Continuous speech recognition of text and commands

• US 5,799,279 A
• Filed: 11/13/1995
• Issued: 08/25/1998
• Est. Priority Date: 11/13/1995
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Dragon Systems, Inc. (Microsoft Corporation)
• Inventors: Gould, Joel M., Young, Jonathan H.
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): EDOUARD, PATRICK NESTOR
• Application Number:
  US08/559,207
• Time in Patent Office: 1,016 Days
• Field of Search: 395/2.84, 395/2.86, 395/2.87, 395/2.79, 395/2.6, 395/2.55, 395/2.4
• US Class Current: 704/275
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 2015/088 Word spotting
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2015/228 of application context

Abstract

In a method for use in recognizing continuous speech, signals are accepted corresponding to interspersed speech elements including text elements corresponding to text to be recognized and command elements corresponding to commands to be executed. The elements are recognized. The recognized elements are acted on in a manner which depends on whether they represent text or commands.

Claims

1. A method for use in recognizing continuous speech comprising

accepting signals corresponding to interspersed speech elements including text elements corresponding to text to be recognized and command elements corresponding to commands to be executed, wherein a particular one of the speech elements may correspond to a text element in one context and to a command element in another context,

recognizing the speech elements,

when a recognized one of the speech elements may be either a command element or a text element, designating the recognized speech element as corresponding to a text element or to a command element based on a context in which the recognized speech element appears, and

acting on the recognized speech elements in a manner which depends on whether the speech elements correspond to text elements or command elements.

2. The method of claim 1 in which a text element is acted on by providing the text element to a text processing application.

3. The method of claim 1 in which a command element acted upon by causing an application to perform a step.

4. The method of claim 1 in which the designating is based on natural characteristics of spoken text versus spoken commands.

5. The method of claim 1 in which the designating includes evaluating a likelihood that the recognized speech element is either a command element or a text element.

6. The method of claim 1 further including biasing the designating in favor of the recognized speech element being a text element or a command element.

7. The method of claim 6 in which the biasing includes determining if the recognized speech element reflects a command reject.

8. The method of claim 6 in which the biasing includes determining if the recognized speech element conforms to a command template.

9. The method of claim 6 in which the biasing includes comparing recognition scores of the recognized speech element as a command or as text.

10. The method of claim 6 in which the biasing includes determining the length of silence between successive ones of the speech elements.

12. The method of claim 1 in which the recognizing comprises, in parallel

recognizing the speech elements as if they were text, and

recognizing the speech elements as if they were commands.

13. The method of claim 12 further comprising temporarily stopping the recognizing of speech elements as if they were text (or commands) upon designating a speech element as being a command element (or a text element).

14. The method of claim 1 further comprising displaying to a user the results of the recognition.

15. The method of claim 14 wherein the results are partial results.

16. The method of claim 1 further comprising enabling the user to cause a re-recognition if a speech element is incorrectly recognized as a text element or a command element.

17. The method of claim 16 in which

the user may cause a rerecognition if a command element is recognized as a text element, and

in response to the re-recognition a text processing application may undo the inclusion of the text element in text being processed.

18. The method of claim 1 in which prior to acting on a recognized command element, information associated with the command element is displayed to a user.

19. The method of claim 1 further comprising

accepting from a user a direction to consider previous or subsequent speech elements as either text elements or command elements but not both.

11. The method of 6 in which the biasing includes determining whether actions of the user imply that the recognized speech element cannot be text.

20. Software stored on a medium for use in recognizing speech comprising

instructions for accepting signals corresponding to interspersed speech elements including text elements corresponding to text to be recognized and command elements corresponding to commands to be executed, wherein a particular one of the speech elements may correspond to a text element in one context and to a command element in another context,

instructions for recognizing the speech elements,

instructions for, when a recognized one of the speech element may be either a command element or a text element, designating the recognized speech element as corresponding to a text element or to a command element based on a context in which the speech element appears, and

instructions for acting on the recognized speech elements in a manner which depends on whether the speech elements correspond to text elements or command elements.

21. A method for use in recognizing speech comprising

accepting signals corresponding to interspersed speech elements including text elements corresponding to text to be recognized and command elements corresponding to commands to be executed, wherein a particular one of the speech elements may correspond to a text element in one context and to a command element in another context,

recognizing the speech elements,

when a recognized one of the speech elements may be either a command element or a text element, biasing the recognizing in favor of the given element being text or a command,

acting on the text by providing it to a text processing application,

acting on the command by causing an application to perform a step, and

displaying to a user the results of the recognition.

22. A method for use in recognizing continuous speech comprising

accepting signals corresponding to interspersed speech elements including text elements corresponding to text to be recognized and command elements corresponding to commands to be executed, wherein a particular one of the speech elements may correspond to a text element in one context and to a command element in another context;

recognizing the speech elements;

when a recognized one of the speech elements may be either a command element or a text element, designating the recognized speech element as corresponding to a text element or to a command element based on a context in which the recognized speech element appears, the designating including evaluating a likelihood that the recognized speech element is either a command element or a text element;

acting on a recognized speech element that corresponds to a text element by providing the text element to a text processing application;

acting on a recognized speech element that corresponds to a command element by causing an application to perform a step;

displaying to a user results of the recognition; and

enabling the user to cause a re-recognition if a speech element is incorrectly recognized as a text element or a command element.