Virtual voiceprint system and method for generating voiceprints
View Patent ↗A system and method for generating on-demand voiceprints are presented wherein voiceprints are created on the fly using voice recordings and associated metadata specified by an application. The application requests a voiceprint and specifies a description of the data necessary to generate the voiceprint, including the appropriate voice recordings, the requisite verification engine and other parameters that should be utilized to generate the voiceprint. The specified voice recordings are accessed from storage and a voiceprint is produced using the designated speech engine and application-specified parameters.
1. A system for generating voiceprints, comprising:
a first specification for designating a first set of data to be used in producing a first voiceprint;
stored voice recordings annotated with metadata describing characteristics associated with said stored voice recordings; and
a voice print generator for receiving said first specification and using said first set of data to retrieve selected voice recordings from said stored voice recordings, and further for generating said first voiceprint using said selected voice recordings;
wherein the first specification comprises a first voiceprint description received from a first application, the first voiceprint description identifying a first subset of said stored voice recordings to be utilized by the voice print generator to generate the first voiceprint; and
wherein the voice print generator is further operative to generate a second voiceprint responsive to a second specification comprising a second voiceprint description received from a second application different than the first application, the second voiceprint description identifying a second subset of said stored voice recordings, different than the first subset, to be utilized by the voice print generator to generate the second voiceprint;
the first and second voiceprints thereby being generated on demand by the voiceprint generator from the stored voice recordings responsive to the respective first and second voiceprint descriptions received from the respective first and second applications.
2. The system of claim 1 , wherein said first specification includes a designation of the specific speech engine to be used in producing said first voiceprint, and further wherein said voice print generator creates said first voiceprint using said specific speech engine.
3. The system of claim 2 , wherein said first specification is provided by the first application, said first application being coupled to said system, said first application requesting said first voiceprint.
4. The system of claim 3 , wherein said first application collects current voice recordings from a first application user and annotates said current voice recordings with additional metadata, and further wherein said first application provides said current voice recordings and said additional metadata to said system for storage and subsequent use.
5. A system for generating on-demand voiceprints from stored voice recordings, comprising:
a data manager for receiving requests from a plurality of applications, wherein said applications provide said data manager with data specifications related to specific voiceprints requested by said applications;
a storage system for storing said voice recordings, wherein said voice recordings are annotated with metadata describing the characteristics of said voice recordings; and
a voice print generator for generating said specific voiceprints using said voice recordings;
wherein said data manager retrieves said voice recordings by correlating said data specifications with said metadata, and further wherein said data manager provides said voice print generator with additional information for use by said voice print generator in producing said voiceprints;
wherein a first specification comprises a first voiceprint description received from a first application, the first voiceprint description identifying a first subset of said stored voice recordings to be utilized by the voice print generator to generate a first voiceprint; and
wherein the voice print generator is further operative to generate a second voiceprint responsive to a second specification comprising a second voiceprint description received from a second application different than the first application, the second voiceprint description identifying a second subset of said stored voice recordings, different than the first subset, to be utilized by the voice print generator to generate the second voiceprint;
the first and second voiceprints thereby being generated on demand by the voiceprint generator from the stored voice recordings responsive to the respective first and second voiceprint descriptions received from the respective first and second applications.
6. The system of claim 5 wherein said additional information includes a specification of the particular speech engine to be used by said voice print generator in producing said voiceprints.
7. The system of claim 5 wherein said data manager is operative to receive additional voice recordings from said applications for storage in said storage system.
8. The system of claim 7 wherein said additional voice recordings are annotated by said applications with additional metadata describing the characteristics of said additional voice recordings.
9. The system of claim 7 wherein said additional voice recordings are annotated by said data manager with additional metadata describing the characteristics of said additional voice recordings.
10. A method for generating on-demand voiceprints, comprising the steps of:
receiving a first specification from a first application describing data necessary to generate a first voiceprint;
retrieving voice recordings from a storage system by correlating said data with metadata associated with said voice recordings in said storage system;
processing said voice recordings in accordance with said specification and generating said first voiceprint using said stored voice recordings; and
providing said first voiceprint to said first application;
wherein the first specification comprises a first voiceprint description received from the first application, the first voiceprint description identifying a first subset of said stored voice recordings to be utilized by a voice print generator to generate the first voiceprint; and
wherein the voice print generator is further operative to generate a second voiceprint responsive to a second specification comprising a second voiceprint description received from a second application different than the first application, the second voiceprint description identifying a second subset of said stored voice recordings, different than the first subset, to be utilized by the voice print generator to generate the second voiceprint;
the first and second voiceprints thereby being generated on demand by the voiceprint generator from the stored voice recordings responsive to the respective first and second voiceprint descriptions received from the respective first and second applications.
11. The method of claim 10 , wherein said first specification identifies a specific speech engine for use in creating said first voiceprint.
12. The method of claim 11 , wherein said first application utilizes said first voiceprint to authenticate a user.
13. The method of claim 10 , wherein said first application collects current voice recordings from said user and annotates said current voice recordings with additional metadata, and further wherein said first application provides said current voice recordings and said additional metadata to said storage system for storage and subsequent use.