SYSTEMS AND METHODS FOR SPEECH-BASED SEARCHING OF CONTENT REPOSITORIES

• US 20170147585A1
• Filed: 07/22/2014
• Published: 05/25/2017
• Est. Priority Date: 07/22/2014
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Kunc, Ladislav, Kleindienst, Jan, Labsky, Martin, Macek, Tomas
• Granted Patent: US 10,922,322 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06F 16/367 Ontology
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command

Abstract

According to some aspects, a method of searching for content in response to a user voice query is provided. The method may comprise receiving the user voice query, performing speech recognition to generate N best speech recognition results comprising a first speech recognition result, performing a supervised search of at least one content repository to identify one or more supervised search results using one or more classifiers that classify the first speech recognition result into at least one class that identifies previously classified content in the at least one content repository, performing an unsupervised search of the at least one content repository to identify one or more unsupervised search results, wherein performing the unsupervised search comprises performing a word search of the at least one content repository, and generating combined results from among the one or more supervised search results and the one or more unsupervised search results.

Claims

1. A method of searching for content in at least one content repository in response to a user voice query, the method comprising acts of:

receiving the user voice query;

performing speech recognition on the user voice query to generate N best speech recognition results for the user voice query, wherein N is equal to one or more, and wherein the N best speech recognition results comprise a first speech recognition result;

performing a supervised search of the at least one content repository to identify a set of one or more supervised search results, wherein each one of the one or more supervised search results is associated with a score indicative of a predicted relevance of the one of the one or more supervised search results to the user voice query, wherein performing the supervised search comprises processing the first speech recognition result using one or more classifiers that classify the first speech recognition result into at least one class that identifies previously classified content in the at least one content repository;

performing an unsupervised search of the at least one content repository to identify a set of one or more unsupervised search results, wherein each one of the one or more unsupervised search results is associated with a score indicative of a predicted relevance of the one of the one or more unsupervised search results to the user voice query, wherein performing the unsupervised search comprises performing a word search of the at least one content repository using one or more words derived from the first speech recognition result; and

generating a set of combined results from among the set of one or more supervised search results and the set of one or more unsupervised search results.

2. The method of claim 1, wherein the generating comprises evaluating the scores of the supervised search results and the scores of the unsupervised search results, and selecting for the combined results those search results that have scores indicative of the highest relevance among the supervised and unsupervised search results.

3. The method of claim 1, wherein the generating comprises including in the combined search results at least one result from the set of one or more unsupervised results and at least one result from the set of one or more supervised results.

4. The method of claim 1, further comprising identifying one or more elements of an ontology as related to one or more words of the first speech recognition result, and wherein the performing the unsupervised search and/or the performing the supervised search is based at least in part on the identified one or more elements.

5. The method of claim 1, further comprising entropy-weighting the scores of the supervised search results to generate entropy-weighted supervised search results, and wherein generating the set of combined results comprises selecting from among the entropy-weighting supervised search results.

6. The method of claim 1, further comprising presenting at least a portion of the set of combined results to a user using speech synthesis.

7. The method of claim 1, wherein the content repository comprises content describing operation of a motor vehicle.

8. An apparatus comprising:

at least one processor; and

at least one storage medium storing processor-executable instructions that, when executed by the at least one processor, perform a method of searching for content in at least one content repository in response to a user voice query, the method comprising acts of;

receiving the user voice query;

performing speech recognition on the user voice query to generate N best speech recognition results for the user voice query, wherein N is equal to one or more, and wherein the N best speech recognition results comprise a first speech recognition result;

performing a supervised search of the at least one content repository to identify a set of one or more supervised search results, wherein each one of the one or more supervised search results is associated with a score indicative of a predicted relevance of the one of the one or more supervised search results to the user voice query, wherein performing the supervised search comprises processing the first speech recognition result using one or more classifiers that classify the first speech recognition result into at least one class that identifies previously classified content in the at least one content repository;

performing an unsupervised search of the at least one content repository to identify a set of one or more unsupervised search results, wherein each one of the one or more unsupervised search results is associated with a score indicative of a predicted relevance of the one of the one or more unsupervised search results to the user voice query, wherein performing the unsupervised search comprises performing a word search of the at least one content repository using one or more words derived from the first speech recognition result; and

generating a set of combined results from among the set of one or more supervised search results and the set of one or more unsupervised search results.

9. The apparatus of claim 8, wherein the generating comprises evaluating the scores of the supervised search results and the scores of the unsupervised search results, and selecting for the combined results those search results that have scores indicative of the highest relevance among the supervised and unsupervised search results.

10. The apparatus of claim 8, wherein the generating comprises including in the combined search results at least one result from the set of one or more unsupervised results and at least one result from the set of one or more supervised results.

11. The apparatus of claim 8, wherein the method further comprises identifying one or more elements of an ontology as related to one or more words of the first speech recognition result, and wherein the performing the unsupervised search and/or the performing the supervised search is based at least in part on the identified one or more elements.

12. The apparatus of claim 8, wherein the method further comprises entropy-weighting the scores of the supervised search results to generate entropy-weighted supervised search results, and wherein generating the set of combined results comprises selecting from among the entropy-weighting supervised search results.

13. The apparatus of claim 8, wherein the method further comprises presenting at least a portion of the set of combined results to a user using speech synthesis.

14. The apparatus of claim 8, wherein the content repository comprises content describing operation of a motor vehicle.

15. At least one computer-readable storage medium storing computer-executable instructions that, when executed, perform a method of searching for content in at least one content repository in response to a user voice query, the method comprising acts of:

receiving the user voice query;

performing speech recognition on the user voice query to generate N best speech recognition results for the user voice query, wherein N is equal to one or more, and wherein the N best speech recognition results comprise a first speech recognition result;

performing a supervised search of the at least one content repository to identify a set of one or more supervised search results, wherein each one of the one or more supervised search results is associated with a score indicative of a predicted relevance of the one of the one or more supervised search results to the user voice query, wherein performing the supervised search comprises processing the first speech recognition result using one or more classifiers that classify the first speech recognition result into at least one class that identifies previously classified content in the at least one content repository;

performing an unsupervised search of the at least one content repository to identify a set of one or more unsupervised search results, wherein each one of the one or more unsupervised search results is associated with a score indicative of a predicted relevance of the one of the one or more unsupervised search results to the user voice query, wherein performing the unsupervised search comprises performing a word search of the at least one content repository using one or more words derived from the first speech recognition result; and

generating a set of combined results from among the set of one or more supervised search results and the set of one or more unsupervised search results.

16. The at least one computer-readable storage medium of claim 15, wherein the generating comprises evaluating the scores of the supervised and unsupervised search results and selecting for the combined results those search results that have scores indicative of the highest relevance among the supervised and unsupervised search results.

17. The at least one computer-readable storage medium of claim 15, wherein the generating comprises including in the combined search results at least one result from the set of one or more unsupervised results and at least one result from the set of one or more supervised results.

18. The at least one computer-readable storage medium of claim 15, wherein the method further comprises identifying one or more elements of an ontology as related to one or more words of the first speech recognition result and performing the unsupervised search or the supervised search based at least in part on the identified one or more elements.

19. The at least one computer-readable storage medium of claim 15, wherein the method further comprises entropy-weighting the scores of the supervised search results prior to generating the set of combined results.

20. The at least one computer-readable storage medium of claim 15, wherein the method further comprises presenting at least a portion of the set of combined results using speech synthesis.