Voice browser with weighting of browser-level grammar to enhance usability

• US 7,302,392 B1
• Filed: 10/07/2003
• Issued: 11/27/2007
• Est. Priority Date: 10/07/2003
• Status: Active Grant

• Current Assignee: Sprint Spectrum LP (Deutsche Telekom AG)
• Original Assignee: Sprint Spectrum LP (Deutsche Telekom AG)
• Inventors: Thenthiruperai, Balaji S., Roche, Elizabeth R., Kim, Hyong-Gyun
• Primary Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US10/680,961
• Time in Patent Office: 1,512 Days
• Field of Search: 704/251, 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 2015/228 of application context

Abstract

A computer system in the form of a voice command platform includes a voice browser and voice-based applications. The voice browser has global-level grammar elements and the voice applications have application-level grammar and grammar elements. A programming feature is provided by which developers of the voice applications can programmably weigh or weight global-level grammar elements relative to the application-level grammar or grammar elements. As a consequence of the weighting, a speech recognition engine for the voice browser is more likely to accurately recognize voice input from a user. The weighting can be applied on the application as a whole, or at any given state in the application. Also, the weighting can be made to the global level grammar elements as a group, or on an individual basis.

Claims

1. In a computer system comprising a processor and software comprising a set of instructions stored on a machine-readable medium implementing a voice command platform for voice-based applications, said voice command platform including a speech recognition engine, a voice browser having global-level grammar comprising a set of grammar elements and said voice applications having application-level grammar comprising a set of application-level grammar elements, the improvement comprising:

providing in said voice command platform a programming feature by which developers of said voice applications can weight global-level grammar elements relative to the application-level grammar and/or application level grammar elements,

whereby said speech recognition engine is more likely to properly determine whether voice input from a user phonetically similar to said global-level grammar elements and/or said application level grammar or grammar elements should trigger events associated with said global-level grammar elements or said application level grammar or grammar elements depending on the weighting of said global level grammar elements.

2. The improvement of claim 1, wherein said weighting programming feature comprises a variable having a value of either “

low”

or “

high”

.

3. The improvement of claim 1, wherein said weighting programming feature is provided in said voice command platform such that it is available at a plurality of application states in said voice applications.

4. The improvement of claim 1, wherein said weighting programming feature comprises a variable having a numerical value.

5. The improvement of claim 1, wherein said weighting feature is programmably applied to said voice application as a whole.

6. The improvement of claim 1, wherein the improvement further comprises providing a default/override table for resolving conflicts where the voice input from the user matches both a global level grammar element and an application level grammar element, the default/override table determining which grammar element controls the processing of the voice input.

7. The improvement of claim 1, wherein said global level grammar elements are individually weighted.

8. The improvement of claim 1, wherein said global level grammar elements are weighted as a group.

9. The improvement of claim 1, wherein the weighting is implemented by plurality of applications executed by said voice command platform.

10. A computer program comprising a set of instructions stored on a machine-readable medium implementing a voice-based application for execution in a voice command platform, said voice command platform including a speech recognition engine and a voice browser having global-level grammar elements, said voice-based application having application-level grammar and application level grammar elements, the set of instructions comprising:

instructions weighting one or more global-level grammar elements relative to the application-level grammar and/or application level grammar elements,

whereby said speech recognition engine is more likely to properly determine whether voice input from a user phonetically similar to said global-level grammar elements and/or said application level grammar or grammar elements should trigger events associated with said global-level grammar elements or said application level grammar or grammar elements depending on the weighting of said global level grammar elements.

11. The computer program of claim 10, wherein said instructions weighting one or more global level grammar elements comprises a variable having a value of either “

low”

or “

high”

.

12. The computer program of claim 10, wherein said instructions weighting one or more global level grammar elements are provided at a plurality of application states in said voice application.

13. The computer program of claim 10, wherein said instructions weighting one or more global level grammar elements comprises a variable having a numerical value.

14. The computer program of claim 10, wherein said instructions weighting said global level grammar elements are applied to said voice-based application as whole.

15. The computer program of claim 10, wherein the instructions further comprise instructions referring to a default/override table for resolving conflicts where the voice input from the user matches both a global level grammar element and an application level grammar element, the default/override table determining which grammar element controls the processing of the voice input.

16. The computer program of claim 10, wherein said global level grammar elements are individually weighted.

17. The computer program of claim 10, wherein said global level grammar elements are weighted as a group.

18. A method of processing speech input in a voice application executed by a voice command platform, said voice command platform comprising a speech recognition engine and a voice browser having global-level grammar comprising a set of global level grammar elements, said voice application comprising application-level grammar comprising a set of application level grammar elements, the method comprising the steps of:

a) receiving an utterance comprising speech input from a user;

b) applying a weighting of global level grammar elements relative to application level grammar elements;

c) performing a speech recognition process to determine whether the utterance is recognized by a speech recognition engine and the confidence level at which said utterance is recognized;

d) if said utterance matches a global level grammar element and an application level grammar element, invoking an event handler for said utterance by reference to a default/override table; and

e) if said utterance matches only one of a global level grammar element or an application level grammar element, invoking an event handler associated with the global level grammar element or application level grammar element that matches the utterance.

19. The method of claim 18, wherein said weighting of global level grammar elements is performed by processing a variable having a value of either “

low”

or “

high”

.

20. The method of claim 18, wherein said weighting of global level grammar elements is provided at a plurality of application states in said voice application.

21. The method of claim 18, wherein said weighting of global level grammar elements is performed by processing a variable having a numerical value.

22. The method of claim 18, wherein said weighting of global level grammar elements are applied to said voice application as whole.

23. The method of claim 18, wherein said global level grammar elements are individually weighted.

24. The method of claim 18, wherein said global level grammar elements are weighted as a group.

25. A method of processing speech input in a voice application executed by a voice command platform, said voice command platform comprising a speech recognition engine and a voice browser having global-level grammar comprising a set of global level grammar elements, said voice application comprising application-level grammar comprising a set of application level grammar elements, the method comprising the steps of:

a) receiving an utterance comprising speech input from a user;

b) applying a weighting of global level grammar elements relative to application level grammar elements;

c) performing a speech recognition process to determine whether the utterance is recognized by a speech recognition engine and the confidence level at which said utterance is recognized;

d) if said utterance is not recognized in step c), invoking an out of grammar event handler for said utterance.

26. The method of claim 25, wherein said weighting of global level grammar elements is performed by processing a variable having a value of either “

low”

or “

high”

.

27. The method of claim 25, wherein said weighting of global level grammar elements is provided at a plurality of application states in said voice application.

28. The method of claim 25, wherein said weighting of global level grammar elements is performed by processing a variable having a numerical value.

29. The method of claim 25, wherein said weighting of global level grammar elements are applied to said voice application as whole.

30. The method of claim 25, wherein said global level grammar elements are individually weighted.

31. The method of claim 25, wherein said global level grammar elements are weighted as a group.

32. A method of processing speech input in a voice application executed by a voice command platform, said voice command platform comprising a speech recognition engine and a voice browser having global-level grammar comprising a set of global level grammar elements, said voice application comprising application-level grammar comprising a set of application level grammar elements, the method comprising the steps of:

a) receiving an utterance comprising speech input from a user;

b) applying a weighting of global level grammar elements relative to application level grammar elements;

c) performing a speech recognition process to determine whether the utterance is recognized by a speech recognition engine and the confidence level at which said utterance is recognized, said speech recognition process determining whether the utterance matches any global level grammar elements and/or application level grammar elements.

33. The method of claim 32, wherein said weighting of global level grammar elements is provided at a plurality of application states in said voice application.

34. The method of claim 32, wherein said weighting of global level grammar elements is performed by processing a variable having a numerical value.

35. The method of claim 32, wherein said weighting of global level grammar elements are applied to said voice application as whole.

36. The method of claim 32, wherein said global level grammar elements are individually weighted.

37. The method of claim 32, wherein said global level grammar elements are weighted as a group.