Method and system for automatically adjusting prompt feedback based on predicted recognition accuracy

• US 6,505,155 B1
• Filed: 05/06/1999
• Issued: 01/07/2003
• Est. Priority Date: 05/06/1999
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Vanbuskirk, Ronald, Ortega, Kerry A., Wang, Huifang, Wolf, Catherine G.
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): Opsasnick, Michael N.
• Application Number:
  US09/306,355
• Time in Patent Office: 1,342 Days
• Field of Search: 704/231, 704/246, 704/251, 704/255
• US Class Current: 704/246
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/225 Feedback of the input speech

Abstract

In a computer speech user interface, a method and computer apparatus for automatically adjusting the content of feedback in a responsive prompt based upon predicted recognition accuracy by a speech recognizer. The method includes the steps of receiving a user voice command from the speech recognizer; calculating present speech recognition accuracy based upon the received user voice command; predicting future recognition accuracy based upon the calculated present speech recognition accuracy; and, generating feedback in a responsive prompt responsive to the predicted recognition accuracy. For predicting future poor recognition accuracy based upon poor present recognition accuracy, the calculating step can include monitoring the received user voice command; detecting a reduced accuracy condition in the monitored user voice command; and, determining poor present recognition accuracy if the reduced accuracy condition is detected in the detecting step.

Claims

1. In a computer speech user interface, a method for automatically adjusting the content of feedback in a responsive prompt based upon predicted recognition accuracy by a speech recognizer, comprising the steps of:

receiving a user voice command from said speech recognizer;

calculating a present speech recognition accuracy based upon said received user voice command;

predicting a future recognition accuracy based upon said calculated present speech recognition accuracy; and

,

controlling semantic content of said responsive prompt to be provided to said user in accordance with said predicted recognition accuracy.

2. The method according to claim 1, wherein said calculating step comprises:

monitoring said received user voice command;

detecting a reduced accuracy condition in said monitored user voice command; and

,

determining poor present recognition accuracy if said reduced accuracy condition is detected in said detecting step.

3. The method according to claim 2, wherein said detecting step comprises:

obtaining a confidence value associated with said received user voice command from said speech recognizer;

identifying as a reduced confidence value each said obtained confidence value having a value below a preset threshold; and

,

triggering a reduced accuracy condition if said obtained reduced confidence value is identified in said identifying step.

4. The method according to claim 2, wherein said detecting step comprises:

storing said user voice command in a command history;

calculating a sum of user undo voice commands stored in said command history; and

,

triggering a reduced accuracy condition if said sum exceeds a preset threshold value.

5. The method according to claim 2, wherein said detecting step comprises:

storing said user voice command in a command history;

calculating a sum of user cancel voice commands stored in said command history; and

,

triggering a reduced accuracy condition if said sum exceeds a preset threshold value.

6. The method according to claim 2, wherein said detecting step comprises:

comparing said user voice command to a corresponding computer responsive prompt;

determining whether said voice command is inconsistent with said corresponding computer responsive prompt;

identifying said user voice command determined to be inconsistent with said corresponding computer responsive prompt as an unexpected user voice command;

storing said unexpected user voice command in a command history;

calculating a sum of unexpected user voice commands stored in said command history; and

,

triggering a reduced accuracy condition if said sum exceeds a preset threshold value.

7. The method according to claim 2, wherein the detecting step comprises:

obtaining measured acoustical signal quality data from said speech recognizer; and

,

triggering a reduced accuracy condition if said measured acoustical signal quality evaluates below a preset threshold value.

8. The method according to claim 2, wherein said controlling step comprises providing said received user voice command as a computer responsive prompt.

9. The method according to claim 2, wherein said controlling step comprises incorporating said received user voice command as part of a subsequent computer responsive prompt.

10. The method according to claim 2, wherein said controlling step comprises providing a request for confirmation of said received voice command as a computer responsive prompt.

11. The method according to claim 2, wherein said controlling step, comprises:

concatenating said received user voice command with a subsequent computer responsive prompt; and

,

providing said concatenation as a single computer responsive prompt.

12. The method according to claim 2, wherein said controlling step comprises providing a list of high-recognition user voice commands as a computer responsive prompt.

13. A computer apparatus programmed with a routine set of instructions stored in a fixed medium, said computer apparatus comprising:

means for receiving a user voice command from a speech recognizer in a speech interface;

means for calculating a present speech recognition accuracy based upon said received user voice command;

means for predicting a future recognition accuracy based upon said calculated present speech recognition accuracy; and

,

means for controlling semantic content of a responsive prompt to be provided to said user in accordance with said predicted recognition accuracy.

14. The computer apparatus according to claim 13, wherein said calculating means comprises:

means for monitoring said received user voice command;

means for detecting a reduced accuracy condition in said monitored user voice command; and

,

means for determining poor present recognition accuracy if said reduced accuracy condition is detected by said detecting means.

15. The computer apparatus according to claim 14, wherein said detecting means comprises:

means for obtaining a confidence value associated with said received user voice command from said speech recognizer;

means for identifying as a reduced confidence value each said obtained confidence value having a value below a preset threshold; and

,

means for triggering a reduced accuracy condition if said obtained reduced confidence value is identified by said identifying means.

16. The computer apparatus according to claim 14, wherein said detecting means comprises:

means for storing said user voice command in a command history;

means for calculating a sum of user undo voice commands stored in said command history; and

,

means for triggering a reduced accuracy condition if said calculated sum exceeds a preset threshold value.

17. The computer apparatus according to claim 14, wherein said detecting means comprises:

means for storing said user voice command in a command history;

means for calculating a sum of user cancel voice commands stored in said command history; and

,

means for triggering a reduced accuracy condition if said calculated sum exceeds a preset threshold value.

18. The computer apparatus according to claim 14, wherein said detecting means comprises:

means for comparing said user voice command to a corresponding computer responsive prompt;

means for determining whether said voice command is inconsistent with said corresponding computer responsive prompt;

means for identifying said user voice command determined by said determining means to be inconsistent with said corresponding computer responsive prompt as an unexpected user voice command;

means for storing said identified unexpected user voice command in a command history;

means for calculating a sum of unexpected user voice commands stored in said command history; and

,

means for triggering a reduced accuracy condition if said calculated sum exceeds a preset threshold value.

19. The computer apparatus according to claim 14, wherein said detecting step means comprises:

means for obtaining measured acoustical signal quality data from said speech recognizer; and

,

means for triggering a reduced accuracy condition if said obtained measured acoustical signal quality evaluates below a preset threshold value.

20. The computer apparatus according to claim 14, wherein said controlling means comprises means for providing said received user voice command as a computer responsive prompt.

21. The computer apparatus according to claim 14, wherein said controlling means comprises means for incorporating said received user voice command as part of a subsequent computer responsive prompt.

22. The computer apparatus according to claim 14, wherein said controlling means comprises means for providing a request for confirmation of said received voice command as a computer responsive prompt.

23. The computer apparatus according to claim 14, wherein said controlling means comprises:

means for concatenating said received user voice command with a subsequent computer responsive prompt; and

,

means for providing said concatenation as a single computer responsive prompt.

24. The method according to claim 14, wherein said controlling means comprises providing a list of high-recognition user voice commands as a computer responsive prompt.