Voice interface for a search engine

• US 8,380,502 B1
• Filed: 10/14/2011
• Issued: 02/19/2013
• Est. Priority Date: 02/07/2001
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Franz, Alexander Mark, Henzinger, Monika H., Brin, Sergey, Milch, Brian Christopher
• Primary Examiner(s): Lerner, Martin
• Application Number:
  US13/273,925
• Time in Patent Office: 494 Days
• Field of Search: 704/231, 704/238, 704/240, 704/244, 704/254, 704/255, 704/270, 704/236, 707/723, 707/728, 707/752, 707/765, 707/768
• US Class Current: 704/236
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/085 Methods for reducing search complexity, pruning
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A system receives a voice search query from a user, derives recognition hypotheses from the voice search query, and determines scores associated with the recognition hypotheses, the scores being based on a comparison of the recognition hypotheses to previously received search queries. The system discards at least one of the recognition hypotheses that is associated with a first score that is less than a threshold value, and constructs a first query using at least one non-discarded recognition hypothesis, where the at least one first non-discarded recognition hypothesis is associated with a second score that at least meets the threshold value. The system forwards the first query to a search system, receives first results associated with the first query, and provides the first results to the user.

Claims

1. A method comprising:

receiving, by one or more processors, a voice search query from a user;

deriving, by the one or more processors, a plurality of recognition hypotheses from the voice search query;

determining, by the one or more processors, a plurality of scores associated with the plurality of recognition hypotheses, the plurality of scores being based on a comparison of the plurality of recognition hypotheses to previously received search queries;

discarding, by the one or more processors, at least one of the plurality of recognition hypotheses that is associated with at least one first score, of the plurality of scores, that is less than a threshold value;

constructing, by the one or more processors, a first query using at least one first non-discarded recognition hypothesis of the plurality of recognition hypotheses, where the at least one first non-discarded recognition hypothesis is associated with at least one second score, of the plurality of scores, that at least meets the threshold value, constructing the first query including;

forming an initial query based on the at least one first non-discarded recognition hypothesis,

identifying a plurality of stop words in the initial query, and

pruning, from the initial query, one or more of the plurality of stop words to form the first query, the first query satisfying a length threshold;

forwarding, by the one or more processors, the first query to a search system;

receiving, by the one or more processors and from the search system, first results associated with the first query; and

providing, by the one or more processors and to the user, the first results.

2. The method of claim 1, where deriving the plurality of recognition hypotheses from the voice search query comprises:

deriving the plurality of recognition hypotheses using at least one of a language model, a phonetic dictionary, or an acoustic model.

3. The method of claim 2, further comprising:

updating, based on the voice search query, the at least one of the language model, the phonetic dictionary, or the acoustic model.

4. The method of claim 2, where the plurality of scores are determined based on the at least one of the language model, the phonetic dictionary, or the acoustic model used to derive the plurality of recognition hypotheses.

5. The method of claim 1, further comprising:

identifying a language model, of a plurality of language models, based on at least one characteristic associated with the user,

where deriving the plurality of recognition hypotheses comprises;

deriving the plurality of recognition hypotheses using the identified language model.

6. The method of claim 1, where providing the first results comprises:

adjusting a displayed order of the first results based on the at least one second score.

7. The method of claim 1, where providing the first results comprises:

adjusting a displayed quantity of the first results based on the at least one second score.

8. The method of claim 1, where providing the first results comprises:

grouping sets of the first results based on the at least one second score.

9. The method of claim 1, where providing the first results further comprises:

determining a quantity of the first results; and

when the quantity of the first results is less than a particular number;

constructing a second query based on the voice search query,

forwarding the second query to the search system,

receiving, from the search system, second results associated with the second query, and

providing, to the user, the second results.

10. A device comprising:

a memory to store instructions; and

at least one processor to execute the instructions to;

receive a voice search query from a user,

derive a plurality of recognition hypotheses from the voice search query,

determine a plurality of scores associated with the plurality of recognition hypotheses, the at least one processor determining the plurality of scores based on a comparison of the plurality of recognition hypotheses to previously received search queries,

discard at least one of the plurality of recognition hypotheses, that is associated with at least one first score, of the plurality of scores, that is less than a threshold value,

construct a first query using at least one first non-discarded recognition hypothesis of the plurality of recognition hypotheses, where the at least one first non-discarded recognition hypothesis is associated with at least one second score, of the plurality of scores, that at least meets the threshold value, the at least one processor, when constructing the first query, being further to;

form an initial query based on the at least one first non-discarded recognition hypothesis,

identify a plurality of stop words in the initial query, and

prune, from the initial query, one or more of the plurality of stop words to form the first query, the first query satisfying a length threshold, forward the first query to a search system,

receive, from the search system, first results associated with the first query, and

provide, to the user, the first results.

11. The device of claim 10, where the at least one processor, when deriving the plurality of recognition hypotheses from the voice search query, is further to:

use at least one of a language model, a phonetic dictionary, or an acoustic model to derive the plurality of recognition hypotheses.

12. The device of claim 11, where the at least one processor is further to:

update, based on the voice search query, at least one of the language model, the phonetic dictionary, or the acoustic model used to derive the plurality of recognition hypotheses.

13. The device of claim 11, where the at least one processor, when determining the plurality of scores, is further to:

determine the plurality of scores based on the at least one of the language model, the phonetic dictionary, or the acoustic model used to derive the plurality of recognition hypotheses.

14. The device of claim 10, where the at least one processor, when deriving the plurality of recognition hypotheses, is further to:

identify a language model, of a plurality of language models, based on at least one characteristic associated with the user, and

derive the plurality of recognition hypotheses using the identified language model.

15. The device of claim 10, where the at least one processor, when providing the first results, is further to:

adjust a displayed order of the first results based on the at least one second score.

16. The device of claim 10, where the at least one processor, when providing the first results, is further to:

adjust a displayed quantity of the first results based on the at least one second score.

17. The device of claim 10, where the at least one processor, when providing the first results, is further to:

group sets of the first results based on the at least one second score.

18. The device of claim 10, where the at least one processor, when providing the first results, is further to:

determine a quantity of the first results; and

when the quantity of the first results is less than a particular number;

construct a second query based on the voice search query,

forward the second query to the search system,

receive, from the search system, second search results associated with the second query, and

provide, to the user, the second results.

19. A non-transitory computer-readable medium comprising:

one or more instructions that, when executed by at least one processor, cause the at least one processor to receive a voice search query from a user;

one or more instructions that, when executed by at least one processor, cause the at least one processor to derive a plurality of recognition hypotheses from the voice search query;

one or more instructions that, when executed by at least one processor, cause the at least one processor to determine a plurality of scores associated with the plurality of recognition hypotheses, the plurality of scores being based on a comparison of the plurality of recognition hypotheses to previously received search queries;

one or more instructions that, when executed by at least one processor, cause the at least one processor to discard at least one of the plurality of recognition hypotheses, that is associated with at least one first score, of the plurality of scores, that is less than a threshold value;

one or more instructions that, when executed by at least one processor, cause the at least one processor to construct a first query using at least one first non-discarded recognition hypothesis, of the plurality of recognition hypotheses, where the at least one first non-discarded recognition hypothesis is associated with at least one second score, of the plurality of scores, that at least meets the threshold value, the one or more instructions to construct the first query, being further including;

one or more instructions form an initial query based on the at least one first non-discarded recognition hypothesis,

one or more instructions identify a plurality of stop words in the initial query, and

one or more instructions prune, from the initial query, one or more of the plurality of stop words to form the first query, the first query satisfying a length threshold;

one or more instructions that, when executed by at least one processor, cause the at least one processor to forward the first query to a search system;

one or more instructions that, when executed by at least one processor, cause the at least one processor to receive, from the search system, results associated with the first query; and

one or more instructions that, when executed by at least one processor, cause the at least one processor to provide, to the user, the first results.

20. The non-transitory computer-readable medium of claim 19, where the one or more instructions to derive the plurality of recognition hypotheses from the voice search query further include:

one or more instructions to use at least one of a language model, a phonetic dictionary, or an acoustic model to derive the plurality of recognition hypotheses.

21. The non-transitory computer-readable medium of claim 20, further comprising:

one or more instructions to update, based on the voice search query, the at least one of the language model, the phonetic dictionary, or the acoustic model used to derive the plurality of recognition hypotheses.

22. The non-transitory computer-readable medium of claim 20, where the plurality of scores are determined based on the at least one of the language model, the phonetic dictionary, or the acoustic model used to derive the plurality of recognition hypotheses.

23. The non-transitory computer-readable medium of claim 19, further comprising

one or more instructions to identify a language model, of a plurality of language models, based on at least one characteristic associated with the user, and

where the plurality of recognition hypotheses are derived using the identified language model.

24. The non-transitory computer-readable medium of claim 19, where the one or more instructions to provide the first results further include:

one or more instructions to adjust a displayed order of the first results based on the at least one second score.

25. The non-transitory computer-readable medium of claim 19, where the one or more instructions to provide the first results further include:

one or more instructions to adjust a displayed quantity of the first results based on the at least one second score.

26. The non-transitory computer-readable medium of claim 19, where the one or more instructions to provide the first results further include:

one or more instructions to group sets of the first results based on the at least one second score.

27. The non-transitory computer-readable medium of claim 19, where the one or more instructions to provide the first results further include:

one or more instructions to determine a quantity of the first results; and

when the quantity of the first results is less than a particular number;

one or more instructions to construct a second query using the voice search query,

one or more instructions to forward the second query to the search system, and

one or more instructions to cause second results, from the search system and associated with the second query, to be provided to the user.