Speaker verification system and process

• US 5,677,989 A
• Filed: 04/30/1993
• Issued: 10/14/1997
• Est. Priority Date: 04/30/1993
• Status: Expired due to Term

• Current Assignee: Alcatel-Lucent USA, Inc. (Nokia Corporation)
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Schoeffler, Max S., Rabin, Michael D.
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US08/056,044
• Time in Patent Office: 1,628 Days
• Field of Search: 381/42, 395/2, 395/2.12, 395/2.55, 395/2.59, 395/2.82, 395/2.14
• US Class Current: 704/246
• CPC Class Codes: G07C 9/37 using biometric data, e.g. fingerprints, iris scans or voice recognition
• CPC Class Codes: G10L 17/22 Interactive procedures; Man-machine interfaces

Abstract

An individual'"'"'s speech sample, obtained for the purpose of speaker verification, is used to create a "protected" model of the speech. The protected model, which is stored in a database in association with a personal identifier for that individual, is arranged so that the characteristics of the individual'"'"'s speech cannot be ascertained from the protected model, without access to an encryption key or other secured information stored in the system.

Claims

1. A method for confirming the claimed identity of an individual, said individual having a personal identifier, said method comprising the steps of:

receiving a sample of the individual'"'"'s speech;

creating a protected model of said speech using said sample by transforming said sample using a transformation method such that characteristics of said individual'"'"'s speech cannot be ascertained by accessing said protected speech model, wherein the transformation method is based on at least the personal identifier;

storing said protected model in association with said personal identifier;

in response to a request for identity confirmation, receiving a sample of a claimant'"'"'s speech and a personal identifier provided by said claimant;

retrieving the protected model associated with said personal identifier received from said claimant;

transforming said claimant'"'"'s speech sample with the same transformation method used in said creating step;

performing a comparison to determine if said retrieved model corresponds to said claimant'"'"'s transformed speech sample; and

confirming the identity of said claimant only if said comparison indicates a positive match.

2. A method for verifying the claimed identity of an individual, comprising the steps of:

receiving a speech sample from said individual;

generating a protected model of said speech sample by transforming said sample using a transformation method such that characteristics of said individual'"'"'s speech cannot be recreated by accessing said protected model alone, wherein the transformation method is based on at least data particular to the individual;

storing said protected model;

receiving a request to verify the identity of a claimant claiming to be said individual;

prompting said claimant to input a sample of said claimant'"'"'s speech;

retrieving said protected model;

processing said claimant'"'"'s speech sample with the same transformation method used in said generating step to enable direct comparison between said claimant'"'"'s speech sample and said retrieved protected model; and

verifying the identity of said claimant only if said comparing step indicates a positive match.

3. A method of speaker verification comprising the steps of:

transforming a speech sample received from a user to be verified using a transformation method to generate a protected model that is secured against unauthorized use, wherein the transformation method is based on at least data particular to the user;

processing a speech sample received from a claimant claiming to be said user with said transformation method so that said claimant'"'"'s speech sample can be compared with said protected model received from said user; and

verifying that said claimant is said user if said protected model matches said claimant'"'"'s processed speech sample.

4. A system for confirming the claimed identity of an individual using speech verification, said individual having a personal identifier, said system comprising:

receiving means for receiving an individual'"'"'s speech sample of said individual'"'"'s speech, said individual'"'"'s personal identifier, a claimant'"'"'s speech sample of a claimant'"'"'s speech, and a personal identifier from said claimant;

means for creating a protected model of said individual'"'"'s speech using a transformation method that transforms said individual'"'"'s speech sample such that the characteristics of said individual'"'"'s speech cannot be ascertained by accessing said protected model, wherein the transformation method is based on at least the personal identifier;

means for storing said protected model in association with said individual'"'"'s personal identifier and for retrieving the protected model associated with said personal identifier received from said claimant in response to a request for identity confirmation;

transforming means for transforming said claimant'"'"'s speech sample with said transformation method; and

comparing means for performing a comparison to determine if said retrieved protected model corresponds to said claimant'"'"'s speech sample, and confirming the identity of said claimant only if said comparison indicates a positive match.

5. Apparatus for verifying the claimed identity of an individual, comprising:

means for generating a protected model of a speech sample received from said individual by transforming said speech sample using a transformation method such that characteristics of said individual'"'"'s speech cannot be ascertained by accessing said protected model, wherein the transformation method is based on at least data particular to the individual;

means for storing said protected model;

means for receiving a speech sample from a person claiming to be said individual;

means for transforming said claimant'"'"'s speech sample using said transformation method; and

means for comparing said protected model of said individual'"'"'s transformed speech and said transformed speech sample received from said person.

6. Apparatus for providing speaker verification wherein a stored speech sample is secured from misappropriation, comprising:

means for receiving a speech sample from a user of said apparatus;

means for irreversibly transforming said speech sample using a transformation method to generate a protected speech model which is unique to said user but which cannot be further used for speaker verification without additional information, wherein the transformation method is based on at least data particular to the user;

means for storing said protected speech model for later comparison with a speech sample received from a person claiming to be said user; and

means for receiving and transforming a claimant'"'"'s speech sample using said transformation method.