TEXT-DEPENDENT SPEAKER VERIFICATION
First Claim
Patent Images
1. A computer-implemented process for verifying the identity of a speaker, comprising:
- asking a speaker to repeat a pass phrase;
adapting a speaker-independent model to create a speaker-specific model to be used in verifying that the pass phrase is spoken by a specific speaker.
2 Assignments
0 Petitions
Accused Products
Abstract
A text-dependent speaker verification technique that uses a generic speaker-independent speech recognizer for robust speaker verification, and uses the acoustical model of a speaker-independent speech recognizer as a background model. Instead of using a likelihood ratio test (LRT) at the utterance level (e.g., the sentence level), which is typical of most speaker verification systems, the present text-dependent speaker verification technique uses weighted sum of likelihood ratios at the sub-unit level (word, tri-phone, or phone) as well as at the utterance level.
119 Citations
20 Claims
-
1. A computer-implemented process for verifying the identity of a speaker, comprising:
-
asking a speaker to repeat a pass phrase; adapting a speaker-independent model to create a speaker-specific model to be used in verifying that the pass phrase is spoken by a specific speaker. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
-
16. A system for verifying the identity of speaker, comprising:
-
a general purpose computing device; a computer program comprising program modules executable by the general purpose computing device, wherein the computing device is directed by the program modules of the computer program to, adapt a speaker-independent model to create a speaker-specific model to be used in verifying the identity of a speaker; input a speech signal containing a pass phrase as spoken by a user who claims to be a specific speaker; extract features from the speech signal; input the features and the speaker-independent model into a speech recognizer and outputting a first recognizer score; input the features and the speaker-specific model into a speech recognizer and outputting a second recognizer score; and use the first and second recognizer scores to determine if the pass phrase is spoken by the specific speaker. - View Dependent Claims (17, 18, 19)
-
-
20. A system for verifying that a speaker who claims to be a specific person is that specific person, comprising:
-
a general purpose computing device; a computer program comprising program modules executable by the general purpose computing device, wherein the computing device is directed by the program modules of the computer program to, input an audio signal of a phrase spoken by a speaker claiming to be a given person into a trained speech recognizer; input speaker-independent models into the trained speech recognizer; input speaker-specific models of the given person'"'"'s speech that were derived by adapting the speaker-independent models into the trained speech recognizer; and use the trained speech recognizer and the input speaker-independent models and the speaker-specific models to determine if the input audio signal of the phrase was spoken by the given person.
-
Specification