Speech recognition using categories and speech prefixing

• US 7,624,018 B2
• Filed: 03/12/2004
• Issued: 11/24/2009
• Est. Priority Date: 03/12/2004
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Hsu, Bo-June, Schmid, Philipp H., Wood, David Jeremy Guy, Newkerk, Oscar H. III, Chambers, Robert L.
• Primary Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US10/799,356
• Publication Number:
  US 20050203740A1
• Time in Patent Office: 2,083 Days
• Field of Search: 704/275, 704/270, 704/257
• US Class Current: 704/275
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/193 Formal grammars, e.g. finite state automata, context free grammars or word networks
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams
• CPC Class Codes: G10L 2015/088 Word spotting
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Speech recognition utilizing categories and prefixes is disclosed. Categories identify types of recognition and allow different grammars and prefixes for each category. Categories can be directed to specific applications and/or program modules. Uttering a prefix allows users to easily direct text to specific grammars for enhanced recognition, and also to direct the recognized text to the appropriate application/module.

Claims

1. A computer-implemented method of recognizing speech, the method comprising:

employing a computing device to recognize a predefined prefix in user speech, the predefined prefix being associated with a single grammar in a set of grammars, wherein the predefined prefix is also associated with a target software application;

recognizing text from speech following the prefix using the single grammar associated with the recognized predefined prefix;

identifying the target software application that is associated with the single grammar, the target software application being selected from a plurality of target software applications, associated with other grammars, based on the predefined prefix; and

directing recognized text to the target software application associated with the single grammar and the predefined prefix.

2. The method of claim 1, wherein the prefix is a word.

3. The method of claim 1, wherein the prefix is a phrase.

4. The method of claim 1, wherein a category is provided associating the predefined prefix, single grammar, the target software application, and an identification of a parent of the category together.

5. The method of claim 1, executed upon a desktop computer.

6. The method of claim 1, executed upon a mobile computing device.

11. The method of claim 4, wherein the category is defined as an ActiveCategory.

12. The method of claim 11, and further comprising:

receiving a portion of speech;

identifying whether the portion of speech includes a prefix;

if the portion of speech does not include an identified prefix, automatically recognizing the portion of speech input using the grammar associated with the ActiveCategory; and

if the portion of speech does include an identified prefix, recognizing the portion of speech using a grammar, associated with the identified prefix, that is different than the grammar associated with the ActiveCategory.

7. A speech recognition system comprising:

an input for receiving speech;

a processor for recognizing speech using a set of one or more grammars;

wherein the processor is adapted to receive a first speech input and recognize a prefix associated with a desired category in the first speech input, and wherein the processor is adapted to recognize the first speech input using a first grammar associated with the desired category when the prefix is recognized to generate an output;

wherein the output is provided to a target software application associated with the grammar associated with the prefix; and

wherein, for a subsequent speech input,

the processor recognizes the subsequent speech input using the first grammar associated with the desired category if a subsequent prefix is not identified in the subsequent speech input; and

the processor recognizes the subsequent speech input using a second grammar that is different than the first grammar if a subsequent prefix that indicates another category is recognized in the subsequent speech input.

8. The speech recognition system of claim 7, wherein the prefix must be recognized in order to activate the first grammar to recognize the first speech input.

9. The speech recognition system of claim 7, wherein the desired category is identified as an ActiveCategory.

10. The speech recognition system of claim 9, and wherein the processor receives the subsequent speech input that does not include an identified prefix and automatically recognizes the subsequent portion of input speech using the first grammar that is associated with the ActiveCategory.

13. A system comprising:

an input for receiving speech;

a plurality of target software applications;

a processor for recognizing received speech using a set of grammars;

a plurality of grammar categories, wherein each of the plurality of grammar categories is represented by a data structure that indicates a predefined prefix associated with the grammar category, a single grammar of the set of grammars associated with the grammar category, an identification of a parent of the grammar category, and a particular target software application of the plurality of target software applications associated with the grammar category;

wherein the processor receives a portion of input speech and accesses the data structures that represent the plurality of grammar categories to identify a prefix in the portion of input speech and to generate an output by recognizing the portion of input speech, the portion of input speech being recognized using a single grammar selected from the set of grammars based on the identified prefix, wherein the output is provided to the particular target software application that is identified in the plurality of grammar categories as being associated with the identified prefix.

14. The speech recognition system of claim 13, wherein, for each of the plurality of grammar categories, the data structure representing the grammar category includes a flag that indicates whether the prefix associated with the grammar category is required to invoke the grammar category.

15. The speech recognition system of claim 13, wherein the at least one grammar category associated with the identified prefix is defined as an ActiveCategory, and wherein the processor receives a subsequent portion of input speech that does not include an identified prefix and automatically recognizes the subsequent portion of input speech using the at least one grammar associated with the ActiveCategory.

16. The speech recognition system of claim 15, wherein the plurality of grammar categories are arranged in a hierarchy including at least one parent category having at least one descendent category.

17. The speech recognition system of claim 16, wherein the processor is configured to access all grammars associated with the ActiveCategory and any descendent categories that descend from the ActiveCategory for a subsequent speech input that does not include an identified prefix.

18. The speech recognition system of claim 16, wherein for subsequent portions of input speech a prefix is only required to invoke any grammar categories that are not included in the ActiveCategory and are not descendent categories that descend from the ActiveCategory.