SYSTEM AND METHOD FOR RAPID CUSTOMIZATION OF SPEECH RECOGNITION MODELS

• US 20120253799A1
• Filed: 03/28/2011
• Published: 10/04/2012
• Est. Priority Date: 03/28/2011
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Inventors: BANGALORE, Srinivas, Bell, Robert, Caseiro, Diamantino Antonio, Gilbert, Mazin, Haffner, Patrick
• Granted Patent: US 9,679,561 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 15/065 Adaptation
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
• CPC Class Codes: G10L 2015/0635 updating or merging of old and new templates; Mean values; Weighting
• CPC Class Codes: G10L 2015/0636 Threshold criteria for the updating
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Disclosed herein are systems, methods, and non-transitory computer-readable storage media for generating domain-specific speech recognition models for a domain of interest by combining and tuning existing speech recognition models when a speech recognizer does not have access to a speech recognition model for that domain of interest and when available domain-specific data is below a minimum desired threshold to create a new domain-specific speech recognition model. A system configured to practice the method identifies a speech recognition domain and combines a set of speech recognition models, each speech recognition model of the set of speech recognition models being from a respective speech recognition domain. The system receives an amount of data specific to the speech recognition domain, wherein the amount of data is less than a minimum threshold to create a new domain-specific model, and tunes the combined speech recognition model for the speech recognition domain based on the data.

Claims

1. A method of generating a domain-specific speech recognition model, the method comprising:

identifying a speech recognition domain;

combining a plurality of speech recognition models to yield a combined speech recognition model, each speech recognition model of the plurality of speech recognition models being from a respective speech recognition domain;

receiving an amount of data specific to the speech recognition domain, wherein the amount of data is less than a minimum threshold to create a new domain-specific model; and

tuning the combined speech recognition model for the speech recognition domain based on the amount of data.

2. The method of claim 1, wherein each respective speech recognition domain comprises at least one of business, finance, travel, medical, sports, news, politics, entertainment, and education.

3. The method of claim 1, wherein tuning the combined speech recognition model is performed in a cloud computing environment.

4. The method of claim 1, wherein tuning the combined speech recognition model is performed on-demand in response to a request.

5. The method of claim 1, wherein the plurality of speech recognition models comprises at least two speech recognition models from different domains.

6. The method of claim 1, wherein the combined speech recognition model and at least one of the plurality of speech recognition models are from different domains.

7. The method of claim 1, wherein the amount of data comprises at least one of text, speech, transition data, metadata, and audio.

8. The method of claim 1, wherein the speech recognition domain is specific to a particular user.

9. The method of claim 1, wherein tuning the combined speech recognition model further comprises sampling the amount of data.

10. The method of claim 1, further comprising recognizing speech using the combined speech recognition model.

11. A system for recognizing speech, the system comprising:

a processor;

a first module configured to control the processor to identify a speech recognition domain;

a second module configured to control the processor to combine a plurality of speech recognition models to yield a combined speech recognition model, each speech recognition model of the plurality of speech recognition models being from a respective speech recognition domain;

a third module configured to control the processor to receive an amount of data specific to the speech recognition domain, wherein the amount of data is less than a minimum threshold to create a new domain-specific model;

a fourth module configured to control the processor to tune the combined speech recognition model for the speech recognition domain based on the amount of data; and

a fifth module configured to control the processor to recognize speech using the combined speech recognition model.

12. The system of claim 11, wherein the fourth module is further configured to control the processor to tune the combined speech recognition model on-demand in response to a request.

13. The system of claim 11, wherein the plurality of speech recognition models comprises at least two speech recognition models from different domains.

14. The system of claim 11, wherein the combined speech recognition model and at least one of the plurality of speech recognition models are from different domains.

15. The system of claim 11, wherein the amount of data comprises at least one of text, speech, transition data, metadata, and audio.

16. A non-transitory computer-readable storage medium storing instructions which, when executed by a computing device, cause the computing device to generate a speech recognition model for a specific recognition domain, the instructions comprising:

combining a plurality of speech recognition models to yield a combined speech recognition model, each speech recognition model of the plurality of speech recognition models being from a respective speech recognition domain;

receiving an amount of data specific to a speech recognition domain, wherein the amount of data is less than a minimum threshold to create a new domain-specific model; and

tuning the combined speech recognition model for the speech recognition domain based on the amount of data.

17. The non-transitory computer-readable storage medium of claim 16, wherein combining the plurality of speech recognition models is performed at at least one of a core n-gram level and a sentence level.

18. The non-transitory computer-readable storage medium of claim 16, wherein tuning the combined speech recognition model further comprises sampling the amount of data.

19. The non-transitory computer-readable storage medium of claim 16, further comprising recognizing speech using the combined speech recognition model.

20. The non-transitory computer-readable storage medium of claim 16, wherein the speech recognition domain is specific to a particular user.