Word training interface

• US 7,587,317 B2
• Filed: 02/15/2002
• Issued: 09/08/2009
• Est. Priority Date: 02/15/2002
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Pong Yip, Clement Chun, Falcon, Stephen Russell, Miller, David Michael
• Primary Examiner(s): Han; Qi
• Application Number:
  US10/076,825
• Publication Number:
  US 20030158731A1
• Time in Patent Office: 2,762 Days
• Field of Search: 704/231, 704/270, 704/275, 704/257, 704/270.1, 704/256, 704/256.2, 704/3
• US Class Current: 704/231
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

A method for exposing speech engine features to one or more independent applications wherein the features relate to word training and/or wherein the method optionally exposes the speech engine features without invoking a user interface. A word training interface to expose speech engine features to one or more independent applications wherein the interface is optionally an application programming interface.

Claims

1. A method implemented by a computing device configured to expose speech engine features via a word training interface, the method comprising:

exposing, via the interface, speech engine features for one or more speech engines to a plurality of independent applications wherein the speech engine features comprise one or more features for word training to train a word;

receiving, via the interface, a call issued by one of the plurality of independent applications to train a word using one of the one or more speech engines; and

,

after training the word, receiving, via the interface, a call issued by the one of the plurality of independent applications for adding or replacing information associated with the word in a lexicon associated with at least the one independent applications, the lexicon stored in memory of the computing device and for use during speech recognition.

2. The method of claim 1, wherein the exposing exposes the speech engine features without invoking a user interface to thereby allow each of the plurality of independent applications to control a user'"'"'s experience of that independent application.

3. The method of claim 1 comprising receiving a call issued by one of the plurality of independent applications, the call having a parameter;

parsing the call to retrieve the parameter; and

communicating the parameter to one of the one or more speech engines.

4. The method of claim 1 comprising receiving a call from a speech server, the call issued by one of the plurality of independent applications and having a parameter;

parsing the call to retrieve the parameter; and

communicating the parameter to one or the one or more speech engines.

5. The method of claim 1 comprising receiving a call for a callable method wherein the callable method comprises at least one method selected from the group consisting of a method to begin word training, a method to start audio recording, a method to get audio data, a method to acknowledge a recording, a method to commit a word training, a method to cancel a word training, a method to remove a word, a method to get a number of utterances, and a method to query whether a word is trained.

6. The method of claim 1, wherein the word training interface is an application programming interface.

7. One or more computer-readable storage media having computer-readable instructions thereon which, when executed by a programmable computing device, perform the method of claim 1.

8. An application programming interface for exposing speech engine features for word training to a plurality of independent applications, the interface implemented by a computing device and comprising:

one or more methods of the interface, stored in the computer device, callable by one of the plurality of independent applications, the one or more callable methods having parameters, and the parameters passable to one or more speech engines wherein the speech engine features comprise one or more features for word training to train a word and, after training a word, adding or replacing information associated with the word in a lexicon associated with at least one of the plurality of independent applications, the lexicon stored in memory of a computing device and for use during speech recognition.

9. A word training interface for exposing speech engine features, implemented by a computing device, the interface comprising:

one or more callable methods of the interface, stored in the computer device, to train a word using a speech engine and then adding or replacing information associated with the word in a lexicon associated with an application, stored in memory of a computing device and for use during speech recognition wherein the one or more callable methods are callable through instructions issued by at least two applications and wherein the one or more callable methods, when called by an application to train a word, cause at least one communication with at least one of a plurality of speech engines without invoking a user interface.

10. The word training interface of claim 9 comprising an application programming interface.

11. The word training interface of claim 9 wherein the one or more callable methods comprise at least one method selected from the group consisting of a method to begin word training, a method to start audio recording, a method to get audio data, a method to acknowledge a recording, a method to commit a word training that allows for adding or replacing an item in a lexicon associated with an application, a method to cancel a word training, a method to remove a word, a method to get a number of utterances, and a method to query whether a word is trained.

12. A speech recognition system implemented by a computing device, the system comprising:

user interaction means for allowing a user to initiate interactions from any of a plurality of applications;

word training interface means for providing methods, the methods callable through the plurality of applications and providing for word training to train a word using a speech engine and, after word training, providing for adding or replacing information associated with the word in a lexicon associated with any of the plurality of applications, the lexicon stored in memory of a computing device and for use during speech recognition; and

speech engine means for communicating with the word training interface means wherein the speech engine means comprises one or more speech engines.

13. The system of claim 12, further comprising speech server means for prioritizing the interactions from any of the plurality of applications to the word training interface means.

14. The speech recognition system of claim 13, wherein the methods callable through the plurality of applications execute without invoking a user interface to thereby allow each of the plurality of applications to control a user'"'"'s experience of that application.

15. The system of claim 12, wherein the word training interface means comprises an application programming interface.

16. A method of exposing speech engine features related to word training for one or more speech engines to plurality of independent applications, the method implemented by a computing device and comprising:

based on an interaction with one of the plurality of independent applications, issuing an instruction to a word training interface having word training interface features that correspond with the speech engine features related to word training wherein receipt of the instruction results in a call to one of the speech engine features to train a word; and

based on an interaction with the one of the plurality of independent applications, issuing an instruction to the word training interface wherein receipt of the instruction results in a call to one of the speech engine features for adding or replacing information associated with the word in a lexicon associated with at least the one application, the lexicon stored in memory of a computing device and for use during speech recognition.

17. A computing system configured to execute a word training procedure, comprising:

one or more speech engines;

a plurality of applications, each of the plurality of applications configured to receive a user instruction through an associated user interface, and to issue an instruction related to training of a word; and

a word training interface configured to receive an instruction related to training of the word, issue an instruction to one of the one or more speech engines for training of the word, to thereby produce a new training, and to issue an instruction to commit the new training and wherein the instruction to commit the new training allows for adding or replacing information associated with the word in a lexicon associated with at least one of the plurality of applications, the lexicon stored in memory of a computing device and for use during speech recognition.

18. The computing system of claim 17, wherein the word training interface is an application programming interface.

19. The computing system of claim 17, wherein the word training interface comprises callable methods.

20. The computing system of claim 17, wherein the word training interface is configured as an engine extension.