Simultaneous speaker-independent voice recognition and verification over a telephone network

• US 5,127,043 A
• Filed: 05/15/1990
• Issued: 06/30/1992
• Est. Priority Date: 05/15/1990
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: VCS Industries, Inc. (Koninklijke Philips N.V.)
• Inventors: Schalk, Thomas B., Hunt, Alan K.
• Primary Examiner(s): BROWN, THOMAS
• Application Number:
  US07/523,486
• Time in Patent Office: 777 Days
• Field of Search: 381/42, 381/43, 379/88, 379/91, 379/189, 379/199, 379/89, 379/188
• US Class Current: 379/88.02
• CPC Class Codes: G06F 21/32 using biometric data, e.g. fingerprints, iris scans or voiceprints
• CPC Class Codes: G06F 21/40 by quorum, i.e. whereby two or more security principals are required
• CPC Class Codes: G06F 21/83 input devices, e.g. keyboards, mice or controllers thereof
• CPC Class Codes: G06F 2221/2105 Dual mode as a secondary aspect
• CPC Class Codes: G06F 2221/2113 Multi-level security, e.g. mandatory access control
• CPC Class Codes: G06F 2221/2117 User registration
• CPC Class Codes: G07C 9/33 by means of a password
• CPC Class Codes: G07C 9/37 using biometric data, e.g. fingerprints, iris scans or voice recognition
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• 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: H04M 2201/40 using speech recognition speech recognition per se G10L15/00
• CPC Class Codes: H04M 3/382 using authorisation codes or passwords

Abstract

The present invention describes a system and method for enabling a caller to obtain access to services via a telephone network by entering a spoken password having a plurality of digits. Preferably, the method includes the steps of: (1) prompting the caller to speak the password beginning with a first digit and ending with a last digit thereof, (2) recognizing each spoken digit of the password using a speaker-independent voice recognition algorithm, (3) following entry of the last digit of the password, determining whether the password is valid, and (4) if the password is valid, verifying the caller'"'"'s identity using a voice verification algorithm.

Claims

1. A system for enabling an authorized user to obtain access to one or more services via a telephone network by speaking a password having a plurality of digits, comprising:

a digital processor;

storage means connected to the digital processor for storing;

voice recognition feature transformation data and voice recognition class reference data derived from a first plurality of training speakers over a telephone network;

voice verification feature transformation data derived from a second plurality of training speakers over a telephone network;

voice verification class reference data comprising data derived from users authorized to access the services;

prompt means controlled by the digital process for prompting a caller to speak a password beginning with a first digit and ending with a last digit thereof;

speech processing means controlled by the digital processor for receiving each digit and generating speech feature data;

voice recognition means controlled by the digital processor, comprising;

transformation means for receiving the speech feature data generated for each digit and the voice recognition feature transformation data and in response thereto generating voice recognition parameter data for

each digit;

decision means connected to the transformation means for receiving the voice recognition parameter data and the voice recognition class reference data and in response thereto generating an output indicating the digit; and

password validation means responsive to entry of the last digit of the password for determining if the password is valid; and

voice verification means controlled by the digital processor and responsive to a determination that the password is valid for determining whether the caller is an authorized user, comprising;

transformation means for receiving the speech feature data generated for each digit and the voice recognition feature transformation data and in response thereto generating voice verification parameter data for each digit; and

verifier means connected to the transformation means of the voice verification means for receiving the voice verification parameter data and the voice verification class reference data and in response thereto generating an output indicating whether the caller is an authorized user.

2. A method for enabling a caller to obtain access to one or more services via a telephone network by speaking a password having a plurality of digits, comprising the steps of:

prompting the caller to speak the password beginning with a first digit and ending with a last digit thereof;

generating speech feature data for each spoken digit;

applying the speech feature data and voice recognition feature transformation data to a voice recognition feature transformation to generate a first set of parameters for each spoken digit, the first set of parameters for use in a voice recognition system;

applying the speech feature data and voice verification feature transformation data to a voice verification feature transformation to generate a second set of parameters for each spoken digit, the second set of parameters for use in a voice verification system;

recognizing each spoken digit of the password using the first set of parameters;

following entry of the last digit of the password, determining whether the password is valid; and

if the password is valid, verifying the caller'"'"'s identity using the second set of parameters.

8. The method as described in claim 2 wherein the voice recognition feature transformation data is derived from a first plurality of training speakers over the telephone network and the voice verification feature transformation data is derived from a second plurality of training speakers over the telephone network.

3. A method for enabling a caller to obtain access to one or more services via a telephone network by speaking a password having a plurality of digits, comprising the steps of:

prompting the caller to speak the password beginning with a first digit and ending with a last digit thereof;

recognizing each spoken digit of the password by;

continuously generating a set of primary features from a received speech signal representing the spoken digit;

capturing a speech utterance from the set of primary features;

generating a set of secondary features from the captured speech utterance;

segmenting the captured speech utterance into phonetic units;

generating a set of tertiary features from the secondary features within the phonetic units;

transforming the tertiary features for each phonetic unit by voice recognition feature transformation data to generate voice recognition parameters data therefor; and

comparing the voice recognition parameter data to predetermined voice recognition class reference data to determine the spoken digit;

following entry of the last digit of the password, determining whether the password is valid; and

if the password is valid, verifying the caller'"'"'s identity using a voice verification system.

4. The method as described in claim 3 wherein the voice recognition feature transformation data and the voice recognition class reference data are generated in an off-line process from a training database of a plurality of training speakers derived over a telephone network.

5. The method as described in claim 3 wherein the step of verifying the caller'"'"'s identity includes the steps of:

(a) transforming the tertiary features for each phonetic unit by voice verification feature data to generate voice verification parameter data therefor;

(b) comparing the voice verification parameter data to predetermined voice verification class reference data for each spoken digit;

(c) computing a weighted Euclidean distance between the voice verification parameter data and the voice verification class reference data for each spoken digit;

(d) repeating steps (b)-(c) for each spoken digit;

(e) computing an ensemble distance for the entire password from the weighted Euclidean distances generated for each spoken digit;

(f) comparing the ensemble distance to predetermined thresholds to verify the caller'"'"'s identity.

6. The method as described in claim 5 wherein the step of comparing the ensemble distance includes the steps of:

(g) determining whether the ensemble distance is less than a first threshold;

(h) if the ensemble distance is less than the first threshold, accepting the caller'"'"'s identity;

(i) if the ensemble distance is greater than the first threshold, determining whether the ensemble distance is greater than a second threshold; and

(j) if the ensemble distance is greater than the second threshold, rejecting the caller'"'"'s identity.

7. The method as described in claim 6 further including the step of:

(k) if the ensemble distance is greater than the first threshold and less than the second threshold, prompting the caller to answer one or more personal questions associated with the password; and

(l) determining whether the questions are answered correctly; and

(m) if the questions are answered correctly, accepting the caller'"'"'s identity.

9. A method for enabling a caller to obtain access to one or more services via a telephone network by speaking a password having a plurality of characters, comprising the steps of:

prompting the caller to speak the password;

generating first and second sets of parameters for each spoken character using a voice recognition feature transformation and a voice verification feature transformation, respectively, the first set of parameters for use in a voice recognition system and the second set of parameters for use in a voice verification system;

recognizing each spoken character of the password using the first set of parameters;

following entry of the password, determining whether the password is valid;

if the password is valid, using the second set of parameters to attempt to verify the caller'"'"'s identity;

if the caller'"'"'s identity cannot be verified, prompting the caller to answer at least one personal question associated with the password;

determining whether the question is answered correctly; and

if the question is answered correctly, accepting the caller'"'"'identity.

10. The method as described in claim 9 further including the step of:

if the question is answered incorrectly, rejecting the caller'"'"'s identity.