Discrete speech recognition system with ballooning active grammar

• US 6,301,560 B1
• Filed: 01/05/1998
• Issued: 10/09/2001
• Est. Priority Date: 01/05/1998
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Masters, Steven P.
• Primary Examiner(s): Korzuch, William
• Assistant Examiner(s): Lerner, Martin
• Application Number:
  US09/003,055
• Time in Patent Office: 1,373 Days
• Field of Search: 704/231, 704/251, 704/254, 704/255, 704/270, 704/275
• US Class Current: 704/251
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering
• CPC Class Codes: G10L 2015/225 Feedback of the input speech

Abstract

A discrete speech recognition system has a default grammar with keywords and non-keywords. Upon detecting a non-keyword, the speech recognition system performs the function associated with the detected word. Upon detecting a keyword in the default grammar, the speech recognition system temporarily expands its active grammar list from the default grammar to a ballooned grammar that includes both the words in the default grammar and the additional words triggered by detection of the keyword. In this manner, the operator still has the option to select a word from the original grammar, or choose a word from the additional second-tier list. The ballooned grammar remains active until the operator makes a new selection. The speech recognition system then returns to the default grammar.

Claims

1. A discrete speech recognition system comprising:

an application;

a vocabulary accessible by the application, the vocabulary holding a set of utterances applicable to the application;

a speech recognition engine to recognize the utterances in the vocabulary;

a grammar that holds a subset of the utterances in the vocabulary that is active at a particular instance for the speech recognition system to detect;

wherein during a first instance, the grammar contains a default subset of utterances from the vocabulary, the default subset of utterances including at least one keyword utterance associated with a second subset of utterances from the vocabulary, the default subset of utterances further including at least one non-keyword utterance; and

wherein during a second instance following detection by the speech recognition engine of the keyword utterance, the grammar is expanded to contain both the default subset of utterances and the second subset of utterances from the vocabulary, wherein the grammar remains expanded until a non-keyword utterance is detected.

2. A discrete speech recognition system as recited in claim 1, wherein the grammar remains expanded until an utterance from the expanded grammar is detected.

3. A discrete speech recognition system as recited in claim 1, wherein the grammar is returned to the default grammar after an utterance from the expanded grammar is detected.

4. A discrete speech recognition system as recited in claim 1, wherein the second subset of utterances contains another keyword associated with a third subset of utterances from the vocabulary, and during a third instance following detection of said another keyword, the grammar is expanded a second time to contain the default and second subsets of utterances and the third subset of utterances from the vocabulary.

5. A discrete speech recognition system as recited in claim 4, wherein selected utterances from the default or second subsets of utterances are removed from the doubly expanded grammar.

6. An entertainment system incorporating the discrete speech recognition system as recited in claim 1.

7. A computing device incorporating the discrete speech recognition system as recited in claim 1.

26. A discrete speech recognition system as recited in claim 1, wherein all utterances in the default subset of utterances are keyword utterances or non-keyword utterances.

8. A discrete speech recognition system having an active grammar, the active grammar including a default set of utterances, the default set of utterances including at least one keyword utterance associated with a new set of utterances and at least one non-keyword utterance;

wherein the discrete speech recognition system is configured to detect the keyword utterance and in response to expand the active grammar by temporarily adding the new set of utterances to the active grammar so that the active grammar used by the discrete speech recognition system temporarily includes the default set of utterances and the new set of utterances; and

wherein the active grammar is reduced to the default set of utterances upon detection of a non-keyword utterance.

9. The discrete speech recognition system as recited in claim 8, wherein the active grammar is reduced back to include only the default set of utterances after detection of an utterance from the default set of utterances or the new set of utterances.

10. A vehicle system comprising:

an entertainment system for providing entertainment to occupants of the vehicle; and

the discrete speech recognition system as recited in claim 8, configured to detect utterances used to control the entertainment system.

27. A discrete speech recognition system as recited in claim 8, wherein utterances in the default set of utterances that are not keyword utterances are non-keyword utterances.

11. A vehicle computer system comprising:

a computer;

an operating system executing on the computer, the operating system being configured to support multiple applications that are supplied by a vehicle user; and

a discrete speech recognition system to detect utterances used to control at least one of the applications running on the computer, the speech recognition system having an active grammar that contains a default set of utterances, the default set of utterances having at least one keyword and at least one non-keyword;

whereupon detection of the keyword, the speech recognition system is configured to expand the active grammar to include the default set of utterances and a new set of utterances, the active grammar remains expanded until a non-keyword utterance is detected.

12. A vehicle computer system as recited in claim 11, wherein the discrete speech recognition system reduces the active grammar back to the default grammar after an utterance from the expanded grammar is detected.

13. A vehicle computer system as recited in claim 11, wherein the new set of utterances contains another keyword and whereupon detection of said another keyword, the speech recognition system is configured to expand the active grammar a second time to include the default set of utterances, the new set of utterances, and a third set of utterances.

14. A vehicle computer system as recited in claim 13, wherein the speech recognition system removes selected utterances from the active grammar.

28. A vehicle computer system as recited in claim 11, wherein the default set of utterances that are not keyword utterances are non-keyword utterances.

15. In a discrete speech recognition system, a method comprising the following steps:

loading an active grammar with a default set of utterances, the utterances having associated functions, the set of utterances containing at least one keyword associated with a new set of utterances, the set of utterances further containing at least one non-keyword;

detecting an utterance from the active grammar; and

in an event that the detected utterance is a keyword, expanding the active grammar to include the default set of utterances and the new set of utterances, the active grammar remaining expanded until a non-keyword utterance is detected.

16. A method as recited in claim 15, further comprising the following additional steps:

detecting an utterance from the expanded active grammar; and

reducing the active grammar to include only the default set of utterances.

17. A method as recited in claim 15, wherein the new set of utterances contains another keyword and further comprising the following additional steps:

detecting an utterance from the expanded active grammar; and

in an event that the utterance detected from the expanded grammar is a keyword associated with a third set of utterances, expanding the active grammar to include the default set of utterances, the new set of utterances, and the third set of utterances.

18. A method as recited in claim 15, further comprising the step of removing selected utterances from the active grammar that is twice-expanded to include the default set of utterances, the new set of utterances, and the third set of utterances.

19. A computer-readable medium having computer-executable instructions for performing the steps in the method recited in claim 15.

29. A method as recited in claim 15, wherein all utterances in the default set of utterances are non-keyword utterances or keyword utterances.

20. In a discrete speech recognition system having an active grammar, the active grammar including a default set of utterances, and the default set of utterances including at least one keyword associated with a new set of utterances and including at least one non-keyword, a method comprising the step of ballooning the active grammar to temporarily include both the default set of utterances and the new set of utterances upon detection of the keyword, the active grammar remaining ballooned until a non-keyword utterance is detected.

21. A method recited in claim 20, further comprising the step of reducing the active grammar to include only the default set of utterances following recognition of an utterance from the active grammar.

22. A computer-readable medium having computer-executable instructions for performing the steps in the method recited in claim 20.

30. A method as recited in claim 20, wherein all utterances in the default set of utterances are either keyword utterances or non-keyword utterances.

23. One or more computer-readable media having stored thereon a computer program that, when executed by a processor, causes the processor to:

load an active grammar with a default set of utterances, the utterances having associated functions, the default set of utterances containing at least one keyword associated with a new set of utterances, the default set of utterances further containing at least one non-keyword;

detect an utterance from the active grammar; and

expand the active grammar to include the default set of utterances and the new set of utterances if the detected utterance is a keyword, the active grammar remaining expanded until a non-keyword utterance is detected.

24. One or more computer-readable media as recited in claim 23 wherein the computer program further causes the processor to:

detect an utterance from the expanded active grammar; and

reduce the active grammar to include only the default set of utterances.

25. One or more computer-readable media as recited in claim 23 wherein the new set of utterances contains at least one keyword associated with a third set of utterances.

31. One or more computer-readable media as recited in claim 23, wherein the default set of utterances that are not keyword utterances are non-keyword utterances.