MAINTAINING AND SUPPLYING SPEECH MODELS

• US 20130030802A1
• Filed: 07/09/2012
• Published: 01/31/2013
• Est. Priority Date: 07/25/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Jia, Bin, Liu, Ying, Wu, Jia, Zhang, Zhen, Lu, E. Feng
• Granted Patent: US 8,938,388 B2
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• CPC Class Codes: G10L 15/07 to the speaker
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Maintaining and supplying a plurality of speech models is provided. A plurality of speech models and metadata for each speech model are stored. A query for a speech model is received from a source. The query includes one or more conditions. The speech model with metadata most closely matching the supplied one or more conditions is determined. The determined speech model is provided to the source. A refined speech model is received from the source, and the refined speech model is stored.

Claims

1. A method of maintaining and supplying a plurality of speech models comprising:

storing a plurality of speech models;

storing metadata for each stored speech model;

receiving a query for a speech model from a source, the query comprising one or more conditions;

determining the speech model with metadata most closely matching the supplied one or more conditions;

supplying the determined speech model to the source;

receiving a refined speech model from the source; and

storing the refined speech model.

2. The method of claim 1, wherein storing the refined speech model comprises replacing the determined speech model with the refined speech model.

3. The method of claim 1, wherein the query for the speech model comprises a condition identifying a user at the source and the metadata for the determined speech model includes details of the identified user.

4. The method of claim 1, further comprising:

receiving new metadata for the refined speech model from the source; and

storing the new metadata for the refined speech model.

5. The method of claim 4, wherein the query for the speech model comprises a condition identifying an environment at the source, the metadata for the determined speech model does not include details of the identified environment and the storing of the new metadata for the refined speech model includes details of the identified environment.

6. A system for maintaining and supplying a plurality of speech models comprising:

one or more storage devices configured to store a plurality of speech models and metadata for each stored speech model; and

a network interface connected to the one or more storage devices, accessible by an external source and configured to;

receive from a source a query for a speech model, the query comprising one or more conditions;

determine the speech model with metadata most closely matching the supplied one or more conditions;

supply to the source the determined speech model;

receive from the source a refined speech model; and

store the refined speech model on the one or more storage devices.

7. The system of claim 6, wherein the network interface is configured, when storing the refined speech model on the one or more storage devices, to replace the determined speech model with the refined speech model.

8. The system of claim 6, wherein the query for a speech model comprises a condition identifying a user at the source and the metadata for the determined speech model includes details of the identified user.

9. The system of claim 6, wherein the network interface is further configured to receive from the source new metadata for the refined speech model and store the new metadata for the refined speech model on the one or more storage devices.

10. The system of claim 9, wherein the query for the speech model comprises a condition identifying an environment at the source, the metadata for the determined speech model does not include details of the identified environment and the storing of the new metadata for the refined speech model includes details of the identified environment.

11. A computer program product for maintaining and supplying a plurality of speech models, the computer program product comprising:

a computer readable storage medium having computer readable program code embodied therewith, the computer readable program code comprising;

computer readable program code configured to perform;

storing a plurality of speech models;

storing metadata for each stored speech model;

receiving a query for a speech model from a source, the query comprising one or more conditions;

determining the speech model with metadata most closely matching the supplied one or more conditions;

supplying the determined speech model to the source;

receiving a refined speech model from the source; and

storing the refined speech model.

12. The computer program product of claim 11, wherein storing the refined speech model comprises replacing the determined speech model with the refined speech model.

13. The computer program product of claim 11, wherein the query for a speech model comprises a condition identifying a user at the source and the metadata for the determined speech model includes details of the identified user.

14. The computer program product of claim 11, wherein the computer readable program code is configured to perform:

receiving new metadata for the refined speech model from the source; and

storing the new metadata for the refined speech model.

15. The computer program product according to claim 14, wherein the query for the speech model comprises a condition identifying an environment at the source, the metadata for the determined speech model does not include details of the identified environment and storing of the new metadata for the refined speech model includes details of the identified environment.