Voice authentication system having cognitive recall mechanism for password verification

• US 6,671,672 B1
• Filed: 03/30/1999
• Issued: 12/30/2003
• Est. Priority Date: 03/30/1999
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Heck, Larry P.
• Primary Examiner(s): Knepper, David D.
• Application Number:
  US09/281,356
• Time in Patent Office: 1,736 Days
• Field of Search: 704/270-275, 704/246-250, 704/235, 370/352, 375/220, 380/30
• US Class Current: 704/273
• CPC Class Codes: G10L 17/24 the user being prompted to utter a password or a predefined phrase

Abstract

A voice authentication system having a cognitive recall mechanism for password verification is provided. A user is enrolled for password verification by receiving a first voice input from the user representing the password prompt and a second voice input representing a correct response to the password prompt. The first and second voice inputs may be stored as waveforms, as voiceprints, recognized speech data, or a combination thereof. During verification, the identity of the user is verified by outputting the user-provided password prompt and evaluating a response to password prompt against the correct response. Thus, the user is able to select his own password prompt to facilitate cognitive recall of the password during a subsequent verification phase.

Claims

1. A machine-implemented method of enrolling a user for voice authentication, the method comprising:

outputting a first prompt for eliciting a first utterance from the user;

receiving the first utterance;

storing the first utterance as a password prompt for the user;

outputting a second prompt for eliciting a second utterance from the user;

receiving the second utterance; and

storing the second utterance as a password corresponding to the password prompt.

2. A method as recited in claim 1, wherein said storing the first utterance as a password prompt comprises recording the first utterance as one or more waveforms.

3. A method as recited in claim 1, wherein said storing the first utterance as a password prompt comprises:

performing speech recognition on the first utterance to generate a text string; and

storing the text string.

4. A method as recited in claim 1, wherein said storing the second utterance as a password comprises generating a voiceprint from the second utterance.

5. A method as recited in claim 1, wherein said storing the second utterance as a password comprises:

performing speech recognition on the second utterance to generate a text string; and

storing the text string.

6. A machine-implemented method of performing voice authentication, the method comprising:

outputting a password prompt to the user, the password prompt representative of an utterance previously received from the user and corresponding to a password for the user; and

verifying the user based on whether a spoken response to the password prompt is received which matches the password.

7. A method as recited in claim 6, wherein said outputting the password prompt comprises outputting a recording of said utterance previously received from the user.

8. A method as recited in claim 6, further comprising determining whether the spoken response was provided by said user, wherein said verifying comprises establishing a verified condition only if it was determined that the spoken response matches the password and was provided by said user.

9. A method as recited in claim 6, wherein said verifying comprises comparing the spoken response to a voiceprint corresponding to the password.

10. A method as recited in claim 6, wherein said verifying comprises:

performing speech recognition on the spoken response to generate a first text string; and

comparing the first text string to a second text string representing the password.

11. A method of performing voice authentication, the method comprising:

enrolling a user for voice authentication by receiving from the user and storing a first voice input representing a password prompt for the user and a second voice input representing a correct response to the password prompt; and

performing password verification by outputting the password prompt and evaluating a response to the password prompt with respect to the correct response.

12. A method as recited in claim 11, further comprising determining whether the response to the password prompt was provided by said user, wherein said performing verification comprises establishing a verified condition only if it was determined that the response to the password prompt matches the correct response and was provided by said user.

13. A method as recited in claim 11, wherein said storing the first voice input comprises recording the first voice input as one or more waveforms.

14. A method as recited in claim 13, wherein said outputting the password prompt comprises outputting a recording of the first voice input based on said one or more waveforms.

15. A method as recited in claim 11, wherein said storing the first voice input comprises:

performing speech recognition on the first utterance to generate a text string; and

storing the text string.

16. A method as recited in claim 15, wherein said outputting the password prompt comprises:

generating machine-synthesized speech based on the text string; and

outputting the machine-synthesized speech as the password prompt.

17. A method as recited in claim 11, wherein said storing the second voice input comprises generating a voiceprint from the second voice input.

18. A method as recited in claim 11, wherein said storing the second voice input comprises:

performing speech recognition on the second voice input to generate a text string; and

storing the text string.

19. A method of performing voice authentication, the method comprising:

enrolling a user for voice authentication by;

outputting a first prompt to the user for eliciting a first utterance from the user, the first utterance representing a password prompt for the user;

receiving the first utterance;

storing the first utterance;

outputting a second prompt for eliciting a second utterance from the user, the second utterance representing a password corresponding to the password prompt;

receiving the second utterance; and

storing the second utterance; and

performing verification by;

outputting the password prompt; and

establishing a verified condition based on whether a spoken response to the password prompt is received which matches the password.

20. A machine-implemented method of performing voice authentication, the method comprising:

during an enrollment phase;

outputting a first audible prompt for eliciting a first utterance from a user;

receiving the first utterance;

storing the first utterance as a password prompt for the user;

outputting a second audible prompt for eliciting a second utterance from the user;

receiving the second utterance; and

storing the second utterance as a password for the user;

and, during a verification phase;

outputting a third audible prompt to the user, the third audible prompt representative of the password prompt;

determining whether a response to the third audible prompt is received which matches the password;

establishing a verified condition with respect to the user only if it was determined that the response to the third audible prompt matches the password.

21. A method as recited in claim 20, further comprising determining whether the response to the third audible prompt was provided by said user, wherein said establishing a verified condition with respect to the user comprises establishing the verified condition only if it was determined that the response to the third audible prompt matches the password and was provided by said user.

22. A method as recited in claim 20, wherein said storing the first utterance as a password prompt comprises recording the first utterance as one or more waveforms, and wherein said outputting the third audible prompt comprises outputting a recording of the first utterance based on the one or more waveforms.

23. A method as recited in claim 20, wherein said storing the first utterance as a password prompt comprises:

performing speech recognition on the first utterance to generate a text string; and

storing the text string;

and wherein said outputting the third audible prompt comprises;

generating machine-synthesized speech based on the text string; and

outputting the machine-synthesized speech.

24. A method as recited in claim 20, wherein said storing the second utterance as a password comprises generating a voiceprint from the second utterance.

25. A method as recited in claim 20, wherein said storing the second utterance as a password comprises:

performing speech recognition on the second utterance to generate a text string; and

storing the text string.

26. A computer-implemented method of performing voice authentication, the method comprising, in a computer system:

during an enrollment phase,

establishing a telephone connection with a remote telephone unit operated by a user;

transmitting a first audible prompt over the telephone connection, the first audible prompt for eliciting a first utterance from a user;

receiving the first utterance from the user over the telephone connection;

storing the first utterance as a password prompt for the user;

transmitting a second audible prompt over the telephone connection, the second audible prompt for eliciting a second utterance from the user;

receiving the second utterance from the user over the telephone connection; and

storing the second utterance as a password for the user; and

, during a verification phase subsequent to the enrollment phase,

outputting a third audible prompt to the user over the telephone connection, the third audible prompt representative of the password prompt;

determining whether a response to the third audible prompt is received which matches the password;

establishing a verified condition with respect to the user only if it was determined that the response to the third audible prompt matches the password.

27. A method as recited in claim 26, further comprising determining whether the response to the third audible prompt was provided by said user, wherein said establishing a verified condition with respect to the user comprises establishing the verified condition only if it was determined that the response to the third audible prompt matches the password and was provided by said user.

28. A method as recited in claim 26, wherein said storing the first utterance as a password prompt comprises recording the first utterance as one or more waveforms.

29. A method as recited in claim 26, wherein said storing the first utterance as a password prompt comprises:

performing speech recognition on the first utterance to generate a text string; and

storing the text string.

30. A method as recited in claim 26, wherein said storing the second utterance as a password comprises generating a voiceprint from the second utterance.

31. A method as recited in claim 26, wherein said storing the second utterance as a password comprises:

performing speech recognition on the second utterance to generate a text string; and

storing the text string.

32. A device for performing voice authentication, the device comprising:

enrollment means for enrolling a user for voice authentication, the enrollment means comprising;

first output means for outputting a first prompt to the user for eliciting a first utterance from the user, the first utterance representing a password prompt for the user;

first receiving means for receiving the first utterance;

first storage means for storing the first utterance;

second output means for outputting a second prompt for eliciting a second utterance from the user, the second utterance representing a password corresponding to the password prompt; and

second receiving means for receiving the second utterance;

second storage means for storing the second utterance; and

verification means for verifying the user, the verification means comprising;

third output means for outputting the password prompt to the user; and

means for verifying the user based on whether a spoken response to the password prompt is received which matches the password.

33. A computer system comprising:

a processor;

an input/output (I/O) device coupled to the processor; and

a storage device coupled to the processor and having sequences of instructions stored therein which can be executed by the processor to cause the computer system to;

output a first prompt for eliciting a first utterance from a user using the I/O device;

receive the first utterance using the I/O device;

store the first utterance as a password prompt for the user;

output a second prompt for eliciting a second utterance from the user using the I/O device;

receive the second utterance using the I/O device; and

store the second utterance as a password corresponding to the password prompt for verifying the user.

34. A computer system as recited in claim 33, wherein the storage device further has sequences of instructions stored therein which can be executed by the processor to cause the computer system to:

output a third prompt to the user using the I/O device, the third audible prompt representative of the password prompt;

determine whether a response to the third prompt is received which matches the password; and

establish a verified condition with respect to the user only if it was determined that the response to the third prompt matches the password.

35. A computer system as recited in claim 33, wherein the I/O device comprises a first telephone unit for establishing a telephone connection with a remote telephone unit operated by the user, such that the first and second utterances and said response are received by the computer system over the telephone connection.

36. A machine-readable program storage medium tangibly embodying a sequence of instructions executable by the machine to perform a method for performing voice authentication, the method comprising:

during an enrollment phase;

outputting a first audible prompt for eliciting a first utterance from a user;

receiving the first utterance;

storing the first utterance as a password prompt for the user;

outputting a second audible prompt for eliciting a second utterance from the user;

receiving the second utterance; and

storing the second utterance as a password for the user;

and, during a verification phase;

outputting a third audible prompt to the user, the third audible prompt representative of the password prompt;

determining whether a response to the third audible prompt is received which matches the password;

establishing a verified condition with respect to the user only if it was determined that the response to the third audible prompt matches the password.

37. A voice authentication system comprising:

an enroller configured to enroll a user for voice authentication by receiving from the user and storing a first voice input representing a password prompt for the user and a second voice input representing a password response to the password prompt; and

a password verifier configured to output the password prompt and to perform verification based on whether a third voice input representing a response to the password prompt is received which matches the password response.

38. A voice authentication system as recited in claim 37, further comprising:

a feature extractor configured to extract speech features from the first voice input; and

a speech recognizer configured to generate recognized speech based on the extracted features.

39. A voice authentication system as recited in claim 37, further comprising:

a knowledge verifier configured to verify a content of the third voice input; and

a speaker verifier configured to access a voiceprint to verify the content of the third voice input.