Voice recognition rejection scheme

• US 6,574,596 B2
• Filed: 02/08/1999
• Issued: 06/03/2003
• Est. Priority Date: 02/08/1999
• Status: Expired due to Term

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Chang, Chienchung, Garudadri, Harinath, Dejaco, Andrew P., Bi, Ning
• Primary Examiner(s): CHAWAN, VIJAY B
• Application Number:
  US09/248,513
• Publication Number:
  US 20020055841A1
• Time in Patent Office: 1,576 Days
• Field of Search: 704/252, 704/253, 704/254, 704/256, 704/255, 704/246, 704/243, 704/244, 704/240, 704/236, 704/238, 704/241, 704/242, 704/249, 706/12, 379/88.02
• US Class Current: 704/249
• CPC Class Codes: G10L 15/10 using distance or distortion measures between unknown speech and reference templates
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

A voice recognition rejection scheme for capturing an utterance includes the steps accepting the utterance, applying an N-best algorithm to the utterance, or rejecting the utterance. The utterance is accepted if a first predefined relationship exists between one or more closest comparison results for the utterance with respect to a stored word and one or more differences between the one or more closest comparison results and one or more other comparison results between the utterance and one or more other stored words. An N-best algorithm is applied to the utterance if a second predefined relationship exists between the one or more closest comparison results and the one or more differences between the one or more closest comparison results and the one or more other comparison results. The utterance is rejected if a third predefined relationship exists between the one or more closest comparison results and the one or more differences between the one or more closest comparison results and the one or more other comparison results. One of the one or more other comparison results may advantageously be a next-closest comparison result for the utterance and another store word. The first, second, and third predefined relationships may advantageously be linear relationships.

Claims

1. A method of capturing an utterance in a voice recognition system, comprising the steps of:

comparing the utterance to a first stored word to generate a first score;

comparing the utterance to a second stored word to generate a second score;

determining a difference between the first score end the second score;

determining a ratio of the first score to the difference; and

processing the utterance based on the ratio.

2. The method of claim 1, further comprising:

if the ratio of the first score to the difference is within a first range of values, accepting the utterance;

if the ratio of the first score to the difference is within a second range of values, applying an N-best algorithm to verify the utterance; and

if the ratio of the first score to the difference is within a third range of values, rejecting the utterance.

3. The method of claim 1, wherein the difference corresponds to a change of score between the first score and the second score.

4. The method of claim 1, wherein the first stored word comprises a best candidate in a vocabulary of a voice recognition system, and the second stored word comprises a next-best candidate in a vocabulary of a voice recognition system.

5. The method of claim 1, wherein the first score comprises a closest comparison result, and the second score comprises a next-closest comparison result.

6. The method of claim 1, wherein the first score and the second score comprise linear predictive coding coefficients.

7. The method of claim 1, wherein the first score and the second score comprise cepstral coefficients.

8. The method of claim 1, wherein the first score and the second score comprise bandpass filter outputs.

9. The method of claim 2, wherein the first, second, and third range of values define linear relationships between the first score and the difference.

10. The method of claim 1, wherein the difference comprises a difference between a closest comparison result and a next-closest comparison result.

11. A voice recognition system, comprising:

an acoustic processor configured to extract speech parameters from digitized speech samples of an utterance; and

a processor coupled to the acoustic processor and configured to compare the utterance to a first stored word to generate a first score;

compare the utterance to a second stored word to generate a second score;

determine a difference between the first score and the second score;

determine a ratio of the first score to the difference; and

process the utterance based on the relationship.

12. The voice recognition system of claim 11, wherein the processor is further configured to:

accept the utterance if the ratio of the first score to the difference is within a first range of values;

apply an N-best algorithm to verify the utterance if the ratio of the first score to the difference is within a second range of values; and

reject the utterance if the ratio of the first score to the difference is within a third range of values, rejecting the utterance.

13. The voice recognition system of claim 11, wherein the difference corresponds to a change of score between the first score and the second score.

14. The voice recognition system of claim 11, wherein the first stored word comprises a best candidate in a vocabulary of the voice recognition system, and the second stored word comprises a next-best candidate in a vocabulary of the voice recognition system.

15. The voice recognition system of claim 11, wherein the first score comprises a closest comparison result, and the second score comprises a next-closest comparison result.

16. The voice recognition system of claim 11, wherein the first and the second score comprise linear predictive coding coefficients.

17. The voice recognition system of claim 11, wherein the first score and the second score comprise cepstral coefficients.

18. The voice recognition system of claim 11, wherein the first score and the second score comprise bandpass filter outputs.

19. The voice recognition system of claim 12, wherein the first, second, and third range of values define linear relationships between the first score and the difference.

20. The voice recognition system of claim 11, wherein the difference comprises a difference between a closest comparison result and a next-closest comparison result.

21. A voice recognition system, comprising:

means for comparing the utterance to a first stored word to generate a first score;

means for comparing the utterance to a second stored word to generate a second score;

means for determining a difference between the first score and the second score;

means for determining a ratio of the first score to the difference; and

means for processing the utterance based on the relationship.

22. The voice recognition system of claim 21, further comprising:

means for accepting the utterance if the ratio of the first score to the difference is within a first range of values;

means for applying an N-best algorithm to verify the utterance if the ratio of the first score to the difference is within a second range of values; and

means for rejecting the utterance if the ratio of the first score to the difference is within a third range of values, rejecting the utterance.

23. The voice recognition system of claim 22, wherein the first, second, and third range of values define linear relationships between the first score and the difference.

24. The voice recognition system of claim 21, wherein the difference corresponds to a change of score between the first score and the second score.

25. The voice recognition system of claim 21, wherein the first stored word comprises a best candidate in a vocabulary of the voice recognition system, and the second stored word comprises a next-best candidate in a vocabulary of the voice recognition system.

26. The voice recognition system of claim 21, wherein the first score comprises a closest comparison result, and the second comprises a next-closest comparison result.

27. The voice recognition system of claim 21, wherein the first score and the second comprise linear predictive coding coefficients.

28. The voice recognition system of claim 21, wherein the first score and the second score comprise cepstral coefficients.

29. The voice recognition system of claim 21, wherein the first score and the second score comprise bandpass filter outputs.

30. The voice recognition system of claim 21, wherein difference comprises a difference between a closest comparison result and a next-closest comparison result.

31. A voice recognition system, comprising:

means for extracting speech parameters from digitized speech samples of an utterance; and

means for;

comparing the utterance to a first stored word to generate a first score;

comparing the utterance to a second stored word to generate a second score;

determining a difference between the first score and the second score;

determining a ratio of the first score to the difference; and

processing the utterance based on the relationship.

32. The voice recognition system of claim 31, further comprising means for:

accepting the utterance if the ratio of the first score to the difference is within a first range of values;

applying an N-best algorithm to verify the utterance if the ratio of the first score to the difference is within a second range of values; and

rejecting the utterance if the ratio of the first score to the difference is within a third range of values, rejecting the utterance.

33. The voice recognition system of claim 31, wherein the difference comprises a difference between a closest comparison result and a next-closest comparison result.

34. The voice recognition system of claim 31, wherein the difference corresponds to a change of score between the first score and the second score.

35. The voice recognition system of claim 31, wherein the first stored word comprises a best candidate in a vocabulary of the voice recognition system, and the second stored word comprises a next-best candidate in a vocabulary of the voice recognition system.

36. The voice recognition system of claim 31, wherein the first score comprises a closest comparison result, and the second score comprises a next-closest comparison result.

37. The voice recognition system of claim 31, wherein the first score and the second score comprise linear predictive coding coefficients.

38. The voice recognition system of claim 31, wherein the first score and the second score comprise cepstral coefficients.

39. The voice recognition system of claim 31, wherein the first score and the second score comprise bandpass filter outputs.

40. The voice recognition system of claim 32, wherein the first, second, and third range of values define linear relationships between the first score and the difference.