Automatic pruning of grammars in a multi-application speech recognition interface

• US 20080059195A1
• Filed: 08/09/2006
• Published: 03/06/2008
• Est. Priority Date: 08/09/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Brown, Christopher Ralph
• Granted Patent: US 7,899,673 B2
• Time in Patent Office: Days
• US Class Current: 704/270
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Management of multiple voice-controlled applications is provided employing registration and prioritization of applications and associated commands. In a multi-application environment, each application requests registration by a speech recognition interface. Upon registering the application(s), the interface assigns priorities to the active applications. Each application also submits a list of associated commands for registration. The interface selects a subset of submitted commands to be registered and assigned priorities. Speech recognition is performed based on the selected and prioritized commands to reduce interface overload and grammar conflict.

Claims

1. A method to be executed at least in part in a computing device for managing multiple speech enabled applications, comprising:

receiving a request for registering an application;

registering the application;

receiving a request from the registered application to register a set of commands;

selecting a subset of the received set of commands for registration;

adding the selected commands to a global command list; and

performing speech recognition operations associated with the application using the global command list.

2. The method of claim 1, further comprising:

selecting the subset of the received set of commands such that a number of commands from all registered applications does not exceed a maximum number for the global command list.

3. The method of claim 2, wherein the maximum number for the global command list is determined based on one of a predetermined constant value and a heuristic chosen at run-time based on an analysis of available system resources.

4. The method of claim 2, further comprising:

removing registered commands from the global command list, if the maximum number is exceeded based on the assigned priority levels to the commands.

5. The method of claim 1, further comprising:

receiving a set of suggested priority levels associated with the received set of commands from the registered application.

6. The method of claim 1, further comprising:

selecting the subset of the received set of commands such that a grammar conflict between the commands is prevented.

7. The method of claim 6, further comprising:

employing a comparison of commands based on one of a metric of grammatical distance and a number of distinctive phonetic characteristics to prevent the grammar conflict.

8. The method of claim 7, further comprising:

assigning a priority level to the registered application based on whether the application is one of a foreground application and a background application.

9. The method of claim 7, further comprising:

assigning a priority level to the registered application based on a type of the application.

10. The method of claim 1, further comprising:

assigning priority levels to the selected commands based on at least one from a set of;

a priority level of an associated application, a status of the associated application, a type of the associated application, a suggested priority level by the associated application, a number of already registered applications, a number of already registered commands, a status of available system resources, and a category of each command.

11. The method of claim 10, wherein the status of the associated application includes at least one from a set of:

the application being a foreground application, the application being a background application, the application being active, and a system resource associated with the application being available.

12. The method of claim 10, further comprising:

assigning the priority levels to the selected commands based on an input from one of;

a system administrator, a user, and an application with a sufficient permission level.

13. The method of claim 1, further comprising:

dynamically updating the global command list based on at least one of;

a change in a status of registered applications and a predetermined period.

14. The method of claim 1, further comprising:

tracking a frequency of use for each of the selected commands to update their priority level.

15. A computer-readable medium having computer executable instructions for providing a single speech recognition interface in a multi-application environment, the instructions comprising:

in response to a request from each application, registering a plurality of applications by assigning a priority level based on a status and a type of each application;

receiving a request from a registered application to register a set of grammar elements;

selecting a subset of the received set of grammar elements for registration such that a maximum size for a unified grammar is preserved and a conflict between the grammar elements is prevented;

assigning a priority level to each selected grammar element based on a priority level of an associated application, a status of the associated application, a type of the associated application, a suggested priority level by the associated application, a number of already registered applications, a number of already registered grammar elements, a status of available system resources, and a category of each grammar element; and

adding the selected grammar elements to the unified grammar.

16. The computer-readable medium of claim 15, wherein the instructions further comprise:

registering at least one synonym for each registered grammar element; and

removing registered grammar elements and their synonyms based on at least one from a set of;

a priority level associated with the grammar element, a priority level of an associated application, a status of the associated application, a type of the associated application, a suggested priority level by the associated application, a number of already registered applications, a number of already registered grammar elements, a status of available system resources, and a category of each grammar element.

17. The computer-readable medium of claim 15, wherein the instructions further comprise:

in response to the maximum size for the unified grammar being exceeded, registering a new grammar element;

moving a set of grammar elements associated with an application to a submenu such that the grammar elements in the submenu are accessible through the new grammar element; and

removing the grammar elements in the submenu from the unified grammar.

18. A system for managing multiple speech enabled applications, comprising:

a speech recognition engine configured to convert audio signals to textual data;

a grammar module configured to;

in response to a request from an application, register the application by assigning a priority level based on a status and a type of each application;

receive a request from the registered application to register a set of commands;

select a subset of the received set of commands for registration such that a maximum size for a global command list is preserved and a conflict between the commands is prevented;

assign a priority level to each selected command based on a priority level of the application, a status of the application, a type of the application, a suggested priority level for the command by the application, a number of already registered applications, a number of already registered commands, a status of available system resources, and a category of each command;

add the selected commands to the global command list; and

parse commands for the speech enabled applications from the textual data employing the command list.

19. The system of claim 18, wherein the speech recognition engine is further configured to parse the commands based on a feedback from the grammar module.

20. The system of claim 18, wherein the grammar module is further configured to remove registered commands from the global command list, if the maximum size is exceeded based on the assigned priority levels to the commands.