Method and apparatus for performing relational speech recognition

• US 20050234723A1
• Filed: 02/23/2005
• Published: 10/20/2005
• Est. Priority Date: 09/28/2001
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Franco, Horacio, Bercow, Douglas A., Gadde, Venkata Ramana Rao, Venkataraman, Anand, Myers, Gregory K., Bratt, Harry, Israel, David J., Frandsen, Michael W., Arnold, James F.
• Granted Patent: US 7,533,020 B2
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G10L 15/065 Adaptation
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models

Abstract

A method and apparatus are provided for performing speech recognition using observable and meaningful relationships between words within a single utterance and using a structured data source as a source of constraints on the recognition process. Results from a first constrained speech recognition pass can be combined with information about the observable and meaningful word relationships to constrain or simplify subsequent recognition passes. This iterative process greatly reduces the search space required for each recognition pass, making the speech recognition process more efficient, faster and accurate.

Claims

1. A method for performing speech recognition, the method comprising:

acquiring a speech signal from a user;

performing a first recognition pass by applying a first language model to said speech signal, said first language model being constrained in accordance with a structured data source; and

generating a subsequent language model based at least in part on results from said first recognition pass.

2. The method of claim 1, further comprising:

performing a subsequent recognition pass by applying said subsequent language model; and

recognizing said speech signal.

3. The method of claim 1, wherein said structured data source is at least one of a geographic location report, a map, a blueprint, a telephone directory, a service directory, the Internet, a mobile address book and one or more user preferences.

4. The method of claim 3, wherein said one or more user preferences are learned.

5. The method of claim 3, wherein said one or more user preferences are programmed by said user.

6. The method of claim 1, wherein said first language model is constrained to a limited geographic domain encompassing said user'"'"'s current geographic location.

7. The method of claim 6, wherein said user'"'"'s current geographic location is acquired prior to acquiring said speech signal.

8. The method of claim 7, wherein said user'"'"'s current geographic location is acquired periodically based on said user'"'"'s travel.

9. The method of claim 1, wherein said first language model is constrained to a geographic area specified by said user.

10. The method of claim 9, wherein said first language model is constrained to a union of geographic areas surrounding a plurality of waypoints identified by said user.

11. A computer readable medium containing an executable program for performing speech recognition, where the program performs the steps of:

acquiring a speech signal from a user;

performing a first recognition pass by applying a first language model to said speech signal, said first language model being constrained in accordance with a structured data source; and

generating a subsequent language model based at least in part on results from said first recognition pass.

12. The computer readable medium of claim 11, further comprising:

performing a subsequent recognition pass by applying said subsequent language model; and

recognizing said speech signal.

13. The computer readable medium of claim 11, wherein said structured data source is at least one of a geographic location report, a map, a blueprint, a telephone directory, a service directory, the Internet, a mobile address book and one or more user preferences.

14. The computer readable medium of claim 13, wherein said one or more user preferences are learned.

15. The computer readable medium of claim 13, wherein said one or more user preferences are programmed by said user.

16. The computer readable medium of claim 11, wherein said first language model is constrained to a limited geographic domain encompassing said user'"'"'s current geographic location.

17. The computer readable medium of claim 16, wherein said user'"'"'s current geographic location is acquired prior to acquiring said speech signal.

18. The computer readable medium of claim 17, wherein said user'"'"'s current geographic location is acquired periodically based on said user'"'"'s travel.

19. The computer readable medium of claim 11, wherein said first language model is constrained to a geographic area specified by said user.

20. The computer readable medium of claim 19, wherein said first language model is constrained to a union of geographic areas surrounding a plurality of waypoints identified by said user.

21. Apparatus for performing speech recognition, said apparatus comprising:

means for acquiring a speech signal from a user;

means for performing a first recognition pass by applying a first language model to said speech signal, said first language model being constrained in accordance with a structured data source; and

means for generating a subsequent language model based at least in part on results from said first recognition pass.