Determining context for speech recognition

• US 20040128140A1
• Filed: 12/27/2002
• Published: 07/01/2004
• Est. Priority Date: 12/27/2002
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Deisher, Michael E.
• Granted Patent: US 7,392,183 B2
• Time in Patent Office: Days
• US Class Current: 704/275
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 2015/228 of application context

Abstract

A processor-based system obtaining information about an event from schedule data, and using the information to assist speech recognition of speech occurring during at least a portion of the event.

Claims

1. A method comprising:

a processor-based system obtaining information about an event from schedule data; and

the processor-based system using the information to assist speech recognition of speech occurring during at least a portion of the event.

2. The method of claim 1 wherein using the information to assist speech recognition comprises:

using the information to perform one or more of

automatically transcribing speech;

interpreting speech as a command for a command-driven system; and

identifying a speaker.

3. The method of claim 2, wherein:

obtaining information about an event from schedule data comprises

determining a clock time;

selecting the information from a schedule data base depending on the clock time;

determining a speech context for the event from the information; and

using the information to assist speech recognition further comprises

adjusting an acoustic model of a speech recognizer using the speech context for the event to assist speech recognition.

4. The method of claim 3 wherein adjusting an acoustic model of a speech recognizer using the speech context for the event comprises adjusting the acoustic model of the speech recognizer based on parameters describing the environment in which the event is scheduled to occur.

5. The method of claim 4 wherein adjusting an acoustic model of a speech recognizer using the speech context for the event further comprises adjusting the acoustic model of the speech recognizer to account for reverberation based on the reverberation time of the space in which the event is scheduled to occur.

6. The method of claim 4 wherein adjusting an acoustic model of a speech recognizer using the speech context for the event further comprises adjusting the acoustic model of the speech recognizer based on information about noise sources in the environment in which the event is scheduled to occur.

7. The method of claim 1 wherein using the information to assist speech recognition comprises receiving the information from a remote server over a wireless network.

8. The method of claim 1 wherein:

obtaining information about an event from schedule data comprises obtaining information about a person scheduled to attend the event from the schedule data; and

using the information to assist speech recognition comprises using information about the person to assist speech recognition.

9. The method of claim 8 wherein using information about the person to assist speech recognition comprises:

selecting the person as a transcription source based on one or more of

the person'"'"'s name;

the person'"'"'s position in an organization; and

the person'"'"'s membership on a list of transcription sources.

10. The method of claim 8 wherein using information about the persons scheduled to attend the event to assist speech recognition comprises:

identifying a person present at the event based on comparing the speech of the person to a stored acoustic speech identification model of a person scheduled to attend the event; and

using the acoustic speech recognition model of the identified person to assist with speech recognition.

11. The method of claim 8 wherein using information about the persons scheduled to attend the event to assist speech recognition comprises:

initializing a pre-processor to filter sound input based on one or more of;

the number of speakers scheduled to attend the event; and

an estimated pitch range of a speaker scheduled to attend the event based on the gender of the speaker.

12. The method of claim 1 further comprising:

the system interacting with a user to determine one or more of;

the number of persons present at the event;

the identity of a person present at the event; and

the location of the event.

17. A medium having stored thereon data which when accessed by a machine, cause the machine to perform the method of claim 1.

18. The medium of claim 17 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 2.

19. The medium of claim 18 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 3.

20. The medium of claim 19 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 4.

21. The medium of claim 20 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 5.

22. The medium of claim 20 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 6.

23. The medium of claim 17 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 7.

24. The medium of claim 17 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 8.

25. The medium of claim 24 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 9.

26. The medium of claim 24 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 10

27. The medium of claim 24 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 11.

28. The medium of claim 17 having stored thereon further data which when accessed by a machine, cause the machine to perform the method of claim 12.

13. A system comprising:

a processor;

a data store coupled to the processor, having stored therein data which when accessed by the processor cause it to obtain information about an event from schedule data, and

the data store, having stored therein further data which when accessed by the processor causes it to use the information to assist speech recognition.

14. The system of claim 13 further comprising:

a clock coupled to the processor, to allow determination of the time;

a schedule data base accessible from the processor, from which information about the event may be selected by the processor, depending on the time; and

a speech recognizer module coupled to the processor, including a acoustic model adjustable by the processor, using a speech context for the event included in the information about the event.

15. The system of claim 14 further comprising a pre-processing module to filter sound input to the speech recognizer module based on one or more of:

the number of speakers scheduled to attend the event; and

an estimated pitch range of a speaker scheduled to attend the event based on the gender of the speaker.

16. The system of claim 15 wherein the schedule data is accessible from the processor via a wireless network to which the system and the data base are connected.