Method and system for identifying a user by voice

• US 6,978,238 B2
• Filed: 07/12/1999
• Issued: 12/20/2005
• Est. Priority Date: 07/12/1999
• Status: Expired due to Term

• Current Assignee: Charles Schwab & Company Incorporated (Charles Schwab Corporation)
• Original Assignee: Charles Schwab & Company Incorporated (Charles Schwab Corporation)
• Inventors: Wohlsen, Robert C, McNeill, Sue
• Primary Examiner(s): Azad, Abul K.
• Application Number:
  US09/351,723
• Publication Number:
  US 20030125944A1
• Time in Patent Office: 2,353 Days
• Field of Search: 704246-250, 704/273
• US Class Current: 704/246
• CPC Class Codes: G10L 17/24 the user being prompted to utter a password or a predefined phrase

Abstract

A method and apparatus identifies a caller as a user in a group of users. An enrollment process extracts a grammar of the user'"'"'s password and a voiceprint of the user'"'"'s password and name. A caller may identify himself as a user by speaking his name. The name is recognized and a number of users having a name most closely matching the name spoken is identified using voice recognition techniques. The caller then speaks his password and the grammar is identified that most closely matches the grammars of the passwords corresponding to the users identified from the spoken name. A voiceprint is extracted from the name and grammar spoken by the caller, and if the voiceprint matches the voiceprint extracted for the user identified using the grammar during that user'"'"'s enrollment process, the caller is identified as that user.

Claims

1. A method of obtaining a user'"'"'s identity by voice, comprising:

receiving a set of at least one known grammar and a set of at least one known voiceprint corresponding to a plurality of utterances from each of a first plurality of users;

for each of the first plurality of users, associating the set of at least one known grammar and the set of at least one known voiceprint with an identifier of said user;

receiving at least one utterance from a subject user;

performing a voice recognition on at least one of the at least one utterance received from the subject user, said voice recognition being different from extracting a grammar from a first at least one of the at least one utterance received from the subject user;

responsive to the voice recognition technique, selecting from the first plurality of users a second plurality of users, smaller than the first plurality of users by a factor of at least ten, for which the first voice recognition most closely matches at least one selected from the set of at least one grammar and the set of at least one voiceprint associated with the identifiers of the second plurality of users;

from the second plurality of users, selecting the user for which a grammar of the first at least one of the at least one utterance received from the subject user most closely matches at least one of the set of at least one grammar associated with the identifiers of the second plurality of users;

verifying a voiceprint of at least one of the at least one utterance has at least a similarity to the set of at least one voiceprint of the selected user; and

responsive to the verifying step, providing the identifier of the selected user as the identifier of the subject user.

2. The method of claim 1 wherein the voice recognition comprises extracting a grammar from a second at least one of the at least one utterance received from the subject user, the second at least one of the at least one utterance having at least one difference from the first at least one of the at least one utterance.

3. The method of claim 1 wherein the voice recognition technique comprises speaker independent voice recognition.

4. The method of claim 1 wherein the first at least one of the at least one utterance comprises a password.

5. The method of claim 1 wherein a number of the second plurality of users corresponds to a constant.

6. The method of claim 1 wherein the second plurality of users corresponds to users for which the voice recognition technique yields a confidence level exceeding a threshold.

7. The method of claim 1 wherein at least one of the at least one utterance may be other than a real word.

8. A system for obtaining a user'"'"'s identity by voice, comprising:

storage for storing and providing at an output a set of at least one known grammar and a set of at least one known voiceprint corresponding to a plurality, of utterances from each of a first plurality of users, for each of the first plurality of users, the set of at least one known grammar and the set of at least one known voiceprint being associated with an identifier of said user;

a first recognizer having an input operatively coupled for receiving at least one utterance from a subject user, the first recognizer for performing a voice recognition on at least one of the at least one utterance received from the subject user, said voice recognition being different from extracting a grammar from a first at least one of the at least one utterance received from the subject user, the first recognizer additionally for, responsive to the voice recognition technique, selecting from the first plurality of users a second plurality of users, smaller than the first plurality of users by a factor of at least ten, for which the first voice recognition most closely matches at least one selected from the set of at least one grammar and the set of at least one voiceprint associated with the identifiers of the second plurality of users received at a second input coupled to the storage output, and for providing at an output identifiers of the second plurality of users;

a second recognizer having a first input for receiving the identifiers of the second plurality of users, and a second input for receiving at least one of the at least one utterance from the subject user, the second recognizer for extracting a grammar from the at least one of the at least one utterance received at the second second voice recognizer input, and for selecting from the second plurality of users the user for which the grammar extracted most closely matches at least one of the set of at least one grammar associated with the identifiers of the second plurality of users received at a third input coupled to the storage output, and for providing an identifier of the selected user at an output;

a verifier having a first input coupled to the second recognizer output, the verifier for obtaining a voiceprint of at least one of the at least one utterance received at a second input, and for verifying a voiceprint of at least one of the at least one utterance has at least a similarity to the set of at least one voiceprint of the selected user received at a third input coupled to the storage output; and

responsive to said verification, providing at an output the identifier of the selected user as the identifier of the subject user.

9. The system of claim 8 wherein the first recognizer performs the voice recognition by extracting a grammar from a second at least one of the at least one utterance received from the subject user, the second at least one of the at least one utterance having at least one difference from the first at least one of the at least one utterance.

10. The system of claim 8 wherein the first recognizer performs the voice recognition using speaker independent voice recognition.

11. The system of claim 8 wherein the first at least one of the at least one utterance comprises a password.

12. The system of claim 8 wherein a number of the second plurality of users corresponds to a constant.

13. The system of claim 8 wherein the second plurality of users corresponds to users for which the voice recognition performed by the first recognizer yields a confidence level exceeding a threshold.

14. The system of claim 8 wherein at least one of the at least one utterance may be other than a real word.

15. A computer program product comprising a computer useable medium having computer readable program code embodied therein for obtaining a user'"'"'s identity by voice, the computer program product comprising computer readable program code devices configured to cause a computer to:

receive a set of at least one known grammar and a set of at least one known voiceprint corresponding to a plurality of utterances from each of a first plurality of users;

for each of the first plurality of users, associate the set of at least one known grammar and the set of at least one known voiceprint with an identifier of said user;

receive at least one utterance from a subject user;

perform a voice recognition on at least one of the at least one utterance received from the subject user, said voice recognition being different from extracting a grammar from a first at least one of the at least one utterance received from the subject user;

responsive to the voice recognition technique, select from the first plurality of users a second plurality of users, smaller than the first plurality of users by a factor of at least ten, for which the first voice recognition most closely matches at least one selected from the set of at least one grammar and the set of at least one voiceprint associated with the identifiers of the second plurality of users;

from the second plurality of users, select the user for which a grammar of the first at least one of the at least one utterance received from the subject user most closely matches at least one of the set of at least one grammar associated with the identifiers of the second plurality of users;

verify a voiceprint of at least one of the at least one utterance has at least a similarity to the set of at least one voiceprint of the selected user; and

responsive to the computer readable program code devices configured to cause the computer to verify, provide the identifier of the selected user as the identifier of the subject user.

16. The computer program product of claim 15 wherein the computer readable program code devices configured to cause the computer to perform voice recognition comprise computer readable program code devices configured to cause the computer to extract a grammar from a second at least one of the at least one utterance received from the subject user, the second at least one of the at least one utterance having at least one difference from the first at least one of the at least one utterance.

17. The computer program product of claim 15 wherein the computer readable program code devices configured to cause the computer to perform voice recognition comprise computer readable program code devices configured to cause the computer to perform speaker independent voice recognition.

18. The computer program product of claim 15 wherein the first at least one of the at least one utterance comprises a password.

19. The computer program product of claim 15 wherein a number of the second plurality of users corresponds to a constant.

20. The computer program product of claim 15 wherein the second plurality of users corresponds to users for which the voice recognition technique yields a confidence level exceeding a threshold.

21. The computer program product of claim 15 wherein at least one of the at least one utterance may be other than a real word.