Dynamic N-best algorithm to reduce recognition errors

• US 7,421,387 B2
• Filed: 05/18/2004
• Issued: 09/02/2008
• Est. Priority Date: 02/24/2004
• Status: Expired due to Fees

• Current Assignee: GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: General Motors Corporation (Motors Liquidation Co. GUC Trust)
• Inventors: Godden, Kurt S.
• Primary Examiner(s): Vo; Huyen X.
• Application Number:
  US10/847,719
• Publication Number:
  US 20050187768A1
• Time in Patent Office: 1,568 Days
• Field of Search: 704/254, 704/252, 704/231, 704/236, 704/255, 704/256, 704/251, 704/240, 704/200, 704/235, 704/270, 704/249
• US Class Current: 704/200
• CPC Class Codes: G10L 15/08 Speech classification or search

Abstract

A method for reducing recognition errors. The method includes receiving an N-best list associated with an input of a computer based recognition system. The N-best list includes one or more hypotheses and associated confidence values. The input is classified in response to the N-best list, resulting in a classification. A re-scoring algorithm that is tuned for the classification is selected. The re-scoring algorithm is applied to the N-best list to create a re-scored N-best list. A hypothesis for the value of the input is selected based on the re-scored N-best list.

Claims

1. A method for reducing recognition errors, the method comprising:

receiving an N-best list associated with an input of a computer based recognition system, the N-best list including one or more hypotheses and associated confidence values;

classifying the input in response to the N-best list to result in a classification into a bin from a plurality of bins according to characteristics of the N-best list;

selecting a re-scoring algorithm from a plurality of rescoring algorithms that are tuned for the classification, wherein the re-scoring algorithm is selected on a per bin basis;

applying the re-scoring algorithm to the N-best list to create a re-scored N-best list; and

selecting a hypothesis for the value of the input based on the re-scored N-best list.

2. The method of claim 1 wherein:

the computer based recognition system is a speech recognition engine;

the one or more hypotheses and associated confidence values are determined by the speech recognition engine; and

the input is a user utterance.

3. The method of claim 2 wherein the user utterance includes a name of a letter in the alphabet.

4. The method of claim 2 wherein the user utterance includes a name of a number.

5. The method of claim 2 wherein the user utterance includes a word.

6. The method of claim 2 wherein the user utterance includes a phrase.

7. The method of claim 2 wherein the user utterance includes a sentence.

8. The method of claim 1 wherein the computer based recognition system is an optical character reader engine.

9. The method of claim 1 wherein the computer based recognition system is an image recognition engine.

10. The method of claim 1 wherein the re-scoring algorithm was created in response to training data.

11. The method of claim 1 wherein the classifying is based on one or more of the confidence values associated with the one or more hypotheses on the N-best list, an expected frequency of the one or more hypotheses on the N-best list, a conditional probability that the one or more hypotheses are included in the N-best list, confidence value distributions associated with each of the one or more hypotheses, the number of hypotheses on the N-best list, and the order of the hypotheses on the N-best list, where the one or more hypotheses on the N-best list are ordered from highest associated confidence value to lowest associated confidence value.

12. The method of claim 1 wherein the selecting a hypothesis includes selecting the hypothesis with the highest confidence value from the one or more hypotheses on the re-scored N-best list.

13. The method of claim 1 wherein the re-scoring algorithm is based on statistical properties associated with the N-best list.

14. The method of claim 1 wherein the re-scoring algorithm includes re-scoring the N-best list based on the confidence values associated with the one or more hypotheses on the N-best list.

15. The method of claim 1 wherein the re-scoring algorithm includes re-scoring the N-best list based on an expected frequency of the one or more hypotheses on the N-best list.

16. The method of claim 1 wherein the re-scoring algorithm includes re-scoring the N-best list based on a conditional probability that the one or more hypotheses are included in the N-best list.

17. The method of claim 1 wherein the re-scoring algorithm includes re-scoring the N-best list based on confidence value distributions associated with each of the one or more hypotheses.

18. The method of claim 1 wherein the re-scoring algorithm includes re-scoring the N-best list based on the number of hypotheses on the N-best list.

19. The method of claim 1 wherein the re-scoring algorithm includes re-scoring the N-best list based on the order of the hypotheses on the N-best list, where the one or more hypotheses on the N-best list are ordered from highest associated confidence value to lowest associated confidence value.

20. The method of claim 1 wherein the re-scoring algorithm includes one or more of re-scoring the N-best list based on the confidence values associated with the one or more hypotheses on the N-best list, re-scoring the N-best list based on an expected frequency of the one or more hypotheses on the N-best list, re-scoring the N-best list based on a conditional probability that the one or more hypotheses are included in the N-best list, re-scoring the N-best list based on confidence value distributions associated with each of the one or more hypotheses, re-scoring the N-best list based on the number of hypotheses on the N-best list, and re-scoring the N-best list based on the order of the hypotheses on the N-best list, where the one or more hypotheses on the N-best list are ordered from highest associated confidence value to lowest associated confidence value.

21. A computer implemented method for providing a dynamic N-best algorithm to reduce recognition errors, the method comprising:

receiving an N-best list associated with an input of a computer based recognition system, the N-best list including one or more hypotheses and associated confidence values;

classifying the input in response to the N-best list to result in a classification into a bin from a plurality of bins according to characteristics of the N-best list;

selecting a re-scoring algorithm from a plurality of rescoring algorithms that are tuned for the classification, wherein the re-scoring algorithm is selected on a per bin basis;

applying the re-scoring algorithm to the N-best list to create a re-scored N-best list; and

selecting a hypothesis for the value of the input based on the re-scored N-best list.

22. A system for reducing recognition errors, the system comprising a host system in communication with a computer based recognition system, the host system including instructions to implement a method comprising:

receiving an N-best list associated with an input to the computer based recognition system, the N-best list including one or more hypotheses and associated confidence values;

classifying the input in response to the N-best list to result in a classification into a bin from a plurality of bins according to characteristics of the N-best list;

selecting a re-scoring algorithm from a plurality of rescoring algorithms that are tuned for the classification, wherein the re-scoring algorithm is selected on a per bin basis;

applying the re-scoring algorithm to the N-best list to create a re-scored N-best list; and

selecting a hypothesis for the value of the input based on the re-scored N-best list.