SYSTEM FOR RECOGNIZING SPEECH FOR SEARCHING A DATABASE

• US 20090112593A1
• Filed: 10/24/2008
• Published: 04/30/2009
• Est. Priority Date: 10/24/2007
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Harman Becker Automotive Systems GmbH (Samsung Electronics Co. Ltd.)
• Inventors: Low, Andreas, Haiber, Udo, Konig, Lars
• Granted Patent: US 8,380,505 B2
• Time in Patent Office: Days
• US Class Current: 704/251
• CPC Class Codes: G06F 16/2423 Interactive query statement specification based on a database schema
• CPC Class Codes: G06F 16/243 Natural language query formulation
• CPC Class Codes: G06F 16/332 Query formulation
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

A system is provided for recognizing speech for searching a database. The system receives speech input as a spoken search request and then processes the speech input in a speech recognition step using a vocabulary for recognizing the spoken request. By processing the speech input words recognized in the speech input and included in the vocabulary are obtained to form at least one hypothesis. The hypothesis is then utilized to search a database using the at least one hypothesis as a search query. A search result is then received from the database and provided to the user.

Claims

1. A method of recognizing speech for searching a database, where a user enters a search request as a speech input, comprising the following steps:

processing the speech input in a speech recognition step, where at least one hypothesis is obtained comprising recognized words and pseudo-words not recognized as words; and

searching the database using the at least one hypothesis as a search query, where a list of search results is obtained.

2. The method of claim 1, where the pseudo-words are words that are not included in a vocabulary used to recognize words of the speech input.

3. The method of claim 1, where processing the speech input obtains a plurality of hypotheses.

4. The method of claim 1, where the speech recognition step includes the use of a speech model to recognize words.

5. The method of claim 1, where a plurality of pseudo-word hypotheses are generated for a pseudo-word, resulting in a plurality of hypotheses and search queries.

6. The method of claim 1, where a vocabulary is utilized to recognize the words of the speech input and includes one or a combination of the following:

words obtained from a list of the most commonly used search queries for a database, words obtained from a list of the most commonly spoken search requests, pronunciation alternatives of words, words obtained from user operated applications.

7. The method of claim 1, where an exclusion vocabulary is utilized to processing the speech input and to exclude words from the at least one hypothesis.

8. The method of claim 1, where processing the speech input further comprises the step of deriving pseudo-word properties for a pseudo-word.

9. The method of claim 8, where searching the database comprises deriving entry properties from database entries and comparing the pseudo-word properties with the entry properties to find entries most closely corresponding to the pseudo-words.

10. The method of claim 9, where searching the database comprises generating pre-processed database entries by forming entry alternatives and comparing the pre-processed database entries with the at least one hypothesis.

11. The method of claim 9, where entry alternatives are obtained by forming orthographic or pronunciation alternatives of database entries.

12. The method of claim 1, further comprising the steps of estimating the language of words comprised in the at least one hypothesis and performing the search in the database taking into account the estimated language.

13. The method of claim 1, further comprising the step of identifying semantic information in the speech input and performing the search taking into account the semantic information.

14. The method of claim 1, where the search results are weighted in the list of search results, and that the search results are provided to the user according to their weighting.

15. The method of claim 1, further comprising the step of identifying candidate-words in the list of search results as most probable words corresponding to the pseudo-words.

16. The method of claim 15, further comprising the step of providing an extended vocabulary including the candidate words.

17. The method of claim 16, further comprising the steps of recognizing words in the speech input using the extended vocabulary and reordering the list of search results taking into account the recognized words, where a new list of search results is obtained.

18. The method of claim 16, further comprising the steps of recognizing words in the speech input using the extended vocabulary to obtain at least one hypothesis, and performing a search of the database using the at least one hypothesis as a search query, where a new list of search results is obtained

19. The method of claim 17, where the new list of search results is provided to the user after a predetermined amount of time.

20. The method of claim 1, further comprising the step of excluding or devaluating search results or words included in the search query by user input.

21. The method of claim 1, where the database is searched using a search method configured to deliver search results including words that are selected by the search method to be most probable words corresponding to the pseudo-words.

22. A system for recognizing speech for searching a database, comprising:

a speech recognition unit detecting a search request of a user in form of a speech input, the speech recognition unit being formed so as to recognize the speech input, where at least one hypothesis is obtained including recognized words and pseudo-words not recognized as words;

a database; and

a search unit searching the database using the at least one hypothesis as a search query, where a list of search results is obtained.

23. The system of claim 22, where the search unit is formed so as to provide a list of candidate words that are included in the search results and associated with pseudo-words of the hypothesis.

24. The system of claim 23, further including a first reprocessing unit formed so as to recognize words in the speech input using an extended vocabulary including the candidate words and to reorder the list of search results depending on the result of the recognition.

25. The system of claim 23, further including a second reprocessing unit formed so as to recognize words in the speech input using an extended vocabulary including the candidate words to obtain at least one hypothesis, and to provide said hypothesis to the search unit to be utilized as a search query.

26. The system of claim 22, where the speech recognition unit is furthermore formed so as to provide pseudo-word properties associated with pseudo-words, where the search unit is furthermore formed so as to compare the pseudo-word properties to properties derived from data base entries or entry alternatives when searching the database.

27. A method of recognizing speech for searching a database, comprising the steps of:

receiving a speech input comprising a spoken search request;

processing the speech input in a speech recognition step using a vocabulary for recognizing the spoken request, thereby obtaining words recognized in the speech input and included in the vocabulary and pseudo words, which pseudo words are words of the speech input not recognized using the vocabulary;

forming at least one hypothesis including at least one of said pseudo words;

searching the database using the at least one hypothesis as a search query; and

receiving a list of search results from the database.