Bio-phonetic multi-phrase speaker identity verification

• US 20070198264A1
• Filed: 04/13/2007
• Published: 08/23/2007
• Est. Priority Date: 02/13/2003
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Inventors: Chang, Hisao
• Granted Patent: US 7,567,901 B2
• Time in Patent Office: Days
• US Class Current: 704/250
• CPC Class Codes: G10L 17/04 Training, enrolment or model building
• CPC Class Codes: G10L 17/14 Use of phonemic categorisation or speech recognition prior to speaker recognition or verification
• CPC Class Codes: G10L 17/22 Interactive procedures; Man-machine interfaces
• CPC Class Codes: G10L 17/24 the user being prompted to utter a password or a predefined phrase
• CPC Class Codes: G10L 17/26 Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices

Abstract

Systems and methods for bio-phonetic multi-phrase speaker identity verification are disclosed. Generally, a speaker identity verification engine generates a dynamic phrase including at least one dynamically-generated word. The speaker identity verification engine prompts a user to speak the dynamic phrase and receives a dynamic phrase utterance. The speaker identity verification engine extracts at least one voice characteristic from the dynamic phrase utterance and compares the at least one voice characteristic with a voice profile the generate a score. The speaker identity verification engine then determines whether to accept a speaker identity claim based on the score.

Claims

1. A method comprising:

generating and prompting a user to speak a first dynamic phrase (FDP), the FDP comprising at least one dynamically-generated word;

receiving an FDP utterance;

extracting at least one voice characteristic from the FDP utterance and comparing the at least one voice characteristic with a voice profile to generate a score; and

determining whether to accept a speaker identity claim based on the score.

2. The method of claim 1, wherein determining whether to accept a speaker identify claim based on the score comprises:

rejecting the speaker identity claim when the score indicates a low confidence level.

3. The method of claim 2, further comprising:

tagging the FDP as being attempted by an imposter.

4. The method of claim 1, wherein determining whether to accept a speaker identity claim based on the score comprises:

accepting the speaker identity claim when the score indicates a high confidence level.

5. The method of claim 1, wherein determining whether to accept a speaker identity claim based on the score comprises:

determining the score indicates a medium confidence level;

generating and prompting the user to speak a second dynamic phrase (SDP), the SDP comprising at least one dynamically-generated word;

receiving a SDP utterance;

extracting at least one voice characteristic from the SDP utterance and comparing the at least one voice characteristic with the voice profile to generate a second score; and

determining whether to accept a speaker identity claim based on the second score.

6. The method of claim 5, wherein determining whether to accept a speaker identity claim based on the second score comprises:

determining whether to accept a speaker identity claim based on the first and second scores.

7. The method of claim 5, further comprising:

updating a voice profile associated with the speaker identity based on a confidence level indicated by the second score.

8. The method of claim 1, further comprising:

modifying a threshold that the score must exceed to indicates a high confidence level.

9. The method of claim 6, wherein the threshold is modified based on an undesirably high level of imposter attempts.

10. The method of claim 1, wherein the FDP is selected from biographical, geographical and time-sensitive transactional information.

11. A computer-readable storage medium comprising a set of instructions to direct a processor to perform acts of:

generating and prompting a user to speak a first dynamic phrase (FDP), the FDP comprising at least one dynamically-generated word;

receiving an FDP utterance;

extracting at least one voice characteristic from the FDP utterance and comparing the at least one voice characteristic to a voice profile to generate a score; and

determining whether to accept a speaker identity claim based on the score.

12. The computer-readable storage medium of claim 11, wherein determining whether to accept a speaker identify claim based on the score comprises:

rejecting the speaker identity claim when the score indicates a low confidence level.

13. The computer-readable storage medium of claim 12, further comprising:

tagging the FDP as being attempted by an imposter.

14. The computer-readable storage medium of claim 11, wherein determining whether to accept a speaker identity claim based on the score comprises:

accepting the speaker identity claim when the score indicates a high confidence level.

15. The computer-readable storage medium of claim 11, wherein determining whether to accept a speaker identity claim based on the score comprises:

determining the score indicates a medium confidence level;

generating and prompting the user to speak a second dynamic phrase (SDP), the SDP comprising at least one dynamically-generated word;

receiving a SDP utterance;

extracting at least one voice characteristic from the SDP utterance and comparing the at least one voice characteristic with the voice profile to generate a second score; and

determining whether to accept a speaker identity claim based on the second score.

16. A system comprising:

a voice profile database comprising a plurality of voice profiles; and

a speaker identity verification engine in communication with the voice profile database, the speaker identity verification engine operative to generate a first dynamic phrase (FDP), the DP comprising at least one dynamically generated word, to prompt a user to speak the FDP, to receive an FDP utterance, to extract at least one voice characteristic from the FDP utterance, to compare the at least one voice characteristic with at least one of the plurality of voice profiles, and to determine whether to accept a speaker identity claim based on the score.

17. The system of claim 16, wherein the speaker identity verification engine is further operative to reject the speaker identity claim when the score indicates a low confidence level.

18. The system of claim 16, wherein the speaker identity verification engine is further operative to accept the speaker identity claim when the score indicates a high confidence level.

19. The system of claim 16, wherein in response to determining the score indicates a medium confidence level, the speaker identity verification engine is further operative to generate a second dynamic phrase (SDP), the SDP comprising at least one dynamically-generated word, to prompt the user to speak the SDP, to receive a SDP utterance, to extract at least one voice characteristic from the SDP utterance, to compare the at least one voice characteristic with the voice profile to generate a second score, and to determine whether to accept the speaker identity claim based on the second score.

20. The system of claim 16, wherein the speaker identity verification engine is further operative to modify a threshold that the score must exceed to indicate a high confidence level based on a number of imposter attempts.