Flexible schema for language model customization

Patent metadata

Field	Value
Publication number	US-9529794-B2
Application number	US-201414227492-A
Country	US
Kind code	B2
Filing date	Mar 27, 2014
Priority date	Mar 27, 2014
Publication date	Dec 27, 2016
Grant date	Dec 27, 2016

Abstract

The customization of language modeling components for speech recognition is provided. A list of language modeling components may be made available by a computing device. A hint may then be sent to a recognition service provider for combining the multiple language modeling components from the list. The hint may be based on a number of different domains. A customized combination of the language modeling components based on the hint may then be received from the recognition service provider.

First claim

What is claimed is:

1. A computer-implemented method of customizing language modeling components, comprising:
displaying a list of language modeling components;
receiving a selection of one or more language modeling components from the list;
receiving selection of a fixed weight value for the selected one or more of the language modeling components;
generating information based on the selection, wherein the information indicates the selected one or more of the language modeling components based on one or more domains and the selected value for the selected one or more of the language modeling components;
sending the information to a service provider; and
receiving from the service provider a customized combination of the selected language modeling components based on the information.

2. The method of claim 1 , further comprising maintaining an association between the information and the one or more language modeling components.

3. The method of claim 1 , wherein the information further includes a selection of a pre-compiled language model based on the one or more domains.

4. The method of claim 1 , wherein the information further includes a selection of one or more recognition topics from a pre-complied list, the one or more recognition topics corresponding to one or more of the language modeling components.

5. The method of claim 1 , further comprising sending an in-domain text corpus.

6. The method of claim 1 , further comprising sending an in-domain audio corpus.

7. The method of claim 1 , wherein the information includes an existing combination of language modeling components for re-use.

8. The method of claim 1 , wherein the method further comprises sending a recognition request comprising the information.

9. The method of claim 1 , wherein the information is sent prior to initiating an offline initialization process.

10. The computer-implemented method of claim 1 , further comprising displaying a list of fixed weights concurrently with the list of language modeling components.

11. A system for customizing language modeling components, comprising:
a memory for storing executable program code; and
a processor, functionally coupled to the memory, the processor being responsive to computer-executable instructions contained in the program code and operative to:
display a list of language modeling components;
receiving a selection of one or more of the language modeling components from the list;
receive a selection of a fixed weight value for the selected one or more of the language modeling components;
generate information based on the selection, wherein the information indicates the selected one or more of the language modeling components based on one or more domains and the selected fixed weight value for the selected one or more of the language modeling components;
send the information to a service provider; and
receive from the service provider a customized combination of the selected language modeling components based on the information.

12. The system of 11 , wherein the processor is operative to send a selection of a pre-compiled language model based on the one or more of the plurality of domains.

13. The system of claim 11 , wherein the processor is operative to:
send a selection of one or more recognition topics from a pre-complied list, the one or more recognition topics corresponding to one or more of the language modeling components; and
apply one or more weights to the one or more language modeling components.

14. The system of 11 , wherein the processor is operative to display a list of fixed weights concurrently with the list of language modeling components.

15. A computer-readable storage device storing computer executable instructions which, when executed by a computer, will cause computer to perform a method of customizing language modeling components, the method comprising:
displaying a list of distinct language modeling components;
receiving a selection one or more language modeling components from the list;
receiving a selection of a fixed weight value for the selected one or more of the language modeling components;
generating information based on the on the selection, wherein the information indicates the selected one or more of the language modeling components based on one or more domains and the selected fixed weight value for the selected one or more of the language modeling components;
sending the information to a service provider; and
receiving from the service provider a customized combination of the selected distinct language modeling components based on the information.

16. The computer-readable storage device of claim 15 , wherein sending the information comprises sending a selection of a pre-compiled language model based on the one or more domains.

17. The computer-readable storage device of claim 15 , wherein sending the information comprises:
sending a selection of one or more distinct recognition topics from a pre-complied list, the one or more distinct recognition topics corresponding to one or more of the distinct language modeling component.

18. The computer-readable storage device of claim 15 , wherein the information further comprises at least one of an in-domain text corpus and an in-domain audio corpus.

19. The computer-readable storage device of claim 15 , wherein the method further comprises displaying a list of fixed weights concurrently with the list of distinct language modeling components.

20. The computer-readable storage device of claim 15 , wherein the information includes an existing combination of language modeling components for re-use.

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Levit Michael
• Guelman Hernan
• Chang Shuangyu
• Parthasarathy Sarangarajan
• Dumoulin Benoit

Classifications

• G06F40/268

  Morphological analysis · CPC title

• G06F40/30

  Semantic analysis · CPC title

• G10L15/30 Primary

  Distributed recognition, e.g. in client-server systems, for mobile phones or network applications · CPC title

• G10L15/183 Primary

  using context dependencies, e.g. language models · CPC title

• G06F17/2755 Primary

  Physics · mapped topic