,

determining an equal error rate value corresponding to the speech sample based on the calculated ratio,

comparing the determined equal error rate value corresponding to the speech sample to an equal error rate value associated with the claimed identity to select a decision threshold,

generating a match score based on a comparison of the speech sample to a voice sample associated with the claimed identity, and

deciding whether to accept the identity claim of the claimant based on a comparison of the match score to the decision threshold.

13. The system of claim 12, wherein the determining of the equal error rate value corresponding to the speech sample includes using a function that describes a relationship between equal error rate values and ratios of unvoiced frames to a total number of frames in a population.

14. The system of claim 12, wherein the equal error rate value associated with the claimed identity is based on a ratio of unvoiced frames to a total number of frames in the voice sample associated with the claimed identity and the function that describes a relationship between equal error rate values and ratios of unvoiced frames to a total number of frames in a population.

15. The system of claim 12, wherein the match score is modified with a penalty prior to deciding whether to accept the identity claim.

16. The system of claim 15, wherein the penalty is based on a difference between a standard deviation associated with the voice sample of the claimed identity and a ratio of unvoiced frames to the total number of frames in the voice sample associated with the claimed identity.

17. The system of claim 16, wherein the standard deviation associated with the voice sample of the claimed identity is based on the ratio of unvoiced frames to the total number of frames in the voice sample associated with the claimed identity and a function that describes a relationship between standard deviations and ratios of unvoiced frames to the total number of frames in a population.

18. The system of claim 17, wherein the standard deviation and the ratio of unvoiced frames to the total number of frames in the voice sample associated with the claimed identity are stored with the voice sample of the identity in a database residing in a memory device.

19. The system of claim 12, wherein if the equal error rate value corresponding to the speech sample is selected as the decision threshold, calculating a standard deviation of the determined ratio of unvoiced frames for the speech sample based on the ratio of unvoiced frames to the total number of frames in the speech sample and a function that describes a relationship between standard deviations and ratios of unvoiced frames to the total number of frames in a population.

20. A computer readable medium encoded with computer code that performs the following steps:

capturing a speech sample from a claimant claiming an identity;

calculating a ratio of unvoiced frames to a total number of frames in the speech sample;

determining an equal error rate value corresponding to the speech sample based on the calculated ratio;

comparing the determined equal error rate value corresponding to the speech sample to an equal error rate value associated with the claimed identity to select a decision threshold;

generating a match score based on a comparison of the speech sample to a voice sample associated with the claimed identity; and

deciding whether to accept the identity claim of the claimant based on a comparison of the match score to the decision threshold.