Automatic learning of language models

• US 10,535,342 B2
• Filed: 04/10/2017
• Issued: 01/14/2020
• Est. Priority Date: 04/10/2017
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Liensberger, Christian
• Primary Examiner(s): He, Jialong
• Application Number:
  US15/483,977
• Publication Number:
  US 20180293977A1
• Time in Patent Office: 1,009 Days
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/193 Formal grammars, e.g. finite state automata, context free grammars or word networks
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Techniques and systems are disclosed for context-dependent speech recognition. The techniques and systems described enable accurate recognition of speech by accessing sub-libraries associated with the context of the speech to be recognized. These techniques translate audible input into audio data at a smart device and determine context for the speech, such as location-based, temporal-based, recipient-based, and application based context. The smart device then accesses a context-dependent library to compare the audio data with phrase-associated translation data in one or more sub-libraries of the context-dependent library to determine a match. In this way, the techniques allow access to a large quantity of phrases while reducing incorrect matching of the audio data to translation data caused by organizing the phrases into context-dependent sub-libraries.

Claims

1. A method for building a context-dependent library comprising:

analyzing a text-based electronic communication to;

determine a phrase within the text-based electronic communication the phrase being one or more words in length,

determine a context for the text-based electronic communication, wherein the context comprises a plurality of analytical elements that comprise one or more user identification elements associated with an organization and one or more of;

a location-based element associated with the text-based electronic communication and a temporal-based element associated with the text-based electronic communication, and

determine a context vector for the text-based electronic communication based on analysis of the plurality of analytical elements associated with the determined context;

determining, based on the context vector for the text-based electronic communication, a context-dependent library for speech recognition, wherein the context-dependent library is a sub-library of multi-level library for speech recognition;

altering the context-dependent library by one or more of;

adding the phrase into the context-dependent library, and increasing priority of the phrase in the context-dependent library; and

granting access to the context-dependent library to a user within the organization based on a privilege permission of the user within the organization.

2. The method of claim 1, wherein altering the context-dependent library includes comparing the phrase with phrases in a general library and altering the context-dependent library is responsive to determining that the phrase is not within the general library.

3. The method of claim 1, wherein altering the context-dependent library includes comparing the phrase with phrases in the context-dependent library and altering the context-dependent library is responsive to determining that the phrase is already within the context-dependent library, and wherein the altering is effective to raise a priority of the phrase within the context-dependent library.

4. The method of claim 1, wherein altering the context-dependent library includes comparing the phrase with phrases in the context-dependent library and altering the context-dependent library is responsive to determining that the phrase is not already within the context-dependent library, and wherein the altering is effective to add the phrase to the context-dependent library.

5. The method of claim 1, wherein the phrase on which the altering is based is an n-gram.

6. The method of claim 1, wherein the context vector includes one or more of a time at which the text-based electronic communication is sent and a time at which the text-based electronic communication is received.

7. The method of claim 1, wherein the context vector includes a one or more of name of a sender of the text-based electronic communication and a name of a recipient of the text-based electronic communication.

8. The method of claim 1, wherein the context-dependent library is shared and accessed by one or more users of the organization.

9. The method of claim 8, wherein the shared context-dependent library is alterable by the one or more users of the organization.

10. The method of claim 1, further comprising:

organizing the context-dependent library into sub-libraries; and

granting access to one or more sub-libraries of the context-dependent library to users within the organization based on privilege permissions of respective user'"'"'s within the organization.

11. A system comprising:

at least one processor; and

a memory, operatively connected with the at least one processor, storing computer-executable instructions that, when executed by the at least one processor, causing the at least one processor to execute a method that comprises;

analyzing a text-based electronic communication to;

determine a phrase within the text-based electronic communication the phrase being one or more words in length,

determine a context for the text-based electronic communication, wherein the context comprises a plurality of analytical elements that comprise one or more user identification elements associated with an organization and one or more of;

a location-based element associated with the text-based electronic communication and a temporal-based element associated with the text-based electronic communication, and

determine a context vector for the text-based electronic communication based on analysis of the plurality of analytical elements associated with the determined context;

determining, based on the context vector for the text-based electronic communication, a context-dependent library for speech recognition, wherein the context-dependent library is a sub-library of multi-level library for speech recognition;

altering the context-dependent library by one or more of;

adding the phrase into the context-dependent library, and increasing priority of the phrase in the context-dependent library; and

granting access to the context-dependent library to a user within the organization based on a privilege permission of the user within the organization.

12. The system of claim 11, wherein altering the context-dependent library includes comparing the phrase with phrases in a general library and altering the context-dependent library is responsive to determining that the phrase is not within the general library.

13. The system of claim 11, wherein altering the context-dependent library includes comparing the phrase with phrases in the context-dependent library and altering the context-dependent library is responsive to determining that the phrase is already within the context-dependent library, and wherein the altering is effective to raise a priority of the phrase within the context-dependent library.

14. The system of claim 11, wherein the context-dependent library is shared and accessed by one or more users of the organization, and the shared context-dependent library is alterable by the one or more users of the organization.

15. The system of claim 11, wherein the method, executed by the at least one processor, further comprises:

organizing the context-dependent library into sub-libraries; and

granting access to one or more sub-libraries of the context-dependent library to users within the organization based on privilege permissions of respective users within the organization.

16. A computer-readable storage media storing computer-executable instructions that, when executed by at least one processor, causing the at least one processor to execute a method comprising:

analyzing a text-based electronic communication to;

determine a phrase within the text-based electronic communication the phrase being one or more words in length,

determine a context for the text-based electronic communication, wherein the context comprises a plurality of analytical elements that comprise one or more user identification elements associated with an organization and one or more of;

a location-based element associated with the text-based electronic communication and a temporal-based element associated with the text-based electronic communication, and

determine a context vector for the text-based electronic communication based on analysis of the plurality of analytical elements associated with the determined context;

determining, based on the context vector for the text-based electronic communication, a context-dependent library for speech recognition, wherein the context-dependent library is a sub-library of multi-level library for speech recognition;

altering the context-dependent library by one or more of;

adding the phrase into the context-dependent library, and increasing priority of the phrase in the context-dependent library; and

granting access to the context-dependent library to a user within the organization based on a privilege permission of the user within the organization.

17. The computer-readable storage media of claim 16, wherein altering the context-dependent library includes comparing the phrase with phrases in a general library and altering the context-dependent library is responsive to determining that the phrase is not within the general library.

18. The computer-readable storage media of claim 16, wherein altering the context-dependent library includes comparing the phrase with phrases in the context-dependent library and altering the context-dependent library is responsive to determining that the phrase is already within the context-dependent library, and wherein the altering is effective to raise a priority of the phrase within the context-dependent library.

19. The computer-readable storage media of claim 16, wherein the context-dependent library is shared and accessed by one or more users of the organization, and the shared context-dependent library is alterable by the one or more users of the organization.

20. The computer-readable storage media of claim 16, wherein the method, executed by the at least one processor, further comprises:

organizing the context-dependent library into sub-libraries; and

granting access to one or more sub-libraries of the context-dependent library to users within the organization based on privilege permissions of respective users within the organization.