DETECTING POTENTIAL SIGNIFICANT ERRORS IN SPEECH RECOGNITION RESULTS

• US 20140012580A1
• Filed: 07/09/2012
• Published: 01/09/2014
• Est. Priority Date: 07/09/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Ganong, William F. III, Vemula, Raghu, Fleming, Robert
• Granted Patent: US 9,064,493 B2
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G10L 15/01 Assessment or evaluation of speech recognition systems
• 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
• CPC Class Codes: G10L 2015/085 Methods for reducing search complexity, pruning

Abstract

In some embodiments, the recognition results produced by a speech processing system (which may include a top recognition result and one or more alternative recognition results) based on an analysis of a speech input, are evaluated for indications of potential significant errors. In some embodiments, the recognition results may be evaluated to determine whether a meaning of any of the alternative recognition results differs from a meaning of the top recognition result in a manner that is significant for the domain. In some embodiments, one or more of the recognition results may be evaluated to determine whether the result(s) include one or more words or phrases that, when included in a result, would change a meaning of the result in a manner that would be significant for the domain.

Claims

1. A method of processing results of a recognition by an automatic speech recognition (ASR) system on a speech input, the results comprising a first recognition result identified by the ASR system as most likely to be a correct recognition result for the speech input, the results further comprising at least one alternative recognition result identified by the ASR system, the method comprising:

determining whether a meaning of the first recognition result differs in a medically significant way from a meaning of the at least one alternative recognition result; and

in response to determining that that the meaning of the first recognition result differs in a medically significant way from the meaning of the at least one alternative recognition result, triggering an alert.

2. The method of claim 1, wherein the determining whether the meaning of the first recognition result differs in a medically significant way from the meaning of the at least one alternative recognition result comprises making a determination from a group consisting of:

(a) determining whether a symptom expressed in the first recognition result differs from a symptom expressed in one of the at least one alternative recognition result;

(b) determining whether a diagnosis expressed in the first recognition result differs from a diagnosis expressed in one of the at least one alternative recognition result; and

(c) determining whether a treatment expressed in the first recognition result differs from a treatment expressed in one of the at least one alternative recognition result.

3. The method of claim 1, wherein:

the at least one alternative recognition result comprises a second recognition result;

the method further comprises semantically interpreting each of the first recognition result and the second recognition result to determine at least one first fact expressed in the first recognition result and at least one second fact expressed in the second recognition result; and

the determining whether the meaning of the first recognition result differs from the meaning of the at least one alternative recognition result comprises determining whether there is a difference between the at least one first fact and the at least one second fact.

4. The method of claim 3, wherein:

the at least one first fact comprises a fact having a fact type that is marked as significant; and

the determining whether there is a difference between the at least one first fact and the at least one second fact comprises determining whether the at least one second fact comprises the fact having the fact type marked as significant.

5. The method of claim 3, wherein the semantically interpreting the first recognition result and the second recognition result comprises semantically interpreting the first recognition result and the second recognition result using a medical fact extractor to determine at least one first medical fact expressed in the first recognition result and at least one second medical fact expressed in the second recognition result.

6. The method of claim 1, wherein the determining whether the meaning of the first recognition result differs in a medically significant way from the meaning of the at least one alternative recognition result comprises determining whether the first recognition result includes a first member of a set of words or phrases, each member of the set comprising a word or phrase and being associated with at least one other member of the set, and determining whether the at least one alternative recognition result includes at least one other member associated with the first member of the set.

7. The method of claim 6, wherein:

the first member of the set of words or phrases is associated with a second member of the set with which the first member is acoustically confusable and that, when substituted for the first member in a recognition result, changes a meaning of the recognition result; and

the determining whether the at least one alternative recognition result includes the at least one other member associated with the first member of the set comprises determining whether the at least one alternative recognition result includes the second member of the set.

8. The method of claim 7, wherein the second member of the set, when substituted for the first member in a recognition result, changes a medical meaning of the recognition result.

9. An apparatus comprising:

at least one processor; and

at least one storage medium having encoded thereon processor-executable instructions that, when executed by the at least one processor, cause the at least one processor to carry out a method of processing results of a recognition by an automatic speech recognition (ASR) system on a speech input, the results comprising a first recognition result identified by the ASR system as most likely to be a correct recognition result for the speech input, the results further comprising at least one alternative recognition result identified by the ASR system, the method comprising;

determining whether a meaning of the first recognition result differs in a medically significant way from a meaning of the at least one alternative recognition result; and

in response to determining that that the meaning of the first recognition result differs in a medically significant way from the meaning of the at least one alternative recognition result, triggering an alert.

10. The apparatus of claim 9, wherein the determining whether the meaning of the first recognition result differs in a medically significant way from the meaning of the at least one alternative recognition result comprises making a determination from a group consisting of:

(a) determining whether a symptom expressed in the first recognition result differs from a symptom expressed in one of the at least one alternative recognition result;

(b) determining whether a diagnosis expressed in the first recognition result differs from a diagnosis expressed in one of the at least one alternative recognition result; and

(c) determining whether a treatment expressed in the first recognition result differs from a treatment expressed in one of the at least one alternative recognition result.

11. The apparatus of claim 9, wherein:

the at least one alternative recognition result comprises a second recognition result;

the method further comprises semantically interpreting each of the first recognition result and the second recognition result to determine at least one first fact expressed in the first recognition result and at least one second fact expressed in the second recognition result; and

the determining whether the meaning of the first recognition result differs from the meaning of the at least one alternative recognition result comprises determining whether there is a difference between the at least one first fact and the at least one second fact.

12. The apparatus of claim 11, wherein:

the at least one first fact comprises a fact having a fact type that is marked as significant; and

the determining whether there is a difference between the at least one first fact and the at least one second fact comprises determining whether the at least one second fact comprises the fact having the fact type marked as significant.

13. The apparatus of claim 9, wherein the determining whether the meaning of the first recognition result differs in a medically significant way from the meaning of the at least one alternative recognition result comprises determining whether the first recognition result includes a first member of a set of words or phrases, each member of the set comprising a word or phrase and being associated with at least one other member of the set, and determining whether the at least one alternative recognition result includes at least one other member associated with the first member of the set.

14. The apparatus of claim 13, wherein:

the first member of the set of words or phrases is associated with a second member of the set with which the first member is acoustically confusable and that, when substituted for the first member in a recognition result, changes a meaning of the recognition result; and

the determining whether the at least one alternative recognition result includes the at least one other member associated with the first member of the set comprises determining whether the at least one alternative recognition result includes the second member of the set.

15. At least one computer-readable storage medium having encoded thereon computer-executable instructions that, when executed by at least one computer, cause the at least one computer to carry out a method of processing results of a recognition by an automatic speech recognition (ASR) system on a speech input, the results comprising two or more results identified by the ASR system as likely to be an accurate recognition result for the speech input, the method comprising:

evaluating the results to determine whether a first result of the two or more results comprises a word or phrase that appears in a set of words or phrases; and

in response to determining that the first result comprises a word or phrase that appears in the set, determining whether to trigger an alert for the first result.

16. The at least one computer-readable storage medium of claim 15, wherein:

each member of the set of words or phrases comprises a first word or phrase and is associated with at least one other member of the set comprising a second word or phrase, is acoustically confusable with the second word or phrase, and, when substituted for the second word or phrase in a recognition result, changes a meaning of the recognition result; and

the determining whether to trigger the alert comprises determining, when the first result comprises the first word or phrase, whether a second recognition result of the two or more results includes the second word or phrase and triggering the alert when the second recognition result includes the second word or phrase.

17. The at least one computer-readable storage medium of claim 16, wherein determining whether the second recognition result includes second word or phrase comprises determining whether the second recognition result comprises the second word or phrase at a position in the second recognition result aligned with a position of the first word or phrase in the first result.

18. The at least one computer-readable storage medium of claim 15, wherein:

a first member of the set is associated with a second member that, when substituted for the first member in a recognition result, changes a medical meaning of the recognition result; and

the determining whether to trigger the alert comprises determining, when the first result comprises the first member, whether the second recognition result comprises the second member.

19. The at least one computer-readable storage medium of claim 15, wherein:

a first member of the set of words is a null word and is associated with a second member of the set; and

the determining whether to trigger the alert comprises determining whether the second recognition result comprises the second member.

20. The at least one computer-readable storage medium of claim 15, wherein the method further comprises triggering the alert, wherein triggering the alert comprises presenting an audible and/or visible message via a user interface.