Speaker identification using unsupervised speech models
First Claim
Patent Images
1. A method of producing a speech model associated with a speaker for use in determining whether the speaker produced an unidentified speech sample, the method comprising:
- obtaining a sample of speech of the speaker; and
producing a speech model for subsequent use associated with the speaker by;
identifying words in the sample of speech using speech recognition, andderiving a new speech model using the sample of speech and the identified words,the speech model being derived without verifying the accuracy of the identified words and without controlling the content of the sample of speech.
7 Assignments
0 Petitions
Accused Products
Abstract
A speech model is produced for use in determining whether a speaker associated with the speech model produced an unidentified speech sample. First a sample of speech of a particular speaker is obtained. Next, the contents of the sample of speech are identified using speech recognition. Finally, a speech model associated with the particular speaker is produced using the sample of speech and the identified contents thereof. The speech model is produced without using an external mechanism to monitor the accuracy with which the contents were identified.
192 Citations
29 Claims
-
1. A method of producing a speech model associated with a speaker for use in determining whether the speaker produced an unidentified speech sample, the method comprising:
-
obtaining a sample of speech of the speaker; and producing a speech model for subsequent use associated with the speaker by; identifying words in the sample of speech using speech recognition, and deriving a new speech model using the sample of speech and the identified words, the speech model being derived without verifying the accuracy of the identified words and without controlling the content of the sample of speech. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21)
-
-
22. A method of identifying a speaker who produced an unidentified speech sample, comprising:
producing speech models for known speakers, wherein a speech model for each known speaker is produced by; obtaining a sample of speech of the known speaker, identifying words in the sample of speech of the known speaker using speech recognition, and deriving the speech model using the sample of speech and the identified words without verifying the accuracy of the identified words and without controlling the content of the sample of speech; obtaining a sample of speech from an unidentified speaker; identifying words in the sample of speech from the unidentified speaker; and comparing, based on the identified words, the sample of speech from the unidentified speaker to the speech models for the known speakers to determine whether a known speaker is likely to be the unidentified speaker. - View Dependent Claims (23, 24, 25)
-
26. A method of processing voice messages in a voice messaging system, comprising:
-
receiving a first voice message; using speech recognition to identify words in the first voice message without controlling the content of the first voice message; deriving a speech model for the speaker of the first voice message using the identified words; and using the derived speech model to determine if the speaker of the first voice message is the speaker of a second voice message. - View Dependent Claims (27, 28, 29)
-
Specification