Method of comparing voice signals that reduces false alarms

• US 7,650,281 B1
• Filed: 10/11/2006
• Issued: 01/19/2010
• Est. Priority Date: 10/11/2006
• Status: Active Grant

• Current Assignee: National Security Agency
• Original Assignee: US Government As Represented By The Director National Security Agency
• Inventors: Cusmariu, Adolf
• Primary Examiner(s): Chawan; Vijay B
• Application Number:
  US11/545,922
• Time in Patent Office: 1,196 Days
• Field of Search: 704/274, 704/273, 704/243, 704/246, 704/275, 704/251, 704/233, 455/411, 340/825.36, 340/825.49, 340/426.25
• US Class Current: 704/243
• CPC Class Codes: G10L 17/08 Use of distortion metrics or a particular distance between probe pattern and reference templates

Abstract

A method of comparing voice samples to reduce false alarms by receiving a first voice sample, generating a model of the first voice sample, reordering the first voice sample, generating a model of the reordered first voice sample; receiving a second voice sample; generating a model of the second voice sample; reordering the second voice sample, generating a model of the reordered second voice sample, comparing at least one pairings of the models, and determining if the first voice sample matches the second voice sample if the model comparisons are within a user-definable threshold.

Claims

1. A method of comparing voice samples to reduce false alarms:

a) receiving a first voice sample;

b) generating a first model of the first voice sample;

c) reordering the first voice sample using a reordering method selected from the group of reordering methods consisting of time-reversal, segmental rearrangement, and a combination thereof;

d) generating a second model of the result of step (c);

e) receiving a second voice sample;

f) generating a third model of the second voice sample;

g) reordering the second voice sample using a reordering method selected from the group of reordering methods consisting of time-reversal, segmental rearrangement, and a combination thereof;

h) generating a fourth model of the result of step (g);

i) comparing at least one pair of models selected from the group of model pairs consisting of the first model and the third model, the first model and the fourth model, the second model and the third model, and the second model and the fourth model; and

j) determining that the first voice sample matches the second voice sample if the results of step (i) are within a user-definable threshold.

2. The method of claim 1, wherein the step of receiving a first voice sample is comprised of the step of receiving a first voice sample in an analog format.

3. The method of claim 2, further including the step of digitizing the first voice sample.

4. The method of claim 2, further including the step of digitizing the first voice sample.

5. The method of claim 4, wherein the step of receiving a second voice sample is comprised of the step of receiving a first voice sample in an analog format.

6. The method of claim 5, further including the step of digitizing the first voice sample.

7. The method of claim 1, wherein the step of receiving a second voice sample is comprised of the step of receiving a first voice sample in an analog format.

8. A method of comparing voice samples to reduce false alarms:

a) receiving a first voice sample;

b) reordering the first voice sample using a reordering method selected from the group of reordering methods consisting of time-reversal, segmental rearrangement, and a combination thereof;

c) combining the results of step (a) and step (b);

d) generating a first model of the result of step (c);

e) receiving a second voice sample;

f) reordering the second voice sample using a reordering method selected from the group of reordering methods consisting of time-reversal, segmental rearrangement, and a combination thereof;

g) combining the results of step (e) and step (f);

h) generating a second model of the result of step (g);

i) comparing the first model and the second model; and

j) determining that the first voice sample matches the second voice sample if the result of step (i) is within a user-definable threshold.

9. The method of claim 8, wherein the step of receiving a first voice sample is comprised of the step of receiving a first voice sample in an analog format.

10. The method of claim 9, further including the step of digitizing the first voice sample.

11. The method of claim 9, further including the step of digitizing the first voice sample.

12. The method of claim 10, wherein the step of receiving a second voice sample is comprised of the step of receiving a first voice sample in an analog format.

13. The method of claim 8, wherein the step of receiving a second voice sample is comprised of the step of receiving a first voice sample in an analog format.

14. The method of claim 8, wherein the step of combining the results of step (e) and step comprised of the step of appending the result of step (f) to the result of step (e).

15. The method of claim 14, further including the step of digitizing the first voice sample.

16. The method of claim 15, wherein the step of combining the results of step (e) and step (f) is comprised of the step of appending the result of step (f) to the result of step (e).