Method of speech recognition using empirically determined word candidates

• US 6,963,834 B2
• Filed: 05/29/2001
• Issued: 11/08/2005
• Est. Priority Date: 05/29/2001
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Lewis, James R., Hartley, Matthew W., Reich, David E.
• Primary Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US09/867,197
• Publication Number:
  US 20020184016A1
• Time in Patent Office: 1,624 Days
• Field of Search: 704/235, 704/239, 704/250, 704/251, 704/240, 704/270, 704/246, 704/257, 704/236, 704/255, 704/278, 455/426.1
• US Class Current: 704/235
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 2015/221 Announcement of recognition results

Abstract

A method for performing speech recognition can include determining a recognition result for received user speech. The recognition result can include recognized text and a corresponding confidence score. The confidence score of the recognition result can correspond to a predetermined minimum threshold. If the confidence score does not exceed the predetermined minimum threshold, the user can be presented with at least one empirically determined alternate word candidate corresponding to the recognition result.

Claims

1. A method for performing speech recognition comprising:

determining a recognition result for received user speech, said recognition result comprising recognized text and a corresponding confidence score;

comparing said confidence score corresponding to said recognition result to a predetermined minimum threshold; and

if said confidence score does not exceed said predetermuied minimum threshold, presenting said user with at least one empirically determined alternale word candidate corresponding to said recognition result.

2. The method of claim 1, further comprising:

receiving a user input specifying one of said at least one empirically determined alternate word candidates as a correct recognition result corresponding to said received user speech.

3. The method of claim 1, wherein said at least one empirically determined alternate word candidate is phonetically similar to said recognition result.

4. The method of claim 1, wherein said at least one empirically determined alternate word candidate is substantially phonetically equivalent to said recognition result.

5. The method of claim 1, wherein said at least one empirically determined alternate word candidate has a corresponding conditional probability.

6. The method of claim 1, wherein said at least one empirically determined alternate word candidate has a corresponding conditional probability greater than a predetermined threshold.

7. The method of claim 1, wherein said at least one empirically determined alternate word candidate is presented in a graphical format.

8. The method of claim 1, wherein said at least one empirically determined alternate word candidate is presented using an audio user interface.

9. The method of claim 1, wherein said speech specifies a character, and at least one of said at least one empirically determined alternate word candidates is a character.

10. A method for performing speech recognition comprising:

determining a recognition result for received user speech specifying a character, said recognition result comprising a character interpretation of said user speech and a corresponding confidence score;

comparing said confidence score corresponding to said recognition result to a predetermined minimum threshold; and

if said confidence score does not exceed said predetermined minimum threshold, presenting said user with at least one empirically determined alternate character candidate corresponding to said recognition result, said alternate character candidate having a corresponding conditional probability.

11. 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 a recognition result for received user speech, said recognition result comprising recognized text and a corresponding confidence score;

comparing said confidence score corresponding to said recognition result to a predetermined minimum threshold; and

if said confidence score does not exceed said predetermined minimum threshold, presenting said user with at least one empirically determined alternate word candidate corresponding to said recognition result.

12. The machine readable storage of claim 11, further comprising:

receiving a user input specifying one of said at least one empirically determined alternate word candidates as a correct recognition result corresponding to said received user speech.

13. The machine readable storage of claim 11, wherein said at least one empirically determined alternate word candidate is phonetically similar to said recognition result.

14. The machine readable storage of claim 11, wherein said at least one empirically determined alternate word candidate is substantially phonetically equivalent to said recognition result.

15. The machine readable storage of claim 11, wherein said at least one empirically determined alternate word candidate has a corresponding conditional probability.

16. The machine readable storage of claim 11, wherein said at least one empirically determined alternate word candidate has a corresponding conditional probability greater than a predetermined threshold.

17. The machine readable storage of claim 11, wherein said at least one empirically determined alternate word candidate is presented in a graphical format.

18. The machine readable storage of claim 11, wherein said at least one empirically determined alternate word candidate is presented using an audio user interface.

19. The machine readable storage of claim 11, wherein said speech specifies a character, and at least one of said at least one empirically determined alternate word candidates is a character.

20. 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 a recognition result for received user speech specifying a character, said recognition result comprising a character interpretation of said uacr speech and a corresponding confidence score;

comparing said confidence score corresponding to said recognition result to a predetermined minimum threshold; and

if said confidence scare does not exceed said predetermined minimum threshold, presenting said user with at least one empirically determined alternate character candidate corresponding to said recognition result.