Method and system for considering information about an expected response when performing speech recognition

• US 10,068,566 B2
• Filed: 06/10/2014
• Issued: 09/04/2018
• Est. Priority Date: 02/04/2005
• Status: Active Grant

• Current Assignee: Vocollect, Inc. (Honeywell International Inc.)
• Original Assignee: Vocollect, Inc. (Honeywell International Inc.)
• Inventors: Braho, Keith, Pike, Jeffrey
• Primary Examiner(s): McFadden, Susan
• Application Number:
  US14/300,276
• Publication Number:
  US 20140288933A1
• Time in Patent Office: 1,547 Days
• Field of Search: 704275
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning

Abstract

A speech recognition system receives and analyzes speech input from a user in order to recognize and accept a response from the user. Under certain conditions, information about the response expected from the user may be available. In these situations, the available information about the expected response is used to modify the behavior of the speech recognition system by taking this information into account. The modified behavior of the speech recognition system comprises adjusting the rejection threshold when speech input matches the predetermined expected response.

Claims

1. A method for recognizing speech comprising the steps of:

conducting, with a speech recognition system, a speech dialog with a user, comprising receiving, with a speech input device, input speech from the user, and digitizing, with a signal processor, the input speech from the user;

in the speech dialog, determining, with a processor of the speech recognition system, an expected response to be received as input speech from the user;

processing, with the processor, the input speech from the user to generate a hypothesis and confidence factor for the input speech;

comparing, with the processor, the hypothesis to the expected response; and

accepting the hypothesis at least partially in response to the hypothesis favorably comparing to the expected response, comprising

in response to the hypothesis favorably comparing to the expected response, modifying, with the processor, an acceptance threshold from an initial acceptance threshold to a modified acceptance threshold in order to favor the expected response over another possible response that can be received as input speech from the user, and

determining, with the processor, to accept the hypothesis, comprising comparing the confidence factor to the modified acceptance threshold, and

wherein the initial acceptance threshold differs from the modified acceptance threshold such that the hypothesis would not have been accepted if the hypothesis was compared to the initial acceptance threshold.

2. The method of claim 1, wherein the hypothesis is multiple words, and the confidence factor is for at least one of the words.

3. The method of claim 1, comprising decreasing the acceptance threshold in response to the hypothesis favorably comparing to the expected response.

4. The method of claim 1, comprising increasing the acceptance threshold in response to the hypothesis favorably comparing to the expected response.

5. The method of claim 1, wherein an adjustment amount for the acceptance threshold is a variable amount.

6. The method of claim 5, wherein the adjustment amount is dynamically varied based on at least one of the number of words in possible responses or the number of actual vocabulary items in the expected response.

7. The method of claim 5, wherein the adjustment amount is dynamically varied based on the likelihood of the user saying the expected response.

8. The method of claim 7, wherein the likelihood of the user saying the expected response is dynamically determined based upon observance of the previous spoken responses of the user.

9. The method of claim 5, wherein the adjustment amount is dynamically varied based on a cost selected from the group consisting of:

a cost of incorrectly accepting the hypothesis as the expected response; and

a cost of incorrectly rejecting the hypothesis.

10. The method of claim 1, wherein the expected response is obtained by at least one of evaluating an expression or retrieving the expected response from a table or data structure prior to comparing the hypothesis to the expected response.

11. The method of claim 1, wherein:

the speech input device comprises a microphone; and

the signal processor comprises an analog-to-digital converter.

12. An apparatus for recognizing speech, comprising:

a speech input device configured to receive input speech from a user;

a signal processor configured to digitize the input speech from the user;

a speech recognizer configured to process the input speech from the user to generate a hypothesis and confidence factor for the input speech;

a processor and acceptance algorithm executable by the processor to determine whether to accept the hypothesis; and

an adjustment component configured to compare the hypothesis to an expected response, and, if the comparison is favorable, modify an acceptance threshold associated with the acceptance algorithm in order to favor the expected response over another possible response that can be received as input speech from the user, comprising the adjustment component being configured modify an initial acceptance threshold to a modified acceptance threshold,

wherein the acceptance algorithm being executable by the processor to determine whether to accept the hypothesis is comprised of

the acceptance algorithm being executable by the processor to compare the confidence factor to the initial acceptance threshold in response to an unfavorable comparison between the hypothesis and the expected response, and

the acceptance algorithm being executable by the processor to compare the confidence factor to the modified acceptance threshold in response to a favorable comparison between the hypothesis and the expected response, and

wherein the initial acceptance threshold differs from the modified acceptance threshold such that the hypothesis is more likely to accepted if the hypothesis is compared to the modified acceptance threshold.

13. The system of claim 12, wherein the hypothesis is multiple words, and the confidence factor is for at least one of the words.

14. The system of claim 12, wherein the adjustment component is configured to adjust the acceptance threshold by a variable amount.

15. The system of claim 12, wherein:

the speech input device comprises a microphone; and

the signal processor comprises an analog-to-digital converter.

16. A method for recognizing speech, comprising:

conducting, with a speech recognition system, a speech dialog with a user, comprising receiving, with a speech input device, input speech from the user, and digitizing, with a signal processor, the input speech from the user;

in the speech dialog, determining, with a processor of the speech recognition system, an expected response to be received as input speech from a user;

processing, with the processor, input speech from a user to generate a hypothesis and confidence factor; and

accepting the hypothesis at least partially in response to the hypothesis favorably comparing to the expected response, comprising

based upon the determined expected response, modifying, with the processor, an acceptance threshold from an initial acceptance threshold to a modified acceptance threshold in order to favor the expected response over another possible response that can be received as input speech from the user, and

determining, with the processor, to accept the hypothesis, comprising comparing the confidence factor to the modified acceptance threshold,

wherein the initial acceptance threshold differs from the modified acceptance threshold such that the hypothesis would not have been accepted if the hypothesis was compared to the initial acceptance threshold.

17. The method of claim 16, wherein:

the speech input device comprises a microphone; and

the signal processor comprises an analog-to-digital converter.