Performing a safety analysis for user-defined voice commands to ensure that the voice commands do not cause speech recognition ambiguities

• US 8,234,120 B2
• Filed: 07/26/2006
• Issued: 07/31/2012
• Est. Priority Date: 07/26/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Agapi, Ciprian, Blass, Oscar J., Monteiro, Brennan D., Vila, Roberto
• Primary Examiner(s): Smits, Talivaldis Ivars
• Assistant Examiner(s): Kazeminezhad, Farzad
• Application Number:
  US11/460,075
• Publication Number:
  US 20080046250A1
• Time in Patent Office: 2,197 Days
• Field of Search: 704/275, 704/256.2, 395/333, 709/206, 345/156
• US Class Current: 704/275
• CPC Class Codes: G10L 15/075 supervised, i.e. under machine guidance

Abstract

The present invention discloses a solution for assuring user-defined voice commands are unambiguous. The solution can include a step of identifying a user attempt to enter a user-defined voice command into a voice-enabled system. A safety analysis can be performed on the user-defined voice command to determine a likelihood that the user-defined voice command will be confused with preexisting voice commands recognized by the voice-enabled system. When a high likelihood of confusion is determined by the safety analysis, a notification can be presented that the user-defined voice command is subject to confusion. A user can then define a different voice command or can choose to continue to use the potentially confusing command, possibly subject to a system imposed confusion mitigating condition or action.

Claims

1. A method for use in a voice-enabled system that permits a user to define one or more new voice commands, the method comprising acts of:

identifying a user attempt to define a new voice command that is not yet defined in the voice-enabled system, the user attempt comprising a proposed voice command to be associated with a voice-enabled capability in the voice-enabled system, wherein the user attempt fully specifies the proposed voice command;

in response to the user providing the proposed voice command as part of the user attempt to define a new voice command, performing, by at least one processor, a safety analysis on the proposed voice command to determine a likelihood that the proposed voice command would be confused with at least one existing voice command that the voice-enabled system was programmed to recognize prior to the act of identifying a user attempt to define a new voice command, wherein the safety analysis is performed after the act of identifying the user attempt; and

parsing the proposed voice command into a plurality of component parts, wherein the safety analysis is performed against at least one component part; and

when a high likelihood of confusion is determined by the safety analysis, presenting a notification that the proposed voice command is subject to potential confusion.

2. The method of claim 1, wherein the voice-enabled capability comprises a user-defined abstraction.

3. The method of claim 2, further comprising acts of:

accepting the proposed voice command as a new voice command for the user-defined abstraction after the safety analysis is performed;

after the act of accepting the proposed voice command, recording a plurality of actions associated with the user-defined abstraction; and

storing the plurality of actions in association with the new voice command to enable the user to subsequently invoke an execution of the plurality of actions by speaking the new voice command.

4. The method of claim 1, further comprising acts of:

when a high likelihood of confusion is determined by the safety analysis, presenting a warning message to the user that the proposed voice command is likely to be confused with a different voice command; and

permitting the user to retain the proposed voice command as a new voice command recognized by the voice-enabled system.

5. The method of claim 4, further comprising acts of:

requiring the user to preface the new voice command with a keyword when invoking the new voice command; and

allowing other user-defined commands that do not have a high likelihood of confusion as determined by the safety analysis to be invoked without being prefaced with a keyword.

6. The method of claim 1, further comprising an act of:

when a low likelihood of confusion is determined by the safety analysis, accepting the proposed voice command as a new voice command recognized by the voice-enabled system.

7. The method of claim 1, further comprising an act of:

providing an option to propose a different voice command, when a high likelihood of confusion is determined by the safety analysis.

8. The method of claim 1, further comprising acts of:

when a high likelihood of confusion is determined by the safety analysis and when a user elects to accept the proposed voice command as a new voice command recognized by the voice-enabled system, performing at least one programmatic action to reduce recognition ambiguities between the at least one existing voice command and the new voice command.

9. The method of claim 1, wherein the safety analysis is performed against each component part.

10. The method of claim 1, further comprising acts of:

receiving a service request;

selecting at least one service agent; and

manipulating, by the at least one service agent, at least one computing device to enable a safety analysis function.

11. A method of adding user-defined abstractions to a voice-enabled system, the method comprising acts of:

detecting an attempt to record a new abstraction;

receiving a proposed voice command for use when invoking the new abstraction, the proposed voice command being proposed as part of the attempt to record the new abstraction, wherein receiving the proposed voice command comprises receiving the entire proposed voice command;

in response to the act of receiving the proposed voice command for use when invoking the new abstraction, comparing the proposed voice command against a set of existing voice commands that the voice-enabled system was programmed to recognize prior to the act of receiving the proposed voice command to determine whether the proposed voice command is likely to be confused with at least one of the existing voice commands, wherein the act of comparing the proposed voice command against the set of existing commands is performed after the act of receiving the proposed voice command;

parsing the proposed voice command into a plurality of component parts, wherein each the act of comparing the proposed voice command against the existing voice commands comprises comparing at least one component part against one or more of the existing voice commands;

when it is determined that the proposed voice command is likely to be confused with at least one of the existing voice commands, performing at least one programmatic action designed to reduce a likelihood of an occurrence of a potential recognition ambiguity between the proposed voice command and the at least one of the existing voice commands;

recording a plurality of actions associated with the new abstraction; and

associating the new abstraction with the proposed voice command to enable a user to subsequently invoke an execution of the plurality of actions by speaking the proposed voice command.

12. The method of claim 11, wherein said at least one programmatic action comprises an action selected from a set of actions consisting of:

substituting a different voice command for the proposed voice command;

limiting a recognition of the proposed voice command to situations where the proposed voice command is prefaced with a keyword; and

placing the proposed voice command in a first grammar different from a second grammar containing the at least one of the existing voice commands where the first and second grammars are not simultaneously activated by the voice-enabled system.

13. The method of claim 11, further comprising acts of:

receiving a service request;

selecting at least one service agent; and

manipulating, by the at least one service agent, at least one computing device to enable a safety analysis function.

14. A voice-enabled system comprising:

at least one processor programmed to implement;

a speech recognition engine configured to accept speech, to compare the speech with voice commands in a grammar, and to automatically execute programmatic actions associated with at least one voice command responsive to a match between the speech and the at least one voice command; and

a safety analysis engine configured to, in response to and after the at least one processor receiving a proposed voice command for use in invoking at least one action in the voice-enabled system, the proposed voice command being received in its entirety and being not yet defined in the voice-enabled system as a command for invoking the at least one action;

parse the proposed voice command into a plurality of component parts;

determine, based at least in part on a comparison of at least one component part of the plurality of component parts against one or more existing voice commands in the grammar, a likelihood that the proposed voice command would be confused with at least one existing voice command in the grammar so as to determine whether the proposed voice command is permitted to be added to the grammar.

15. The system of claim 14, wherein the at least one processor is further programmed to implement:

a confusion mitigation engine configured to automatically execute at least one programmatic action designed to reduce the likelihood of confusion between the proposed voice command and the at least one existing voice command in the grammar.

16. The system of claim 15, wherein the confusion mitigation engine is configured to present a notification to a user that the proposed voice command is subject to potential confusion whenever a high likelihood of confusion is determined by the safety analysis engine.

17. The system of claim 15, wherein the confusion mitigation engine is configured to, when the safety analysis engine determines that the proposed voice command has a high likelihood of confusion with at least one existing voice command in the grammar, require the proposed voice command to be preceded by a keyword in order to be invoked, wherein at least one other user-defined voice command having a low likelihood of confusion as determined by the safety analysis engine is allowed to be directly invoked without the keyword.

18. The system of claim 15, wherein the confusion mitigation engine is configured to automatically define a restricted set of one or more users who are allowed to utilize the proposed voice command to prevent potential conflicts with at least one other voice command that is not available to any user in the restricted set but is available to at least one other user of the voice-enabled system.

19. The system of claim 14, wherein the at least one processor is further programmed to implement:

an abstraction engine configured to record and execute user-defined abstractions, wherein the at least one action comprises at least one user-defined abstraction recorded by the abstraction engine.

20. The system of claim 14, further comprising:

at least one grammar of system-defined voice commands, wherein the speech recognition engine and the safety analysis engine are adapted to process the system-defined voice commands.