Fixed text speaker verification method and apparatus
First Claim
1. A method for utilizing a speech utterance to verify an identity of a person, comprising the steps of:
- collecting a representation of an identity asserted by the person;
collecting an uttered phrase from the person which has a predetermined sequence and converting the phrase into data characteristic of the person'"'"'s voice;
comparing the characteristic data with other stored data characteristic of the phrase as spoken by the person to produce a match or a mismatch;
confirming the identity of the person on a match of said comparison;
denying the identity of the person on a mismatch of said comparison;
prompting said speaker a second time in response to a mismatch of said comparison to repeat said phrase;
comparing characteristic data of the repeated phrase with said other data to again determine a match or mismatch;
confirming or denying the identity of the person based upon said second comparison; and
updating said other stored data on determining a match of said comparison by averaging said characteristic data with said other stored data and storing the result thereof for use in subsequent comparisons.
1 Assignment
0 Petitions
Accused Products
Abstract
Speaker verification is performed by computing principal components of a fixed text statement comprising a speaker identification code and a two-word phrase, and principal spectral components of a random word phrase. A multi-phrase strategy is utilized in access control to allow successive verification attempts in a single session, if the speaker fails initial attempts. Based upon a verification attempt, the system produces a verification score which is compared with a threshold value. On successive attempts, the criterion for acceptance is changed, and one of a number of criteria must be satisfied for acceptance in subsequent attempts. A speaker normalization function can also be invoked to modify the verification score of persons enrolled with the system who inherently produce scores which result in denial of access. Accuracy of the verification system is enhanced by updating the reference template which then more accurately symbolizes the person'"'"'s speech signature.
-
Citations
83 Claims
-
1. A method for utilizing a speech utterance to verify an identity of a person, comprising the steps of:
-
collecting a representation of an identity asserted by the person; collecting an uttered phrase from the person which has a predetermined sequence and converting the phrase into data characteristic of the person'"'"'s voice; comparing the characteristic data with other stored data characteristic of the phrase as spoken by the person to produce a match or a mismatch; confirming the identity of the person on a match of said comparison; denying the identity of the person on a mismatch of said comparison; prompting said speaker a second time in response to a mismatch of said comparison to repeat said phrase; comparing characteristic data of the repeated phrase with said other data to again determine a match or mismatch; confirming or denying the identity of the person based upon said second comparison; and updating said other stored data on determining a match of said comparison by averaging said characteristic data with said other stored data and storing the result thereof for use in subsequent comparisons. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
-
18. A method of verifying the identity of individuals using voice as an identifying characteristic, comprising the steps of:
-
sampling speech signals and converting said signals into digital data; converting said digital data by a linear predictive coding (LPC) technique to define an LPC vector; converting said LPC vector into simulated amplitude vectors; converting the amplitude vectors into respective uncorrelated features defining principal spectral components (PSC); arranging the principal spectral components into a file having a number of frames; time warping said file to match a stored reference file of principal spectral component data previously developed on enrollment of said individual; processing the time warped principal spectral component data with the stored reference principal spectral component data to produce a numerical result; comparing said numerical result with a threshold value; and confirm or deny identity of the person based upon the result of said comparison. - View Dependent Claims (19, 20, 21, 22, 23, 24, 25, 26)
-
-
27. A method of providing access control utilizing speech utterances in a number of attempts defining a session, to verify an identity of a person, comprising the steps of:
-
forming a reference template of parameters characteristic of the speaker'"'"'s speech; converting signals of the speaker into similar parameters when said speaker seeks access; processing said reference template with said similar parameters to derive a result d1 indicative of the similarity between said parameters; comparing said result d1 with a threshold value θ
1 in a first attempt;granting or denying access to the person based upon the result of said first attempt comparison; and comparing a result d2 with a threshold value θ
2 on a second attempt if said first attempt results in a denial of said access. - View Dependent Claims (28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38)
-
-
39. A method of providing access control utilizing a speech utterance to verify an identity of a person, comprising the steps of:
-
forming a reference template of parameters characteristic of the speaker'"'"'s speech; prompting the speaker to enter a fixed text statement; prompting the person to enter a randomized text statement comprising a number of words which are randomly arranged, and which arrangement is generally unknown to the speaker; converting the speech signals of said fixed text statement and the speech signals of said randomized text statement voiced by the person into test parameters; processing said test parameters with said reference template parameters to derive a result indicative of a match or mismatch therebetween; and allowing access to the person if a match is found and denying access to the person if a mismatch is found. - View Dependent Claims (40, 41, 42, 43, 44, 45, 46)
-
-
47. Apparatus for utilizing a speech utterance to verify an identity of a person, comprising:
-
a prompter responsive to an identity asserted by the person for prompting the person to utter a phrase having a predetermined sequence of words; a converter for collecting a representation of an uttered phrase of a predetermined sequence of words, and converting the representation into data characteristic of the person'"'"'s voice; a comparator for comparing the characteristic data with other stored data characteristic of the phrase as spoken by the person to produce a match or a mismatch; a decision circuit for confirming the identity of the person on a match of said comparison and denying the identity of the person on a mismatch of said comparison; a comparator for comparing characteristic data of a repeated phrase on a second attempt by the person with said other data to again determine a match or mismatch; a decision circuit for confirming or denying the identity of the person based upon said second comparison; and means for updating said other stored data on determining said acceptance by averaging said characteristic data with said other stored data and means for storing the result thereof for use in subsequent comparisons. - View Dependent Claims (48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63)
-
-
64. Apparatus for providing personnel access control utilizing speech utterances in a number of attempts defining a session, to verify an identity of a person, comprising:
-
a processor for forming a reference template of parameters characteristic of the speaker'"'"'s speech; a converter for converting signals of the speaker into similar parameters when said speaker seeks access; a processor for processing said reference template with said similar parameters to derive a result d1 indicative of the similarity between said parameters; a comparator for comparing said result d1 with a threshold value θ
1 in a first attempt for granting or dening access to the person based upon the result of said first attempt comparison; anda comparator for comparing a result d2 with a threshold value θ
2 on a second attempt if said first attempt results in a denial of said access. - View Dependent Claims (65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75)
-
-
76. Apparatus for providing personnel access control utilizing a speech utterance to verify an identity of a person, comprising:
-
a processor for forming a reference template of parameters characteristic of the speaker'"'"'s speech; a prompter for prompting the speaker to enter a fixed text statement; a prompter for prompting the person to enter a variable text statement comprising a number of words which are randomly arranged, and which arrangement is generally unknown to the speaker; a converter for converting the speech signals of said fixed text statement and the speech signals of said variable text statement voiced by the person into test parameters; and a processor for processing said test parameters with said reference template parameters to derive a result indicative of a match or mismatch therebetween and for allowing access to the person if a match is found and for denying access to the person if a mismatch is found. - View Dependent Claims (77, 78, 79, 80, 81, 82, 83)
-
Specification