Methods and systems for speech recognition processing using search query information

• US 8,768,698 B2
• Filed: 09/24/2013
• Issued: 07/01/2014
• Est. Priority Date: 10/18/2012
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Mengibar, Pedro J. Moreno, Sorensen, Jeffrey Scott, Weinstein, Eugene
• Primary Examiner(s): ALBERTALLI, BRIAN LOUIS
• Application Number:
  US14/035,083
• Publication Number:
  US 20140114661A1
• Time in Patent Office: 280 Days
• Field of Search: None
• US Class Current: 704/244
• CPC Class Codes: G06F 16/24522 Translation of natural language queries to structured queries
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• 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 15/14 using statistical models, e.g. Hidden Markov Models [HMMs] G10L15/18 takes precedence
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams
• CPC Class Codes: G10L 2015/0635 updating or merging of old and new templates; Mean values; Weighting

Abstract

Methods and systems for speech recognition processing are described. In an example, a computing device may be configured to receive information indicative of a frequency of submission of a search query to a search engine for a search query composed of a sequence of words. Based on the frequency of submission of the search query exceeding a threshold, the computing device may be configured to determine groupings of one or more words of the search query based on an order in which the one or more words occur in the sequence of words of the search query. Further, the computing device may be configured to provide information indicating the groupings to a speech recognition system.

Claims

1. A method, comprising:

receiving, at a computing device, information indicative of popularity of a search query comprising a sequence of words;

based on the information, determining one or more subsequences of words, each subsequence comprising one or more words of the search query based on an order in which the one or more words occur in the sequence of words of the search query; and

providing information indicative of the one or more subsequences of words to update a speech recognition system configured to convert a given spoken utterance into a given sequence of words.

2. The method of claim 1, wherein the search query includes one or more of a text string and a voice search query.

3. The method of claim 1, wherein the speech recognition system is configured to include probabilities of occurrence for given sequences of words, and wherein providing the information indicative of the one or more subsequences of words to update the speech recognition system comprises:

updating the probabilities of occurrence based on the one or more subsequences and the information indicative of the popularity of the search query.

4. The method of claim 3, wherein providing the information indicative of the one or more subsequences of words to update the speech recognition system comprises:

updating the speech recognition system to include the one or more subsequences; and

assigning to the one or more subsequences respective probabilities of occurrence that are higher than given probabilities of occurrences of other sequences of words.

5. The method of claim 1, further comprising:

for the speech recognition system to convert the given spoken utterance into the given sequence of words, constraining search space of given sequences of words in the speech recognition system to at least the one or more subsequences.

6. The method of claim 1, further comprising:

causing the speech recognition system to attempt to match the given spoken utterance to one of the one or more subsequences before attempting to match the given spoken utterance to other word sequences.

7. The method of claim 1, further comprising:

assigning respective probabilities of occurrence for the one or more subsequences based on the information indicative of the popularity of the search query, wherein the respective probabilities of occurrence are time-varying.

8. The method of claim 1, further comprising:

assigning to the one or more subsequences respective probabilities of occurrence based on the information indicative of the popularity of the search query;

continuously receiving updated information for the popularity of the search query; and

modifying the respective probabilities of occurrence based on the updated information.

9. The method of claim 8, wherein the respective probabilities of occurrence decay over time based on the updated information.

10. The method of claim 1, wherein the computing device includes the speech recognition system.

11. The method of claim 1, wherein the information indicative of the popularity of the search query is based on a frequency of submission of the search query to a search engine.

12. The method of claim 11, wherein determining the one or more subsequences of words comprises:

determining one or more subsequences of words based on the frequency of submission of the search query to the search engine exceeding a threshold.

13. A computer readable memory having stored thereon instructions that, when executed by a computing device, cause the computing device to perform functions comprising:

receiving information indicative of popularity of a search query comprising a sequence of words;

based on the information, determining one or more subsequences of words, each subsequence comprising one or more words of the search query based on an order in which the one or more words occur in the sequence of words of the search query; and

providing information indicative of the one or more subsequences of words to update a speech recognition system configured to convert a given spoken utterance into a given sequence of words.

14. The computer readable medium of claim 13, wherein the function of determining the one or more subsequences comprises:

generating a factor graph that includes automaton states and automaton arcs, each of the automaton arcs corresponding to a word from the sequence of words of the search query.

15. The computer readable medium of claim 14, wherein the function of determining a subsequence of the one or more subsequences further comprises:

selecting a first word represented by a first automaton arc connected to a first automaton state;

continuing to a second automaton state adjacent to the first automaton state; and

selecting a second word represented by a second arc connected to the second automaton state, wherein the second word is adjacent to the first word in the sequence of words of the search query.

16. The computer readable medium of claim 14, wherein a subsequence of words of the one or more subsequences is allowed to begin at any given automaton state in the factor graph.

17. A device comprising:

at least one processor;

data storage; and

program instructions in the data storage that, upon execution by the at least one processor, cause the device to perform functions comprising;

receiving information indicative of popularity of a search query comprising a sequence of words;

based on the information, determining one or more subsequences of words, each subsequence comprising one or more words of the search query based on an order in which the one or more words occur in the sequence of words of the search query; and

providing information indicative of the one or more subsequences of words to update a speech recognition system configured to convert a given spoken utterance into a given sequence of words.

18. The device of claim 17, wherein the functions further comprise:

determining a metric based on a frequency of submission of the search query to a search engine in a given period of time, wherein the function of determining the one or more subsequences is based on a comparison of the metric to a threshold.

19. The device of claim 17, wherein the function of determining the one or more subsequences comprises:

generating a factor graph that includes automaton states and automaton arcs, each of the automaton arcs corresponding to a word from the sequence of words of the search query.

20. The device of claim 19, wherein the function of determining a subsequence of the one or more subsequences comprises:

selecting a first word represented by a first automaton arc connected to a first automaton state;

continuing to a second automaton state adjacent to the first automaton state; and

select a second word represented by a second arc connected to the second automaton state, wherein the second word is adjacent to the first word in the sequence of words of the search query, and wherein the subsequence is allowed to begin at any given automaton state in the factor graph.