Systems and methods for managing multiple grammars in a speech recognition system

• US 7,720,678 B2
• Filed: 12/16/2005
• Issued: 05/18/2010
• Est. Priority Date: 02/04/2002
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Miller, David, Falcon, Steve, Yip, Clement, Banay, Dan
• Primary Examiner(s): Armstrong; Angela A
• Application Number:
  US11/275,193
• Publication Number:
  US 20060161429A1
• Time in Patent Office: 1,614 Days
• Field of Search: 704/231, 704/251, 704/270, 704/275
• US Class Current: 704/231
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Systems and methods are described for a speech system that manages multiple grammars from one or more speech-enabled applications. The speech system includes a speech server that supports different grammars and different types of grammars by exposing several methods to the speech-enabled applications. The speech server supports static grammars that do not change and dynamic grammars that may change after a commit. The speech server provides persistence by supporting persistent grammars that enable a user to issue a command to an application even when the application is not loaded. In such a circumstance, the application is automatically launched and the command is processed. The speech server may enable or disable a grammar in order to limit confusion between grammars. Global and yielding grammars are also supported by the speech server. Global grammars are always active (e.g., “call 9-1-1”) while yielding grammars may be deactivated when an interaction whose grammar requires priority is active.

Claims

1. A speech system, comprising:

memory;

a processor coupled to the memory for managing multiple speech-enabled applications;

a speech engine configured to recognize commands from a user and make announcements to the user;

a speech server having a speech server interface through which multiple speech-enabled applications communicate with the speech system, and a speech application programming interface through which the speech server communicates with the speech engine;

wherein the speech server manages concurrent processing of interactions submitted by the speech-enabled applications while allowing each speech-enabled application to utilize a different speech recognition grammar;

a master grammar table comprising a speech recognition grammar that is registered with the speech server by the speech-enabled applications;

the speech server passing a recognition of a grammar upon receiving an indication that the recognition of the grammar occurred in a speech interaction to a corresponding speech-enabled application that registered the grammar;

wherein when the corresponding speech-enabled application is unloaded and not running, the speech server automatically loads unloaded speech-enabled application in response to receiving a command included in the grammar used by the unloaded speech-enabled application;

an interaction manager moves speech interactions that are considered a high priority to a beginning of an order to be processed, receiving a high priority speech interaction and interrupting a speech interaction that is being processed when the high priority speech interaction is received, while other speech interactions are processed as submitted, on a first-in-first-out basis, wherein the order to be processed is maintained in an interaction list by the interaction manager; and

the interaction manager establishes a chain of interactions that constitutes a conversation, wherein the interaction manager waits during a pre-determined grace period or time out for a submission of a subsequent interaction in the conversation before processing a next interaction in the interaction list;

wherein the speech server includes a self destruct flag that indicates a current interaction terminates when the interaction is interrupted.

2. The speech system as recited in claim 1, wherein the speech server is configured to support persistent grammars for the unloaded speech-enabled application.

3. The speech system as recited in claim 1, wherein the speech server is further configured to:

support yielding grammars for the speech-enabled applications; and

deactivate all but one active yielding grammar.

4. The speech system as recited in claim 1, wherein the speech server is further configured to:

support yielding grammars and global grammars for the speech-enabled applications; and

activate the global grammars and continuously monitor for recognitions in the global grammars.

5. The speech system as recited in claim 1, further comprising a grammar table for each grammar used by the speech-enabled applications, each grammar table containing one or more grammar attributes for the grammar with which it associated.

6. The speech system as recited in claim 5, wherein a grammar attribute in each grammar table is a grammar identifier that uniquely identifies the grammar associated with the grammar table.

7. The speech system as recited in claim 5, wherein a grammar attribute in each grammar table is a global flag that, if set, indicates that the grammar associated with the grammar table is a global grammar that may not be deactivated by the speech server.

8. The speech system as recited in claim 5, wherein:

a grammar attribute in each grammar table is a persistent flag that, if set, indicates that the grammar associated with the grammar table is a persistent grammar that may be launched by the speech server when the speech server recognizes a command belonging to the grammar; and

an application associated with the persistent grammar is launched by the speech server when the speech server recognizes a command belonging to the grammar.

9. The speech system as recited in claim 5, wherein a grammar attribute in each grammar table is an active flag that, if set, indicates that the grammar associated with the grammar is currently active.

10. The speech system as recited in claim 5, further comprising the master grammar table that contains each grammar table used by the speech recognition applications.

11. A method, having computer-executable instructions to be executed by a computing device, the method comprising:

managing multiple speech-enabled applications;

registering a grammar by the multiple speech-enabled applications in a master grammar table, the master grammar table comprising a grammar table;

receiving a first speech interaction from a first speech-enabled application that utilizes a first grammar, wherein a recognition of the first grammar is passed to the first speech-enabled application that registered the first grammar in the master grammar table;

identifying the first grammar associated with the first speech interaction;

processing the first speech interaction according to properties attributable to the identified first grammar;

receiving a second speech interaction from a second speech-enabled application that utilizes a second grammar, the second speech interaction being received while the first speech interaction is processing;

processing the second speech interaction according to properties attributable to the second grammar;

wherein the processing of the first speech interaction is interrupted and the processing of the second speech interaction is immediately commenced when an indication is detected directing that the second speech interaction be processed immediately;

if the second speech-enabled application that registered the second grammar in the master grammar table is not loaded, then automatically loading the second speech-enabled application to execute one or more commands associated with the second speech interaction;

moving the speech interaction that is considered a high priority to a beginning of an order to be processed, while other speech interactions are processed in the order in which the speech interactions are submitted, on a first-in-first-out basis, wherein the order to be processed is maintained in an interaction list;

establishing a chain of interactions that constitutes a conversation, wherein a subsequent interaction in the conversation is waited for during a pre-determined grace period or time out before processing a next interaction in the interaction list; and

setting a self destruct flag that indicates when a current speech interaction is interrupted, the interaction is terminated.

12. The method as recited in claim 11, wherein a grammar attribute included in the grammar table in the master grammar table is an active flag that, when set, indicates that a speech grammar associated with the grammar table is currently active and, when not set, indicates that the speech grammar associated with the grammar table is currently inactive.

13. The method as recited in claim 11, wherein the processing the second speech interaction further comprises processing the second speech interaction after the first speech interaction has concluded processing when a grammar attribute in the master grammar table corresponding to the second grammar indicates that the second grammar is a yielding grammar.

14. The method as recited in claim 11, wherein a grammar attribute included in the grammar table in the master grammar table is a grammar identifier that uniquely identifies a speech grammar associated with the grammar table.

15. The method as recited in claim 11, wherein a grammar attribute included in the grammar table in the master grammar table is an executable command that, when executed, launches a speech recognition application associated with the grammar table.

16. The method as recited in claim 11, wherein a grammar attribute included in the grammar table in the master grammar table is a global flag that, when set, indicates that a speech grammar associated with the grammar table is a global grammar that may only be deactivated by a speech-enabled application that submitted the speech interaction and, when not set, indicates that the speech grammar associated with the grammar table is a yielding grammar.

17. The method as recited in claim 11, wherein a grammar attribute included in the grammar table in the master grammar table is a persistent flag that, when set, indicates that a speech grammar associated with the grammar table is a persistent grammar that is active even when a speech recognition application associated with the speech grammar is not loaded.

18. One or more computer-readable storage media containing computer-executable instructions that, when executed on a computer, perform the following steps:

managing multiple speech-enabled applications;

registering a grammar by the multiple speech-enabled applications in a master grammar table;

receiving a first interaction from a first speech-enabled application that utilizes a first grammar, wherein a recognition of the first grammar is passed to the first speech-enabled application that registered the first grammar in the master grammar table;

processing the first interaction according to properties attributable to the first grammar;

receiving a second interaction from a second speech-enabled application that utilizes a second grammar, the second interaction being received while the first interaction is processing;

processing the second interaction according to properties attributable to the second grammar;

wherein the processing of the first interaction is interrupted and processing of the second interaction is immediately commenced if an indication is detected directing that the second interaction be processed immediately;

moving the speech interaction that is considered a high priority to a beginning of an order, while other speech interactions are processed in the order submitted, on a first-in-first-out basis order;

if the second speech-enabled application that registered the second grammar in the master grammar table is not loaded, then automatically loading the second speech-enabled application to execute one or more commands associated with the second interaction; and

setting a self destruct flag that indicates when a current speech interaction is interrupted, the interaction is terminated.

19. The one or more computer-readable storage media as recited in claim 18, wherein the indication directing that the second interaction be processed immediately is a flag contained in the second interaction.