Automatic generation of efficient grammar for heading selection

• US 7,054,813 B2
• Filed: 03/01/2002
• Issued: 05/30/2006
• Est. Priority Date: 03/01/2002
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Lewis, James R., Polkosky, Melanie D., Sadowski, Wallace J. Jr.
• Primary Examiner(s): Chawan, Vijay
• Assistant Examiner(s): Shortledge, Thomas E.
• Application Number:
  US10/086,359
• Publication Number:
  US 20030167168A1
• Time in Patent Office: 1,551 Days
• Field of Search: 704/251, 704/252, 704/257, 704/7, 704/10, 704/275, 704/270, 704/254, 704/231, 704/243
• US Class Current: 704/251
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules

Abstract

A method of generating a grammar for recognizing headings in a speech recognition system can include identifying, within a data store, at least one heading selection associated with a content item. At least a first word from the identified heading selections can be extracted and a heading grammar automatically can be generated by including each extracted word of the identified heading selections within the heading grammar.

Claims

1. A computer-implemented method of generating a grammar for recognizing headings in a speech recognition system comprising:

determining that said at least one heading section is to be presented to a user;

based on the determination, automatically identifying, within a data store, at least one heading selection associated with a content item, wherein each of said at least one heading selection is able to be used as a selection item for identifying the content item through a speech interface;

automatically extracting at least a first word from each said identified heading selection, wherein said extracted at least a first word includes “

n”

words of the heading section, and wherein “

n”

is less than the total number of words in the heading selection;

automatically generating a heading grammar by including each said extracted word of said identified heading selections within said heading grammar;

presenting said identified headings to the user; and

speech recognizing a spoken user selection using said heading grammar.

2. The method of claim 1, wherein the beading selections are specified within the content item by an identifying tag of a markup language.

3. The method of claim 1, wherein said automatic generating step dynamically generates said heading grammar responsive to a user request for at least one content item.

4. The method of claim 1, wherein said automatic generating step dynamically generates said heading grammar responsive to a presentation of individual ones of said identified heading selections.

5. The method of claim 1, wherein “

n”

is a number between one and two.

6. The method of claim 1, further comprising:

presenting said identified heading selections through a speech interface; and

decoding user speech selecting one of said heading selections according to said heading grammar.

7. The method of claim 6, wherein said user speech comprises a first word of one of said heading selections.

8. The method of claim 1, wherein said heading grammars are automatically generated at designated times and occur during at least one of a system update action and a system synchronization action.

9. A computer-based speech processing system for recognizing, at least in part, heading selections, said speech processing system comprising:

a speech interface;

a data store in communication with said speech interface; and

a speech recognition engine in communication with said data story and speech interface,

wherein said speech recognition system is configured to automatically generate a heading grammar comprising at least a first word from each of said heading selections, wherein said at least a first word includes “

n”

words of the heading section, and wherein “

n”

is less tan the total number of words in the heading selection, wherein each of said heading selections references a particular content item, and wherein spoken user selections are speech recognized using said automatically generated heading grammar.

10. A machine-readable storage, having stored thereon a computer program having a plurality of code sections executable by a machine for causing the machine to perform the steps of:

determining that said at least one heading section is to be presented to a user;

automatically identifying, within a data store, at least one heading selection associated with a content item, wherein each of said at least one heading selection is able to be used as a selection item for identifying the content item through a speech interface;

automatically extracting at least a first word from each said identified heading selection, wherein said extracted at least a first word includes “

n”

words of the heading section, and wherein “

n”

is less than the total number of words in the heading selection;

automatically generating a heading grammar by including each said extracted word of said identified heading selections within said heading grammar;

presenting said identified headings to the user; and

speech recognizing a spoken user selection using said heading grammar.

11. The machine-readable storage of claim 10, wherein the heading selections are specified within the content item by an identifying tag of a markup language.

12. The machine-readable storage of claim 10, wherein said automatic generating step dynamically generates said heading grammar responsive to a user request for at least one content item.

13. The machine-readable storage of claim 10, wherein said automatic generating step dynamically generates said heading grammar responsive to a presentation of individual ones of said identified heading selections.

14. The machine-readable storage of claim 10, wherein “

n”

is a number between one and two.

15. The machine-readable storage of claim 10, further comprising:

presenting said identified heading selections through a speech interface; and

decoding user speech selecting one of said heading selections according to said heading grammar.

16. The machine-readable storage of claim 15, wherein said user speech comprises a first word of one of said heading sections.

17. The machine-readable storage of claim 10, wherein said heading grammars are automatically generated at designated times and occur during at least one of a system update action and a system synchronization action.